From ea8b62e47bdd1a30248bb68bd4be700c701d5685 Mon Sep 17 00:00:00 2001
From: Greg Steuck <blackgnezdo@gmail.com>
Date: Tue, 25 Oct 2016 12:43:06 -0700
Subject: [PATCH] Haddock (#3)

* Trivial script for regenerating haddocks: exclude .haddock files.

* Haddock regen
---
 docs/haddock/doc-index-47.html                |     4 +
 docs/haddock/doc-index-92.html                |     4 +
 docs/haddock/doc-index-95.html                |     4 +
 docs/haddock/doc-index-A.html                 |     4 +
 docs/haddock/doc-index-All.html               |     4 +
 docs/haddock/doc-index-B.html                 |     4 +
 docs/haddock/doc-index-C.html                 |     4 +
 docs/haddock/doc-index-D.html                 |     4 +
 docs/haddock/doc-index-E.html                 |     4 +
 docs/haddock/doc-index-F.html                 |     4 +
 docs/haddock/doc-index-G.html                 |     4 +
 docs/haddock/doc-index-H.html                 |     4 +
 docs/haddock/doc-index-I.html                 |     4 +
 docs/haddock/doc-index-K.html                 |     4 +
 docs/haddock/doc-index-L.html                 |     4 +
 docs/haddock/doc-index-M.html                 |     4 +
 docs/haddock/doc-index-N.html                 |     4 +
 docs/haddock/doc-index-O.html                 |     4 +
 docs/haddock/doc-index-P.html                 |     4 +
 docs/haddock/doc-index-Q.html                 |     4 +
 docs/haddock/doc-index-R.html                 |     4 +
 docs/haddock/doc-index-S.html                 |     4 +
 docs/haddock/doc-index-T.html                 |     4 +
 docs/haddock/doc-index-U.html                 |     4 +
 docs/haddock/doc-index-V.html                 |     4 +
 docs/haddock/doc-index-W.html                 |     4 +
 docs/haddock/doc-index-Z.html                 |     4 +
 docs/haddock/doc-index.html                   |     4 +
 docs/haddock/frames.html                      |    30 +
 docs/haddock/haddock-util.js                  |   344 +
 docs/haddock/hslogo-16.png                    |   Bin 0 -> 1684 bytes
 docs/haddock/index-frames.html                |     4 +
 docs/haddock/index.html                       |     4 +
 docs/haddock/minus.gif                        |   Bin 0 -> 56 bytes
 docs/haddock/ocean.css                        |   600 +
 docs/haddock/plus.gif                         |   Bin 0 -> 59 bytes
 docs/haddock/synopsis.png                     |   Bin 0 -> 11327 bytes
 .../tensorflow-0.1.0.0/TensorFlow-Build.html  |    27 +
 .../TensorFlow-BuildOp.html                   |     6 +
 .../TensorFlow-ControlFlow.html               |     9 +
 .../TensorFlow-Internal-FFI.html              |     8 +
 .../TensorFlow-Internal-VarInt.html           |     4 +
 .../tensorflow-0.1.0.0/TensorFlow-Nodes.html  |     6 +
 .../tensorflow-0.1.0.0/TensorFlow-Output.html |    14 +
 .../TensorFlow-Session.html                   |    23 +
 .../tensorflow-0.1.0.0/TensorFlow-Tensor.html |    17 +
 .../tensorflow-0.1.0.0/TensorFlow-Types.html  |    13 +
 .../haddock/tensorflow-0.1.0.0/doc-index.html |     4 +
 docs/haddock/tensorflow-0.1.0.0/frames.html   |    30 +
 .../tensorflow-0.1.0.0/haddock-util.js        |   344 +
 docs/haddock/tensorflow-0.1.0.0/hslogo-16.png |   Bin 0 -> 1684 bytes
 .../tensorflow-0.1.0.0/index-frames.html      |     4 +
 docs/haddock/tensorflow-0.1.0.0/index.html    |     4 +
 .../mini_TensorFlow-Build.html                |     4 +
 .../mini_TensorFlow-BuildOp.html              |     4 +
 .../mini_TensorFlow-ControlFlow.html          |     4 +
 .../mini_TensorFlow-Internal-FFI.html         |     4 +
 .../mini_TensorFlow-Internal-VarInt.html      |     4 +
 .../mini_TensorFlow-Nodes.html                |     4 +
 .../mini_TensorFlow-Output.html               |     4 +
 .../mini_TensorFlow-Session.html              |     4 +
 .../mini_TensorFlow-Tensor.html               |     4 +
 .../mini_TensorFlow-Types.html                |     4 +
 docs/haddock/tensorflow-0.1.0.0/minus.gif     |   Bin 0 -> 56 bytes
 docs/haddock/tensorflow-0.1.0.0/ocean.css     |   600 +
 docs/haddock/tensorflow-0.1.0.0/plus.gif      |   Bin 0 -> 59 bytes
 .../src/TensorFlow-Build.html                 |   387 +
 .../src/TensorFlow-BuildOp.html               |   210 +
 .../src/TensorFlow-ControlFlow.html           |    98 +
 .../src/TensorFlow-Internal-FFI.html          |   254 +
 .../src/TensorFlow-Internal-Raw.html          |   514 +
 .../src/TensorFlow-Internal-VarInt.html       |    61 +
 .../src/TensorFlow-Nodes.html                 |   152 +
 .../src/TensorFlow-Orphans.html               |    57 +
 .../src/TensorFlow-Output.html                |   167 +
 .../src/TensorFlow-Session.html               |   213 +
 .../src/TensorFlow-Tensor.html                |    96 +
 .../src/TensorFlow-Types.html                 |   393 +
 .../tensorflow-0.1.0.0/src/hscolour.css       |     5 +
 docs/haddock/tensorflow-0.1.0.0/synopsis.png  |   Bin 0 -> 11327 bytes
 .../haddock/tensorflow-0.1.0.0/tensorflow.txt |   626 +
 .../TensorFlow-GenOps-Core.html               |  2059 ++
 .../doc-index-95.html                         |     4 +
 .../doc-index-A.html                          |     4 +
 .../doc-index-All.html                        |     4 +
 .../doc-index-B.html                          |     4 +
 .../doc-index-C.html                          |     4 +
 .../doc-index-D.html                          |     4 +
 .../doc-index-E.html                          |     4 +
 .../doc-index-F.html                          |     4 +
 .../doc-index-G.html                          |     4 +
 .../doc-index-H.html                          |     4 +
 .../doc-index-I.html                          |     4 +
 .../doc-index-L.html                          |     4 +
 .../doc-index-M.html                          |     4 +
 .../doc-index-N.html                          |     4 +
 .../doc-index-O.html                          |     4 +
 .../doc-index-P.html                          |     4 +
 .../doc-index-Q.html                          |     4 +
 .../doc-index-R.html                          |     4 +
 .../doc-index-S.html                          |     4 +
 .../doc-index-T.html                          |     4 +
 .../doc-index-U.html                          |     4 +
 .../doc-index-V.html                          |     4 +
 .../doc-index-W.html                          |     4 +
 .../doc-index-Z.html                          |     4 +
 .../doc-index.html                            |     4 +
 .../tensorflow-core-ops-0.1.0.0/frames.html   |    30 +
 .../haddock-util.js                           |   344 +
 .../tensorflow-core-ops-0.1.0.0/hslogo-16.png |   Bin 0 -> 1684 bytes
 .../index-frames.html                         |     4 +
 .../tensorflow-core-ops-0.1.0.0/index.html    |     4 +
 .../mini_TensorFlow-GenOps-Core.html          |     4 +
 .../tensorflow-core-ops-0.1.0.0/minus.gif     |   Bin 0 -> 56 bytes
 .../tensorflow-core-ops-0.1.0.0/ocean.css     |   600 +
 .../tensorflow-core-ops-0.1.0.0/plus.gif      |   Bin 0 -> 59 bytes
 .../src/TensorFlow-GenOps-Core.html           | 22618 ++++++++++++++++
 .../src/hscolour.css                          |     5 +
 .../tensorflow-core-ops-0.1.0.0/synopsis.png  |   Bin 0 -> 11327 bytes
 .../tensorflow-core-ops.txt                   |  4576 ++++
 .../TensorFlow-Examples-MNIST-Parse.html      |     4 +
 ...ensorFlow-Examples-MNIST-TrainedGraph.html |     4 +
 .../tensorflow-mnist-0.1.0.0/doc-index.html   |     4 +
 .../tensorflow-mnist-0.1.0.0/frames.html      |    30 +
 .../tensorflow-mnist-0.1.0.0/haddock-util.js  |   344 +
 .../tensorflow-mnist-0.1.0.0/hslogo-16.png    |   Bin 0 -> 1684 bytes
 .../index-frames.html                         |     4 +
 .../tensorflow-mnist-0.1.0.0/index.html       |     4 +
 .../mini_TensorFlow-Examples-MNIST-Parse.html |     4 +
 ...ensorFlow-Examples-MNIST-TrainedGraph.html |     4 +
 .../tensorflow-mnist-0.1.0.0/minus.gif        |   Bin 0 -> 56 bytes
 .../tensorflow-mnist-0.1.0.0/ocean.css        |   600 +
 .../haddock/tensorflow-mnist-0.1.0.0/plus.gif |   Bin 0 -> 59 bytes
 .../src/Paths_tensorflow_mnist.html           |    46 +
 .../src/TensorFlow-Examples-MNIST-Parse.html  |   107 +
 ...ensorFlow-Examples-MNIST-TrainedGraph.html |    41 +
 .../tensorflow-mnist-0.1.0.0/src/hscolour.css |     5 +
 .../tensorflow-mnist-0.1.0.0/synopsis.png     |   Bin 0 -> 11327 bytes
 .../tensorflow-mnist.txt                      |    41 +
 .../TensorFlow-Examples-MNIST-InputData.html  |     4 +
 .../doc-index.html                            |     4 +
 .../frames.html                               |    30 +
 .../haddock-util.js                           |   344 +
 .../hslogo-16.png                             |   Bin 0 -> 1684 bytes
 .../index-frames.html                         |     4 +
 .../index.html                                |     4 +
 ...i_TensorFlow-Examples-MNIST-InputData.html |     4 +
 .../minus.gif                                 |   Bin 0 -> 56 bytes
 .../ocean.css                                 |   600 +
 .../plus.gif                                  |   Bin 0 -> 59 bytes
 .../Paths_tensorflow_mnist_input_data.html    |    46 +
 .../TensorFlow-Examples-MNIST-InputData.html  |    42 +
 .../src/hscolour.css                          |     5 +
 .../synopsis.png                              |   Bin 0 -> 11327 bytes
 .../tensorflow-mnist-input-data.txt           |    19 +
 .../TensorFlow-OpGen-AttrVal.html             |     6 +
 .../TensorFlow-OpGen.html                     |     4 +
 .../tensorflow-opgen-0.1.0.0/doc-index.html   |     4 +
 .../tensorflow-opgen-0.1.0.0/frames.html      |    30 +
 .../tensorflow-opgen-0.1.0.0/haddock-util.js  |   344 +
 .../tensorflow-opgen-0.1.0.0/hslogo-16.png    |   Bin 0 -> 1684 bytes
 .../index-frames.html                         |     4 +
 .../tensorflow-opgen-0.1.0.0/index.html       |     4 +
 .../mini_TensorFlow-OpGen-AttrVal.html        |     4 +
 .../mini_TensorFlow-OpGen.html                |     4 +
 .../tensorflow-opgen-0.1.0.0/minus.gif        |   Bin 0 -> 56 bytes
 .../tensorflow-opgen-0.1.0.0/ocean.css        |   600 +
 .../haddock/tensorflow-opgen-0.1.0.0/plus.gif |   Bin 0 -> 59 bytes
 .../src/TensorFlow-OpGen-AttrVal.html         |   131 +
 .../src/TensorFlow-OpGen.html                 |   468 +
 .../tensorflow-opgen-0.1.0.0/src/hscolour.css |     5 +
 .../tensorflow-opgen-0.1.0.0/synopsis.png     |   Bin 0 -> 11327 bytes
 .../tensorflow-opgen.txt                      |    48 +
 .../TensorFlow-EmbeddingOps.html              |    16 +
 .../TensorFlow-Gradient.html                  |     4 +
 .../TensorFlow-Ops.html                       |   122 +
 .../tensorflow-ops-0.1.0.0/doc-index.html     |     4 +
 .../tensorflow-ops-0.1.0.0/frames.html        |    30 +
 .../tensorflow-ops-0.1.0.0/haddock-util.js    |   344 +
 .../tensorflow-ops-0.1.0.0/hslogo-16.png      |   Bin 0 -> 1684 bytes
 .../tensorflow-ops-0.1.0.0/index-frames.html  |     4 +
 .../haddock/tensorflow-ops-0.1.0.0/index.html |     4 +
 .../mini_TensorFlow-EmbeddingOps.html         |     4 +
 .../mini_TensorFlow-Gradient.html             |     4 +
 .../mini_TensorFlow-Ops.html                  |     4 +
 docs/haddock/tensorflow-ops-0.1.0.0/minus.gif |   Bin 0 -> 56 bytes
 docs/haddock/tensorflow-ops-0.1.0.0/ocean.css |   600 +
 docs/haddock/tensorflow-ops-0.1.0.0/plus.gif  |   Bin 0 -> 59 bytes
 .../src/TensorFlow-EmbeddingOps.html          |    87 +
 .../src/TensorFlow-Gradient.html              |   708 +
 .../src/TensorFlow-Ops.html                   |   307 +
 .../tensorflow-ops-0.1.0.0/src/hscolour.css   |     5 +
 .../tensorflow-ops-0.1.0.0/synopsis.png       |   Bin 0 -> 11327 bytes
 .../tensorflow-ops-0.1.0.0/tensorflow-ops.txt |   378 +
 ...o-Tensorflow-Core-Framework-AttrValue.html |     4 +
 ...Proto-Tensorflow-Core-Framework-Graph.html |     4 +
 ...oto-Tensorflow-Core-Framework-NodeDef.html |     4 +
 ...Proto-Tensorflow-Core-Framework-OpDef.html |     4 +
 ...sorflow-Core-Framework-ResourceHandle.html |     4 +
 ...roto-Tensorflow-Core-Framework-Tensor.html |     4 +
 ...Tensorflow-Core-Framework-TensorShape.html |     4 +
 ...Proto-Tensorflow-Core-Framework-Types.html |     4 +
 ...Proto-Tensorflow-Core-Protobuf-Config.html |     4 +
 .../doc-index-95.html                         |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-A.html |     4 +
 .../doc-index-All.html                        |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-B.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-C.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-D.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-E.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-F.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-G.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-H.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-I.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-K.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-L.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-M.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-N.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-O.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-P.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-R.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-S.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-T.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-U.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index-V.html |     4 +
 .../tensorflow-proto-0.1.0.0/doc-index.html   |     4 +
 .../tensorflow-proto-0.1.0.0/frames.html      |    30 +
 .../tensorflow-proto-0.1.0.0/haddock-util.js  |   344 +
 .../tensorflow-proto-0.1.0.0/hslogo-16.png    |   Bin 0 -> 1684 bytes
 .../index-frames.html                         |     4 +
 .../tensorflow-proto-0.1.0.0/index.html       |     4 +
 ...o-Tensorflow-Core-Framework-AttrValue.html |     4 +
 ...Proto-Tensorflow-Core-Framework-Graph.html |     4 +
 ...oto-Tensorflow-Core-Framework-NodeDef.html |     4 +
 ...Proto-Tensorflow-Core-Framework-OpDef.html |     4 +
 ...sorflow-Core-Framework-ResourceHandle.html |     4 +
 ...roto-Tensorflow-Core-Framework-Tensor.html |     4 +
 ...Tensorflow-Core-Framework-TensorShape.html |     4 +
 ...Proto-Tensorflow-Core-Framework-Types.html |     4 +
 ...Proto-Tensorflow-Core-Protobuf-Config.html |     4 +
 .../tensorflow-proto-0.1.0.0/minus.gif        |   Bin 0 -> 56 bytes
 .../tensorflow-proto-0.1.0.0/ocean.css        |   600 +
 .../haddock/tensorflow-proto-0.1.0.0/plus.gif |   Bin 0 -> 59 bytes
 ...-Core-Framework-AllocationDescription.html |   233 +
 ...o-Tensorflow-Core-Framework-AttrValue.html |   762 +
 ...o-Tensorflow-Core-Framework-CostGraph.html |   570 +
 ...to-Tensorflow-Core-Framework-Function.html |   586 +
 ...Proto-Tensorflow-Core-Framework-Graph.html |   198 +
 ...oto-Tensorflow-Core-Framework-NodeDef.html |   257 +
 ...Proto-Tensorflow-Core-Framework-OpDef.html |   854 +
 ...sorflow-Core-Framework-ResourceHandle.html |   182 +
 ...o-Tensorflow-Core-Framework-StepStats.html |   653 +
 ...roto-Tensorflow-Core-Framework-Tensor.html |   448 +
 ...flow-Core-Framework-TensorDescription.html |   187 +
 ...Tensorflow-Core-Framework-TensorShape.html |   166 +
 ...Proto-Tensorflow-Core-Framework-Types.html |   355 +
 ...to-Tensorflow-Core-Framework-Versions.html |   128 +
 ...Proto-Tensorflow-Core-Protobuf-Config.html |  1671 ++
 .../tensorflow-proto-0.1.0.0/src/hscolour.css |     5 +
 .../tensorflow-proto-0.1.0.0/synopsis.png     |   Bin 0 -> 11327 bytes
 .../tensorflow-proto.txt                      |   663 +
 .../TensorFlow-Queue.html                     |     9 +
 .../tensorflow-queue-0.1.0.0/doc-index.html   |     4 +
 .../tensorflow-queue-0.1.0.0/frames.html      |    30 +
 .../tensorflow-queue-0.1.0.0/haddock-util.js  |   344 +
 .../tensorflow-queue-0.1.0.0/hslogo-16.png    |   Bin 0 -> 1684 bytes
 .../index-frames.html                         |     4 +
 .../tensorflow-queue-0.1.0.0/index.html       |     4 +
 .../mini_TensorFlow-Queue.html                |     4 +
 .../tensorflow-queue-0.1.0.0/minus.gif        |   Bin 0 -> 56 bytes
 .../tensorflow-queue-0.1.0.0/ocean.css        |   600 +
 .../haddock/tensorflow-queue-0.1.0.0/plus.gif |   Bin 0 -> 59 bytes
 .../src/TensorFlow-Queue.html                 |    89 +
 .../tensorflow-queue-0.1.0.0/src/hscolour.css |     5 +
 .../tensorflow-queue-0.1.0.0/synopsis.png     |   Bin 0 -> 11327 bytes
 .../tensorflow-queue.txt                      |    26 +
 tools/haddock.sh                              |     3 +-
 277 files changed, 52638 insertions(+), 1 deletion(-)
 create mode 100644 docs/haddock/doc-index-47.html
 create mode 100644 docs/haddock/doc-index-92.html
 create mode 100644 docs/haddock/doc-index-95.html
 create mode 100644 docs/haddock/doc-index-A.html
 create mode 100644 docs/haddock/doc-index-All.html
 create mode 100644 docs/haddock/doc-index-B.html
 create mode 100644 docs/haddock/doc-index-C.html
 create mode 100644 docs/haddock/doc-index-D.html
 create mode 100644 docs/haddock/doc-index-E.html
 create mode 100644 docs/haddock/doc-index-F.html
 create mode 100644 docs/haddock/doc-index-G.html
 create mode 100644 docs/haddock/doc-index-H.html
 create mode 100644 docs/haddock/doc-index-I.html
 create mode 100644 docs/haddock/doc-index-K.html
 create mode 100644 docs/haddock/doc-index-L.html
 create mode 100644 docs/haddock/doc-index-M.html
 create mode 100644 docs/haddock/doc-index-N.html
 create mode 100644 docs/haddock/doc-index-O.html
 create mode 100644 docs/haddock/doc-index-P.html
 create mode 100644 docs/haddock/doc-index-Q.html
 create mode 100644 docs/haddock/doc-index-R.html
 create mode 100644 docs/haddock/doc-index-S.html
 create mode 100644 docs/haddock/doc-index-T.html
 create mode 100644 docs/haddock/doc-index-U.html
 create mode 100644 docs/haddock/doc-index-V.html
 create mode 100644 docs/haddock/doc-index-W.html
 create mode 100644 docs/haddock/doc-index-Z.html
 create mode 100644 docs/haddock/doc-index.html
 create mode 100644 docs/haddock/frames.html
 create mode 100644 docs/haddock/haddock-util.js
 create mode 100644 docs/haddock/hslogo-16.png
 create mode 100644 docs/haddock/index-frames.html
 create mode 100644 docs/haddock/index.html
 create mode 100644 docs/haddock/minus.gif
 create mode 100644 docs/haddock/ocean.css
 create mode 100644 docs/haddock/plus.gif
 create mode 100644 docs/haddock/synopsis.png
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/TensorFlow-Build.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/TensorFlow-BuildOp.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/TensorFlow-ControlFlow.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/TensorFlow-Internal-VarInt.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/TensorFlow-Nodes.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/TensorFlow-Output.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/TensorFlow-Session.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/TensorFlow-Tensor.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/TensorFlow-Types.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/doc-index.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/frames.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/haddock-util.js
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/hslogo-16.png
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/index-frames.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/index.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Build.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-BuildOp.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-ControlFlow.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Internal-FFI.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Internal-VarInt.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Nodes.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Output.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Session.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Tensor.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Types.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/minus.gif
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/ocean.css
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/plus.gif
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Build.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-BuildOp.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-ControlFlow.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Internal-FFI.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Internal-Raw.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Internal-VarInt.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Nodes.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Orphans.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Output.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Session.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Tensor.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Types.html
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/src/hscolour.css
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/synopsis.png
 create mode 100644 docs/haddock/tensorflow-0.1.0.0/tensorflow.txt
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-95.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-A.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-All.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-B.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-C.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-D.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-E.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-F.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-G.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-H.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-I.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-L.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-M.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-N.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-O.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-P.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-Q.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-R.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-S.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-T.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-U.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-V.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-W.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-Z.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/frames.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/haddock-util.js
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/hslogo-16.png
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/index-frames.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/index.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/mini_TensorFlow-GenOps-Core.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/minus.gif
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/ocean.css
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/plus.gif
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/src/TensorFlow-GenOps-Core.html
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/src/hscolour.css
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/synopsis.png
 create mode 100644 docs/haddock/tensorflow-core-ops-0.1.0.0/tensorflow-core-ops.txt
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-TrainedGraph.html
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/doc-index.html
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/frames.html
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/haddock-util.js
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/hslogo-16.png
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/index-frames.html
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/index.html
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/mini_TensorFlow-Examples-MNIST-Parse.html
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/mini_TensorFlow-Examples-MNIST-TrainedGraph.html
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/minus.gif
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/ocean.css
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/plus.gif
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/src/Paths_tensorflow_mnist.html
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/src/TensorFlow-Examples-MNIST-Parse.html
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/src/TensorFlow-Examples-MNIST-TrainedGraph.html
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/src/hscolour.css
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/synopsis.png
 create mode 100644 docs/haddock/tensorflow-mnist-0.1.0.0/tensorflow-mnist.txt
 create mode 100644 docs/haddock/tensorflow-mnist-input-data-0.1.0.0/TensorFlow-Examples-MNIST-InputData.html
 create mode 100644 docs/haddock/tensorflow-mnist-input-data-0.1.0.0/doc-index.html
 create mode 100644 docs/haddock/tensorflow-mnist-input-data-0.1.0.0/frames.html
 create mode 100644 docs/haddock/tensorflow-mnist-input-data-0.1.0.0/haddock-util.js
 create mode 100644 docs/haddock/tensorflow-mnist-input-data-0.1.0.0/hslogo-16.png
 create mode 100644 docs/haddock/tensorflow-mnist-input-data-0.1.0.0/index-frames.html
 create mode 100644 docs/haddock/tensorflow-mnist-input-data-0.1.0.0/index.html
 create mode 100644 docs/haddock/tensorflow-mnist-input-data-0.1.0.0/mini_TensorFlow-Examples-MNIST-InputData.html
 create mode 100644 docs/haddock/tensorflow-mnist-input-data-0.1.0.0/minus.gif
 create mode 100644 docs/haddock/tensorflow-mnist-input-data-0.1.0.0/ocean.css
 create mode 100644 docs/haddock/tensorflow-mnist-input-data-0.1.0.0/plus.gif
 create mode 100644 docs/haddock/tensorflow-mnist-input-data-0.1.0.0/src/Paths_tensorflow_mnist_input_data.html
 create mode 100644 docs/haddock/tensorflow-mnist-input-data-0.1.0.0/src/TensorFlow-Examples-MNIST-InputData.html
 create mode 100644 docs/haddock/tensorflow-mnist-input-data-0.1.0.0/src/hscolour.css
 create mode 100644 docs/haddock/tensorflow-mnist-input-data-0.1.0.0/synopsis.png
 create mode 100644 docs/haddock/tensorflow-mnist-input-data-0.1.0.0/tensorflow-mnist-input-data.txt
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/doc-index.html
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/frames.html
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/haddock-util.js
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/hslogo-16.png
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/index-frames.html
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/index.html
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/mini_TensorFlow-OpGen-AttrVal.html
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/mini_TensorFlow-OpGen.html
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/minus.gif
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/ocean.css
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/plus.gif
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/src/TensorFlow-OpGen-AttrVal.html
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/src/TensorFlow-OpGen.html
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/src/hscolour.css
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/synopsis.png
 create mode 100644 docs/haddock/tensorflow-opgen-0.1.0.0/tensorflow-opgen.txt
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/TensorFlow-EmbeddingOps.html
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/TensorFlow-Gradient.html
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/TensorFlow-Ops.html
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/doc-index.html
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/frames.html
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/haddock-util.js
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/hslogo-16.png
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/index-frames.html
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/index.html
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/mini_TensorFlow-EmbeddingOps.html
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/mini_TensorFlow-Gradient.html
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/mini_TensorFlow-Ops.html
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/minus.gif
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/ocean.css
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/plus.gif
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/src/TensorFlow-EmbeddingOps.html
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/src/TensorFlow-Gradient.html
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/src/TensorFlow-Ops.html
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/src/hscolour.css
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/synopsis.png
 create mode 100644 docs/haddock/tensorflow-ops-0.1.0.0/tensorflow-ops.txt
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-95.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-A.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-All.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-B.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-C.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-D.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-E.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-F.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-G.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-H.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-I.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-K.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-L.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-M.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-N.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-O.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-P.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-R.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-S.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-T.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-U.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index-V.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/doc-index.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/frames.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/haddock-util.js
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/hslogo-16.png
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/index-frames.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/index.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-AttrValue.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-Graph.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-NodeDef.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-OpDef.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-ResourceHandle.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-Tensor.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-TensorShape.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-Types.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Protobuf-Config.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/minus.gif
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/ocean.css
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/plus.gif
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-AllocationDescription.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-AttrValue.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-CostGraph.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Function.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Graph.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-NodeDef.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-OpDef.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-ResourceHandle.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-StepStats.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Tensor.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-TensorDescription.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-TensorShape.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Types.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Versions.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Protobuf-Config.html
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/src/hscolour.css
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/synopsis.png
 create mode 100644 docs/haddock/tensorflow-proto-0.1.0.0/tensorflow-proto.txt
 create mode 100644 docs/haddock/tensorflow-queue-0.1.0.0/TensorFlow-Queue.html
 create mode 100644 docs/haddock/tensorflow-queue-0.1.0.0/doc-index.html
 create mode 100644 docs/haddock/tensorflow-queue-0.1.0.0/frames.html
 create mode 100644 docs/haddock/tensorflow-queue-0.1.0.0/haddock-util.js
 create mode 100644 docs/haddock/tensorflow-queue-0.1.0.0/hslogo-16.png
 create mode 100644 docs/haddock/tensorflow-queue-0.1.0.0/index-frames.html
 create mode 100644 docs/haddock/tensorflow-queue-0.1.0.0/index.html
 create mode 100644 docs/haddock/tensorflow-queue-0.1.0.0/mini_TensorFlow-Queue.html
 create mode 100644 docs/haddock/tensorflow-queue-0.1.0.0/minus.gif
 create mode 100644 docs/haddock/tensorflow-queue-0.1.0.0/ocean.css
 create mode 100644 docs/haddock/tensorflow-queue-0.1.0.0/plus.gif
 create mode 100644 docs/haddock/tensorflow-queue-0.1.0.0/src/TensorFlow-Queue.html
 create mode 100644 docs/haddock/tensorflow-queue-0.1.0.0/src/hscolour.css
 create mode 100644 docs/haddock/tensorflow-queue-0.1.0.0/synopsis.png
 create mode 100644 docs/haddock/tensorflow-queue-0.1.0.0/tensorflow-queue.txt

diff --git a/docs/haddock/doc-index-47.html b/docs/haddock/doc-index-47.html
new file mode 100644
index 0000000..2de9f3b
--- /dev/null
+++ b/docs/haddock/doc-index-47.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - /)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - /</p><table><tr><td class="src">/=</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:-47--61-">TensorFlow.Types</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-92.html b/docs/haddock/doc-index-92.html
new file mode 100644
index 0000000..d5185e5
--- /dev/null
+++ b/docs/haddock/doc-index-92.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - \)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - \</p><table><tr><td class="src">\\</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:-92--92-">TensorFlow.Types</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-95.html b/docs/haddock/doc-index-95.html
new file mode 100644
index 0000000..7a1c25d
--- /dev/null
+++ b/docs/haddock/doc-index-95.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - _)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - _</p><table><tr><td class="src">_Arg</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:_Arg">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_AttrValue'b</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-b">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'f</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-f">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'func</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-func">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'i</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-i">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'list</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-list">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'b</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-b">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'f</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-f">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'i</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-i">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue's</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-s">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'shape</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-shape">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'tensor</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-tensor">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'type'</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-type-39-">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'placeholder</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-placeholder">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue's</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-s">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'shape</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-shape">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'tensor</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-tensor">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'type'</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-type-39-">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_ConfigProto'allowSoftPlacement</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-allowSoftPlacement">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'deviceCount</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-deviceCount">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'DeviceCountEntry'key</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-DeviceCountEntry-39-key">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'DeviceCountEntry'value</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-DeviceCountEntry-39-value">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'deviceFilters</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-deviceFilters">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'gpuOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-gpuOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'graphOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-graphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'interOpParallelismThreads</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-interOpParallelismThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'intraOpParallelismThreads</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-intraOpParallelismThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'logDevicePlacement</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-logDevicePlacement">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'operationTimeoutInMs</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-operationTimeoutInMs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'placementPeriod</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-placementPeriod">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'sessionInterOpThreadPool</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-sessionInterOpThreadPool">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'usePerSessionThreads</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-usePerSessionThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_DebugTensorWatch'debugOps</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_DebugTensorWatch-39-debugOps">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_DebugTensorWatch'debugUrls</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_DebugTensorWatch-39-debugUrls">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_DebugTensorWatch'nodeName</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_DebugTensorWatch-39-nodeName">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_DebugTensorWatch'outputSlot</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_DebugTensorWatch-39-outputSlot">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'allocatorType</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-allocatorType">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'allowGrowth</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-allowGrowth">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'deferredDeletionBytes</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-deferredDeletionBytes">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'perProcessGpuMemoryFraction</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-perProcessGpuMemoryFraction">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'visibleDeviceList</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-visibleDeviceList">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphDef'library</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:_GraphDef-39-library">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">_GraphDef'node</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:_GraphDef-39-node">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">_GraphDef'version</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:_GraphDef-39-version">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">_GraphDef'versions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:_GraphDef-39-versions">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">_GraphOptions'buildCostModel</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-buildCostModel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'buildCostModelAfter</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-buildCostModelAfter">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'enableBfloat16Sendrecv</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-enableBfloat16Sendrecv">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'enableRecvScheduling</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-enableRecvScheduling">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'inferShapes</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-inferShapes">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'optimizerOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-optimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'placePrunedGraph</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-placePrunedGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'timelineStep</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-timelineStep">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_HostCast</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:_HostCast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_HostRecv</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:_HostRecv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_HostSend</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:_HostSend">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_NameAttrList'attr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_NameAttrList-39-attr">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_NameAttrList'AttrEntry'key</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_NameAttrList-39-AttrEntry-39-key">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_NameAttrList'AttrEntry'value</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_NameAttrList-39-AttrEntry-39-value">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_NameAttrList'name</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_NameAttrList-39-name">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_NodeDef'attr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-attr">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'AttrEntry'key</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-AttrEntry-39-key">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'AttrEntry'value</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-AttrEntry-39-value">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'device</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-device">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'input</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-input">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'name</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-name">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'op</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-op">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_opAttrs</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:_opAttrs">TensorFlow.Output</a></td></tr><tr><td class="src">_opControlInputs</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:_opControlInputs">TensorFlow.Output</a></td></tr><tr><td class="src">_OpDef'allowsUninitializedInput</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-allowsUninitializedInput">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'description</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-description">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'isRef</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-isRef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'name</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-name">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'numberAttr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-numberAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'type'</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-type-39-">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'typeAttr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-typeAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'typeListAttr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-typeListAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'attr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-attr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'allowedValues</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-allowedValues">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'defaultValue</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-defaultValue">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'description</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-description">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'hasMinimum</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-hasMinimum">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'minimum</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-minimum">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'name</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-name">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'type'</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-type-39-">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'deprecation</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-deprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'description</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-description">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'inputArg</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-inputArg">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'isAggregate</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-isAggregate">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'isCommutative</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-isCommutative">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'isStateful</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-isStateful">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'name</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-name">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'outputArg</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-outputArg">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'summary</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-summary">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDeprecation'explanation</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDeprecation-39-explanation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDeprecation'version</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDeprecation-39-version">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_opInputs</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:_opInputs">TensorFlow.Output</a></td></tr><tr><td class="src">_OpList'op</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpList-39-op">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_opName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:_opName">TensorFlow.Output</a></td></tr><tr><td class="src">_OptimizerOptions'doCommonSubexpressionElimination</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_OptimizerOptions-39-doCommonSubexpressionElimination">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_OptimizerOptions'doConstantFolding</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_OptimizerOptions-39-doConstantFolding">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_OptimizerOptions'doFunctionInlining</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_OptimizerOptions-39-doFunctionInlining">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_OptimizerOptions'optLevel</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_OptimizerOptions-39-optLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_opType</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:_opType">TensorFlow.Output</a></td></tr><tr><td class="src">_Recv</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:_Recv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_ResourceHandle'container</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-container">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_ResourceHandle'device</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-device">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_ResourceHandle'hashCode</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-hashCode">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_ResourceHandle'maybeTypeName</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-maybeTypeName">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_ResourceHandle'name</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-name">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_Retval</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:_Retval">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_RunMetadata'costGraph</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunMetadata-39-costGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunMetadata'partitionGraphs</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunMetadata-39-partitionGraphs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunMetadata'stepStats</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunMetadata-39-stepStats">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'debugTensorWatchOpts</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-debugTensorWatchOpts">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'interOpThreadPool</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-interOpThreadPool">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'outputPartitionGraphs</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-outputPartitionGraphs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'timeoutInMs</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-timeoutInMs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'traceLevel</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-traceLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_Send</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:_Send">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_TensorProto'boolVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-boolVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'dcomplexVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-dcomplexVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'doubleVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-doubleVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'dtype</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-dtype">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'floatVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-floatVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'halfVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-halfVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'int64Val</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-int64Val">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'intVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-intVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'resourceHandleVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-resourceHandleVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'scomplexVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-scomplexVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'stringVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-stringVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'tensorContent</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-tensorContent">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'tensorShape</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-tensorShape">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'versionNumber</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-versionNumber">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorShapeProto'dim</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:_TensorShapeProto-39-dim">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">_TensorShapeProto'Dim'name</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:_TensorShapeProto-39-Dim-39-name">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">_TensorShapeProto'Dim'size</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:_TensorShapeProto-39-Dim-39-size">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">_TensorShapeProto'unknownRank</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:_TensorShapeProto-39-unknownRank">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">_ThreadPoolOptionProto'numThreads</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ThreadPoolOptionProto-39-numThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-A.html b/docs/haddock/doc-index-A.html
new file mode 100644
index 0000000..a0546b7
--- /dev/null
+++ b/docs/haddock/doc-index-A.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - A)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - A</p><table><tr><td class="src">abort</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:abort">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">abs</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:abs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:abs">TensorFlow.Ops</a></td></tr><tr><td class="src">acos</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:acos">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">add</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:add">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:add">TensorFlow.Ops</a></td></tr><tr><td class="src">addGraphDef</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:addGraphDef">TensorFlow.Build</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:addGraphDef">TensorFlow.Session</a></td></tr><tr><td class="src">addInitializer</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:addInitializer">TensorFlow.Build</a></td></tr><tr><td class="src">addN</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:addN">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:addN">TensorFlow.Ops</a></td></tr><tr><td class="src">addNewOp</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:addNewOp">TensorFlow.Build</a></td></tr><tr><td class="src">addSummary</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:addSummary">TensorFlow.Build</a></td></tr><tr><td class="src">adjustContrast</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:adjustContrast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">adjustContrastv2</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:adjustContrastv2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">all</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:all">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">allCandidateSampler</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:allCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">allocatorType</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:allocatorType">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">allowedValues</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:allowedValues">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">allowGrowth</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:allowGrowth">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">allowSoftPlacement</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:allowSoftPlacement">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">allowsUninitializedInput</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:allowsUninitializedInput">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">AllTensorTypes</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:AllTensorTypes">TensorFlow.Types</a></td></tr><tr><td class="src">any</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:any">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyAdadelta</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyAdadelta">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyAdagrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyAdagrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyAdagradDA</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyAdagradDA">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyAdam</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyAdam">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyFtrl</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyFtrl">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyGradientDescent</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyGradientDescent">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyMomentum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyMomentum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyProximalAdagrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyProximalAdagrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyProximalGradientDescent</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyProximalGradientDescent">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyRMSProp</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyRMSProp">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">argMax</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:argMax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:argMax">TensorFlow.Ops</a></td></tr><tr><td class="src">argMin</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:argMin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">asGraphDef</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:asGraphDef">TensorFlow.Build</a></td></tr><tr><td class="src">asin</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:asin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">assign</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:assign">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:assign">TensorFlow.Ops</a></td></tr><tr><td class="src">assignAdd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:assignAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">assignSub</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:assignSub">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">asString</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:asString">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">asyncProdNodes</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:asyncProdNodes">TensorFlow.Session</a></td></tr><tr><td class="src">atan</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:atan">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">attr</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:attr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:attr">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:attr">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">AttrBool</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrBool">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrBytes</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrBytes">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrCase</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#t:AttrCase">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrDef</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#t:AttrDef">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">attrDef</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:attrDef">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrFloat</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrFloat">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">Attribute</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:Attribute">TensorFlow.Types</a></td></tr><tr><td class="src">AttrInt64</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrInt64">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">attrLens</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:attrLens">TensorFlow.Types</a></td></tr><tr><td class="src">AttrList</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrList">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">attrOriginal</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:attrOriginal">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrShape</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrShape">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrSingle</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrSingle">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrTemplate</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#t:AttrTemplate">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">attrTemplate</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:attrTemplate">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrTensor</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrTensor">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrType</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrType">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrValue</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:AttrValue">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">AttrValue'ListValue</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:AttrValue-39-ListValue">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">audioSummary</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:audioSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">avgPool</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:avgPool">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">avgPool3D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:avgPool3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">avgPool3DGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:avgPool3DGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">avgPoolGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:avgPoolGrad">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-All.html b/docs/haddock/doc-index-All.html
new file mode 100644
index 0000000..1550959
--- /dev/null
+++ b/docs/haddock/doc-index-All.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="index"><p class="caption">Index</p><table><tr><td class="src">/=</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:-47--61-">TensorFlow.Types</a></td></tr><tr><td class="src">abort</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:abort">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">abs</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:abs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:abs">TensorFlow.Ops</a></td></tr><tr><td class="src">acos</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:acos">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">add</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:add">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:add">TensorFlow.Ops</a></td></tr><tr><td class="src">addGraphDef</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:addGraphDef">TensorFlow.Build</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:addGraphDef">TensorFlow.Session</a></td></tr><tr><td class="src">addInitializer</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:addInitializer">TensorFlow.Build</a></td></tr><tr><td class="src">addN</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:addN">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:addN">TensorFlow.Ops</a></td></tr><tr><td class="src">addNewOp</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:addNewOp">TensorFlow.Build</a></td></tr><tr><td class="src">addSummary</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:addSummary">TensorFlow.Build</a></td></tr><tr><td class="src">adjustContrast</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:adjustContrast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">adjustContrastv2</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:adjustContrastv2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">all</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:all">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">allCandidateSampler</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:allCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">allocatorType</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:allocatorType">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">allowedValues</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:allowedValues">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">allowGrowth</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:allowGrowth">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">allowSoftPlacement</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:allowSoftPlacement">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">allowsUninitializedInput</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:allowsUninitializedInput">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">AllTensorTypes</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:AllTensorTypes">TensorFlow.Types</a></td></tr><tr><td class="src">any</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:any">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyAdadelta</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyAdadelta">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyAdagrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyAdagrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyAdagradDA</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyAdagradDA">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyAdam</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyAdam">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyFtrl</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyFtrl">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyGradientDescent</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyGradientDescent">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyMomentum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyMomentum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyProximalAdagrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyProximalAdagrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyProximalGradientDescent</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyProximalGradientDescent">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyRMSProp</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:applyRMSProp">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">argMax</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:argMax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:argMax">TensorFlow.Ops</a></td></tr><tr><td class="src">argMin</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:argMin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">asGraphDef</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:asGraphDef">TensorFlow.Build</a></td></tr><tr><td class="src">asin</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:asin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">assign</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:assign">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:assign">TensorFlow.Ops</a></td></tr><tr><td class="src">assignAdd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:assignAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">assignSub</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:assignSub">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">asString</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:asString">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">asyncProdNodes</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:asyncProdNodes">TensorFlow.Session</a></td></tr><tr><td class="src">atan</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:atan">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">attr</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:attr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:attr">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:attr">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">AttrBool</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrBool">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrBytes</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrBytes">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrCase</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#t:AttrCase">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrDef</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#t:AttrDef">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">attrDef</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:attrDef">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrFloat</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrFloat">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">Attribute</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:Attribute">TensorFlow.Types</a></td></tr><tr><td class="src">AttrInt64</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrInt64">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">attrLens</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:attrLens">TensorFlow.Types</a></td></tr><tr><td class="src">AttrList</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrList">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">attrOriginal</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:attrOriginal">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrShape</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrShape">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrSingle</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrSingle">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrTemplate</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#t:AttrTemplate">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">attrTemplate</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:attrTemplate">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrTensor</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrTensor">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrType</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:AttrType">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrValue</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:AttrValue">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">AttrValue'ListValue</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:AttrValue-39-ListValue">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">audioSummary</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:audioSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">avgPool</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:avgPool">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">avgPool3D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:avgPool3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">avgPool3DGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:avgPool3DGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">avgPoolGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:avgPoolGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">b</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:b">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">barrier</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:barrier">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrierClose</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:barrierClose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrierIncompleteSize</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:barrierIncompleteSize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrierInsertMany</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:barrierInsertMany">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrierReadySize</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:barrierReadySize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchCholesky</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchCholesky">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchCholeskyGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchCholeskyGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchFFT</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchFFT">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchFFT2D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchFFT2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchFFT3D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchFFT3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchIFFT</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchIFFT">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchIFFT2D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchIFFT2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchIFFT3D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchIFFT3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatMul</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixBandPart</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixBandPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixDeterminant</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixDeterminant">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixDiag</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixDiag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixDiagPart</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixDiagPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixInverse</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixInverse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixSetDiag</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixSetDiag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixSolve</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixSolve">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixSolveLs</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixSolveLs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixTriangularSolve</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixTriangularSolve">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchNormWithGlobalNormalization</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchNormWithGlobalNormalization">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchNormWithGlobalNormalizationGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchNormWithGlobalNormalizationGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchSelfAdjointEig</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchSelfAdjointEig">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchSelfAdjointEigV2</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchSelfAdjointEigV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchSvd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchSvd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchToSpace</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchToSpace">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchToSpaceND</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchToSpaceND">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">betainc</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:betainc">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">biasAdd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:biasAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">biasAddGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:biasAddGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">biasAddV1</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:biasAddV1">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">biasCkpt</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-TrainedGraph.html#v:biasCkpt">TensorFlow.Examples.MNIST.TrainedGraph</a></td></tr><tr><td class="src">bitcast</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:bitcast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">boolVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:boolVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">broadcastGradientArgs</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:broadcastGradientArgs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:broadcastGradientArgs">TensorFlow.Ops</a></td></tr><tr><td class="src">Build</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">TensorFlow.Build</a></td></tr><tr><td class="src">build</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:build">TensorFlow.Session</a></td></tr><tr><td class="src">buildAnd</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:buildAnd">TensorFlow.Session</a></td></tr><tr><td class="src">buildCostModel</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:buildCostModel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">buildCostModelAfter</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:buildCostModelAfter">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">buildListOp</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-BuildOp.html#v:buildListOp">TensorFlow.BuildOp</a></td></tr><tr><td class="src">BuildOp</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-BuildOp.html#t:BuildOp">TensorFlow.BuildOp</a></td></tr><tr><td class="src">buildOp</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-BuildOp.html#v:buildOp">TensorFlow.BuildOp</a></td></tr><tr><td class="src">BuildT</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#t:BuildT">TensorFlow.Build</a></td></tr><tr><td class="src">buildWithSummary</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:buildWithSummary">TensorFlow.Session</a></td></tr><tr><td class="src">cast</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:cast">TensorFlow.Ops</a></td></tr><tr><td class="src">ceil</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:ceil">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">checkEndian</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html#v:checkEndian">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">checkNumerics</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:checkNumerics">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cholesky</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cholesky">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">choleskyGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:choleskyGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">collectAllSummaries</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:collectAllSummaries">TensorFlow.Build</a></td></tr><tr><td class="src">colocateWith</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:colocateWith">TensorFlow.Build</a></td></tr><tr><td class="src">complex</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:complex">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">complexAbs</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:complexAbs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">computeAccidentalHits</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:computeAccidentalHits">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">concat</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:concat">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:concat">TensorFlow.Ops</a></td></tr><tr><td class="src">concatOffset</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:concatOffset">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">ConfigProto</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:ConfigProto">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">ConfigProto'DeviceCountEntry</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:ConfigProto-39-DeviceCountEntry">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto-39-DeviceCountEntry">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">conj</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conj">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">const</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:const">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">constant</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:constant">TensorFlow.Ops</a></td></tr><tr><td class="src">container</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:container">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">ControlNode</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:ControlNode">TensorFlow.Output</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:ControlNode">TensorFlow.Build</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">TensorFlow.Output</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#t:ControlNode">TensorFlow.Build</a></td></tr><tr><td class="src">controlTrigger</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:controlTrigger">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv2D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conv2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv2DBackpropFilter</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conv2DBackpropFilter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv2DBackpropInput</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conv2DBackpropInput">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conv3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3DBackpropFilter</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conv3DBackpropFilter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3DBackpropFilterV2</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conv3DBackpropFilterV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3DBackpropInput</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conv3DBackpropInput">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3DBackpropInputV2</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conv3DBackpropInputV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">copy</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:copy">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">copyHost</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:copyHost">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cos</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cos">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">costGraph</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:costGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">countUpTo</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:countUpTo">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cropAndResize</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cropAndResize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cropAndResizeGradBoxes</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cropAndResizeGradBoxes">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cropAndResizeGradImage</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cropAndResizeGradImage">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cross</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cross">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cTCBeamSearchDecoder</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cTCBeamSearchDecoder">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cTCGreedyDecoder</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cTCGreedyDecoder">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cTCLoss</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cTCLoss">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cumprod</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cumprod">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cumsum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cumsum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">DataType</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#t:DataType">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">dcomplexVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:dcomplexVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">debugIdentity</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:debugIdentity">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">debugNanCount</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:debugNanCount">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">debugOps</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:debugOps">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">DebugTensorWatch</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:DebugTensorWatch">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">debugTensorWatchOpts</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:debugTensorWatchOpts">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">debugUrls</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:debugUrls">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">decodeBase64</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:decodeBase64">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeGif</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:decodeGif">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeJpeg</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:decodeJpeg">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeJSONExample</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:decodeJSONExample">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodePng</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:decodePng">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeRaw</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:decodeRaw">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeTensorData</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:decodeTensorData">TensorFlow.Types</a></td></tr><tr><td class="src">defaultValue</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:defaultValue">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">deferredDeletionBytes</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:deferredDeletionBytes">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">Delete</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:Delete">TensorFlow.Types</a></td></tr><tr><td class="src">deleteSessionTensor</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:deleteSessionTensor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">deprecation</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:deprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">depthToSpace</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:depthToSpace">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">depthwiseConv2dNative</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:depthwiseConv2dNative">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">depthwiseConv2dNativeBackpropFilter</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:depthwiseConv2dNativeBackpropFilter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">depthwiseConv2dNativeBackpropInput</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:depthwiseConv2dNativeBackpropInput">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dequeue</td><td class="module"><a href="./tensorflow-queue-0.1.0.0/TensorFlow-Queue.html#v:dequeue">TensorFlow.Queue</a></td></tr><tr><td class="src">description</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:description">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">deserializeManySparse</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:deserializeManySparse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">destroyTemporaryVariable</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:destroyTemporaryVariable">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Device</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:Device">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:Device">TensorFlow.Output</a></td></tr><tr><td class="src">device</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:device">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:device">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">deviceCount</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:deviceCount">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">deviceFilters</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:deviceFilters">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">deviceName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:deviceName">TensorFlow.Output</a></td></tr><tr><td class="src">diag</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:diag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">diagPart</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:diagPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">digamma</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:digamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dilation2D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:dilation2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dilation2DBackpropFilter</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:dilation2DBackpropFilter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dilation2DBackpropInput</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:dilation2DBackpropInput">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dim</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:dim">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">div</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:div">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">doCommonSubexpressionElimination</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:doCommonSubexpressionElimination">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">doConstantFolding</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:doConstantFolding">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">docOpList</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html#v:docOpList">TensorFlow.OpGen</a></td></tr><tr><td class="src">doFunctionInlining</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:doFunctionInlining">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">doubleVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:doubleVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">drawBoundingBoxes</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:drawBoundingBoxes">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">drawMNIST</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html#v:drawMNIST">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">dtype</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:dtype">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">DT_BFLOAT16</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_BFLOAT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_BFLOAT16_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_BFLOAT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_BOOL</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_BOOL">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_BOOL_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_BOOL_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_COMPLEX128</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_COMPLEX128">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_COMPLEX128_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_COMPLEX128_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_COMPLEX64</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_COMPLEX64">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_COMPLEX64_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_COMPLEX64_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_DOUBLE</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_DOUBLE">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_DOUBLE_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_DOUBLE_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_FLOAT</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_FLOAT">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_FLOAT_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_FLOAT_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_HALF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_HALF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_HALF_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_HALF_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT16</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT16_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT32</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT32">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT32_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT32_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT64</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT64">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT64_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT64_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT8</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT8">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT8_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT8_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INVALID</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INVALID">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT16</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT16_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT32</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT32">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT32_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT32_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT8</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT8">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT8_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT8_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QUINT16</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QUINT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QUINT16_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QUINT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QUINT8</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QUINT8">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QUINT8_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QUINT8_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_RESOURCE</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_RESOURCE">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_RESOURCE_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_RESOURCE_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_STRING</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_STRING">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_STRING_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_STRING_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_UINT16</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_UINT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_UINT16_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_UINT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_UINT8</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_UINT8">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_UINT8_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_UINT8_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">dynamicPartition</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:dynamicPartition">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dynamicStitch</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:dynamicStitch">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">editDistance</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:editDistance">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">elu</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:elu">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">eluGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:eluGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">embeddingLookup</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-EmbeddingOps.html#v:embeddingLookup">TensorFlow.EmbeddingOps</a></td></tr><tr><td class="src">enableBfloat16Sendrecv</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:enableBfloat16Sendrecv">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">enableRecvScheduling</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:enableRecvScheduling">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">encodeBase64</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:encodeBase64">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">encodeJpeg</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:encodeJpeg">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">encodePng</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:encodePng">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">encodeTensorData</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:encodeTensorData">TensorFlow.Types</a></td></tr><tr><td class="src">enqueue</td><td class="module"><a href="./tensorflow-queue-0.1.0.0/TensorFlow-Queue.html#v:enqueue">TensorFlow.Queue</a></td></tr><tr><td class="src">enter</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:enter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">eqLengthGuard</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-BuildOp.html#v:eqLengthGuard">TensorFlow.BuildOp</a></td></tr><tr><td class="src">equal</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:equal">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">erf</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:erf">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">erfc</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:erfc">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">evalBuildT</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:evalBuildT">TensorFlow.Build</a></td></tr><tr><td class="src">ExcludedCase</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:ExcludedCase">TensorFlow.Types</a></td></tr><tr><td class="src">excludeList</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html#v:excludeList">TensorFlow.OpGen</a></td></tr><tr><td class="src">exit</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:exit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">exp</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:exp">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">expandDims</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:expandDims">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:expandDims">TensorFlow.Ops</a></td></tr><tr><td class="src">explanation</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:explanation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">ExplicitName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:ExplicitName">TensorFlow.Output</a></td></tr><tr><td class="src">explicitName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:explicitName">TensorFlow.Build</a></td></tr><tr><td class="src">extend</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:extend">TensorFlow.Session</a></td></tr><tr><td class="src">extendGraph</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:extendGraph">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">extractGlimpse</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:extractGlimpse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">extractImagePatches</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:extractImagePatches">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">f</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:f">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">fact</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fact">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Feed</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:Feed">TensorFlow.Tensor</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Feed">TensorFlow.Tensor</a></td></tr><tr><td class="src">feed</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:feed">TensorFlow.Tensor</a></td></tr><tr><td class="src">Fetch</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:Fetch">TensorFlow.Nodes</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#t:Fetch">TensorFlow.Nodes</a></td></tr><tr><td class="src">Fetchable</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#t:Fetchable">TensorFlow.Nodes</a></td></tr><tr><td class="src">fetches</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:fetches">TensorFlow.Nodes</a></td></tr><tr><td class="src">fetchRestore</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:fetchRestore">TensorFlow.Nodes</a></td></tr><tr><td class="src">fetchTensorList</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:fetchTensorList">TensorFlow.Nodes</a></td></tr><tr><td class="src">fetchTensorVector</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:fetchTensorVector">TensorFlow.Nodes</a></td></tr><tr><td class="src">fFT</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fFT">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fFT2D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fFT2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fFT3D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fFT3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fIFOQueue</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fIFOQueue">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fill</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fill">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:fill">TensorFlow.Ops</a></td></tr><tr><td class="src">fixedLengthRecordReader</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fixedLengthRecordReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fixedUnigramCandidateSampler</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fixedUnigramCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">flagParser</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html#v:flagParser">TensorFlow.OpGen</a></td></tr><tr><td class="src">floatVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:floatVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">floor</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:floor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">flushInitializers</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:flushInitializers">TensorFlow.Build</a></td></tr><tr><td class="src">flushNodeBuffer</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:flushNodeBuffer">TensorFlow.Build</a></td></tr><tr><td class="src">fractionalAvgPool</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fractionalAvgPool">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fractionalAvgPoolGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fractionalAvgPoolGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fractionalMaxPool</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fractionalMaxPool">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fractionalMaxPoolGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fractionalMaxPoolGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">func</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:func">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">fusedResizeAndPadConv2D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fusedResizeAndPadConv2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">gather</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:gather">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">gatherNd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:gatherNd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">getAllOpList</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:getAllOpList">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">getFetch</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:getFetch">TensorFlow.Nodes</a></td></tr><tr><td class="src">getNodes</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:getNodes">TensorFlow.Nodes</a></td></tr><tr><td class="src">getOrAddOp</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:getOrAddOp">TensorFlow.Build</a></td></tr><tr><td class="src">getSessionHandle</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:getSessionHandle">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">getSessionTensor</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:getSessionTensor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">getVarInt</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-VarInt.html#v:getVarInt">TensorFlow.Internal.VarInt</a></td></tr><tr><td class="src">GPUOptions</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:GPUOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">gpuOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:gpuOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">gradients</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Gradient.html#v:gradients">TensorFlow.Gradient</a></td></tr><tr><td class="src">GraphDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:GraphDef">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">GraphOptions</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:GraphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">graphOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:graphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">GraphState</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#t:GraphState">TensorFlow.Build</a></td></tr><tr><td class="src">greater</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:greater">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">greaterEqual</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:greaterEqual">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">group</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-ControlFlow.html#v:group">TensorFlow.ControlFlow</a></td></tr><tr><td class="src">halfVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:halfVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">hashCode</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:hashCode">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">hasMinimum</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:hasMinimum">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">histogramSummary</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:histogramSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">hoistBuildT</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:hoistBuildT">TensorFlow.Build</a></td></tr><tr><td class="src">hSVToRGB</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:hSVToRGB">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">i</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:i">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">identity</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-ControlFlow.html#v:identity">TensorFlow.ControlFlow</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:identity">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">identityReader</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:identityReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">iFFT</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:iFFT">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">iFFT2D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:iFFT2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">iFFT3D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:iFFT3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">igamma</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:igamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">igammac</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:igammac">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">imag</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:imag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">imageSummary</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:imageSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">immutableConst</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:immutableConst">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">ImplicitName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:ImplicitName">TensorFlow.Output</a></td></tr><tr><td class="src">implicitName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:implicitName">TensorFlow.Build</a></td></tr><tr><td class="src">inferShapes</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:inferShapes">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">initializedVariable</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:initializedVariable">TensorFlow.Ops</a></td></tr><tr><td class="src">initializeTable</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:initializeTable">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">initializeTableFromTextFile</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:initializeTableFromTextFile">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">input</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:input">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">inputArg</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:inputArg">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">int64Val</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:int64Val">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">interOpParallelismThreads</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:interOpParallelismThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">interOpThreadPool</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:interOpThreadPool">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">inTopK</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:inTopK">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">intraOpParallelismThreads</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:intraOpParallelismThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">intVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:intVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">inv</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:inv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">invertPermutation</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:invertPermutation">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">invGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:invGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">isAggregate</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:isAggregate">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">isCommutative</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:isCommutative">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">isFinite</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:isFinite">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">isInf</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:isInf">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">isNan</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:isNan">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">isRef</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:isRef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">isStateful</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:isStateful">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">isVariableInitialized</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:isVariableInitialized">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">key</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:key">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:key">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:key">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">l2Loss</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:l2Loss">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">learnedUnigramCandidateSampler</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:learnedUnigramCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">less</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:less">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lessEqual</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lessEqual">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lgamma</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lgamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">library</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:library">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">linSpace</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:linSpace">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">list</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:list">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">listDiff</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:listDiff">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">log</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:log">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logDevicePlacement</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:logDevicePlacement">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">logicalAnd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:logicalAnd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logicalNot</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:logicalNot">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logicalOr</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:logicalOr">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logSoftmax</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:logSoftmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logUniformCandidateSampler</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:logUniformCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableExport</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lookupTableExport">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableFind</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lookupTableFind">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableImport</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lookupTableImport">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableInsert</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lookupTableInsert">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableSize</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lookupTableSize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">loopCond</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:loopCond">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lRN</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lRN">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lRNGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lRNGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">makeQueue2</td><td class="module"><a href="./tensorflow-queue-0.1.0.0/TensorFlow-Queue.html#v:makeQueue2">TensorFlow.Queue</a></td></tr><tr><td class="src">matchingFiles</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matchingFiles">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matMul</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:matMul">TensorFlow.Ops</a></td></tr><tr><td class="src">matrixBandPart</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixBandPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixDeterminant</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixDeterminant">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixDiag</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixDiag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixDiagPart</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixDiagPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixInverse</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixInverse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixSetDiag</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixSetDiag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixSolve</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixSolve">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixSolveLs</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixSolveLs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixTriangularSolve</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixTriangularSolve">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matTranspose</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:matTranspose">TensorFlow.Ops</a></td></tr><tr><td class="src">max</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:max">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maximum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:maximum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPool</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:maxPool">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPool3D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:maxPool3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPool3DGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:maxPool3DGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPoolGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:maxPoolGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPoolGradWithArgmax</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:maxPoolGradWithArgmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPoolWithArgmax</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:maxPoolWithArgmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maybe'allowedValues</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:maybe-39-allowedValues">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">maybe'b</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-b">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'costGraph</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-costGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'defaultValue</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:maybe-39-defaultValue">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">maybe'deprecation</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:maybe-39-deprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">maybe'f</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-f">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'func</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-func">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'gpuOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-gpuOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'graphOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-graphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'i</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-i">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'library</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:maybe-39-library">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">maybe'list</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-list">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'optimizerOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-optimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'placeholder</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-placeholder">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe's</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-s">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'shape</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-shape">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'stepStats</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-stepStats">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'tensor</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-tensor">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'tensorShape</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:maybe-39-tensorShape">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">maybe'type'</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-type-39-">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'value</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:maybe-39-value">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-value">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'versions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:maybe-39-versions">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">maybeTypeName</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:maybeTypeName">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">mean</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:mean">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">merge</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:merge">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mergeSummary</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:mergeSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">min</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:min">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">minimum</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:minimum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:minimum">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">mirrorPad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:mirrorPad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mirrorPadGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:mirrorPadGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">MNIST</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html#t:MNIST">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">mnistPb</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-TrainedGraph.html#v:mnistPb">TensorFlow.Examples.MNIST.TrainedGraph</a></td></tr><tr><td class="src">mod</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:mod">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mul</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:mul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:mul">TensorFlow.Ops</a></td></tr><tr><td class="src">multinomial</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:multinomial">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">name</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:name">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:name">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:name">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">4 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:name">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="alt">5 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:name">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">NameAttrList</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:NameAttrList">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">NameAttrList'AttrEntry</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:NameAttrList-39-AttrEntry">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">named</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-ControlFlow.html#v:named">TensorFlow.ControlFlow</a></td></tr><tr><td class="src">neg</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:neg">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:neg">TensorFlow.Ops</a></td></tr><tr><td class="src">negTrain</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:negTrain">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">nextIteration</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:nextIteration">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">node</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:node">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">NodeDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:NodeDef">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">NodeDef'AttrEntry</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:NodeDef-39-AttrEntry">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">NodeName</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:NodeName">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:NodeName">TensorFlow.Output</a></td></tr><tr><td class="src">nodeName</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:nodeName">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">Nodes</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#t:Nodes">TensorFlow.Nodes</a></td></tr><tr><td class="src">nodesUnion</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:nodesUnion">TensorFlow.Nodes</a></td></tr><tr><td class="src">NoneOf</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:NoneOf">TensorFlow.Types</a></td></tr><tr><td class="src">nonMaxSuppression</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:nonMaxSuppression">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">noOp</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-ControlFlow.html#v:noOp">TensorFlow.ControlFlow</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:noOp">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">notEqual</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:notEqual">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">numberAttr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:numberAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">numThreads</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:numThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">oneHot</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:oneHot">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">OneOf</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">TensorFlow.Types</a></td></tr><tr><td class="src">Op</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:Op">TensorFlow.Output</a></td></tr><tr><td class="src">op</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:op">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:op">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">opAttr</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:opAttr">TensorFlow.Output</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:opAttr">TensorFlow.Build</a></td></tr><tr><td class="src">opControlInputs</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:opControlInputs">TensorFlow.Output</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:opControlInputs">TensorFlow.Build</a></td></tr><tr><td class="src">OpDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:OpDef">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:OpDef">TensorFlow.Output</a></td></tr><tr><td class="alt">3 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:OpDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">4 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">opDef</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:opDef">TensorFlow.Build</a></td></tr><tr><td class="src">OpDef'ArgDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:OpDef-39-ArgDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">OpDef'AttrDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:OpDef-39-AttrDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">opDefWithName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:opDefWithName">TensorFlow.Build</a></td></tr><tr><td class="src">OpDeprecation</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:OpDeprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">operationTimeoutInMs</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:operationTimeoutInMs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OpGenFlags</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html#v:OpGenFlags">TensorFlow.OpGen</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html#t:OpGenFlags">TensorFlow.OpGen</a></td></tr><tr><td class="src">opInputs</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:opInputs">TensorFlow.Output</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:opInputs">TensorFlow.Build</a></td></tr><tr><td class="src">OpList</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:OpList">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#t:OpList">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">opName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:opName">TensorFlow.Output</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:opName">TensorFlow.Build</a></td></tr><tr><td class="src">OpResult</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-BuildOp.html#t:OpResult">TensorFlow.BuildOp</a></td></tr><tr><td class="src">OptimizerOptions</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:OptimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">optimizerOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:optimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OptimizerOptions'L0</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:OptimizerOptions-39-L0">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OptimizerOptions'L1</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:OptimizerOptions-39-L1">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OptimizerOptions'Level</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions-39-Level">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">optLevel</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:optLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OpType</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:OpType">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:OpType">TensorFlow.Output</a></td></tr><tr><td class="src">opType</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:opType">TensorFlow.Output</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:opType">TensorFlow.Build</a></td></tr><tr><td class="src">opUnrendered</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:opUnrendered">TensorFlow.Output</a></td></tr><tr><td class="src">Output</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:Output">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:Output">TensorFlow.Output</a></td></tr><tr><td class="src">output</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:output">TensorFlow.Output</a></td></tr><tr><td class="src">outputArg</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:outputArg">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">outputFile</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html#v:outputFile">TensorFlow.OpGen</a></td></tr><tr><td class="src">outputIndex</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:outputIndex">TensorFlow.Output</a></td></tr><tr><td class="src">OutputIx</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:OutputIx">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:OutputIx">TensorFlow.Output</a></td></tr><tr><td class="src">outputOp</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:outputOp">TensorFlow.Output</a></td></tr><tr><td class="src">outputPartitionGraphs</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:outputPartitionGraphs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">outputSlot</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:outputSlot">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">pack</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:pack">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:pack">TensorFlow.Ops</a></td></tr><tr><td class="src">pad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:pad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">paddingFIFOQueue</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:paddingFIFOQueue">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">parameterizedTruncatedNormal</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:parameterizedTruncatedNormal">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">parseTensor</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:parseTensor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">partitionGraphs</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:partitionGraphs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">PendingNodeName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:PendingNodeName">TensorFlow.Output</a></td></tr><tr><td class="src">perProcessGpuMemoryFraction</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:perProcessGpuMemoryFraction">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">placeholder</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:placeholder">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:placeholder">TensorFlow.Ops</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:placeholder">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">placeholderWithDefault</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:placeholderWithDefault">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">placementPeriod</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:placementPeriod">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">placePrunedGraph</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:placePrunedGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">polygamma</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:polygamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">pow</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:pow">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">prefix</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html#v:prefix">TensorFlow.OpGen</a></td></tr><tr><td class="src">priorityQueue</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:priorityQueue">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">prod</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:prod">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">protoShape</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:protoShape">TensorFlow.Types</a></td></tr><tr><td class="src">putVarInt</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-VarInt.html#v:putVarInt">TensorFlow.Internal.VarInt</a></td></tr><tr><td class="src">quantizeAndDequantize</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:quantizeAndDequantize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Queue2</td><td class="module"><a href="./tensorflow-queue-0.1.0.0/TensorFlow-Queue.html#t:Queue2">TensorFlow.Queue</a></td></tr><tr><td class="src">queueClose</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:queueClose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">queueSize</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:queueSize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomCrop</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:randomCrop">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomGamma</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:randomGamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomShuffle</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:randomShuffle">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomShuffleQueue</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:randomShuffleQueue">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomStandardNormal</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:randomStandardNormal">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomUniform</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:randomUniform">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomUniformInt</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:randomUniformInt">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">range</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:range">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:range">TensorFlow.Ops</a></td></tr><tr><td class="src">rank</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:rank">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerNumRecordsProduced</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:readerNumRecordsProduced">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerNumWorkUnitsCompleted</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:readerNumWorkUnitsCompleted">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerRead</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:readerRead">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerReadUpTo</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:readerReadUpTo">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerReset</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:readerReset">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerRestoreState</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:readerRestoreState">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerSerializeState</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:readerSerializeState">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readFile</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:readFile">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readMessageFromFileOrDie</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html#v:readMessageFromFileOrDie">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">readMNISTLabels</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html#v:readMNISTLabels">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">readMNISTSamples</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html#v:readMNISTSamples">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">real</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:real">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reducedShape</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:reducedShape">TensorFlow.Ops</a></td></tr><tr><td class="src">reduceJoin</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:reduceJoin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Ref</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">TensorFlow.Tensor</a></td></tr><tr><td class="src">refEnter</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:refEnter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refExit</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:refExit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refIdentity</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:refIdentity">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">RefKind</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:RefKind">TensorFlow.Tensor</a></td></tr><tr><td class="src">refMerge</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:refMerge">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refNextIteration</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:refNextIteration">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refSelect</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:refSelect">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refSwitch</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:refSwitch">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">relu</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:relu">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:relu">TensorFlow.Ops</a></td></tr><tr><td class="src">relu6</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:relu6">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">relu6Grad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:relu6Grad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reluGrad</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:reluGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:reluGrad">TensorFlow.Ops</a></td></tr><tr><td class="src">render</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:render">TensorFlow.Build</a></td></tr><tr><td class="src">Rendered</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:Rendered">TensorFlow.Output</a></td></tr><tr><td class="src">renderedNodeDefs</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:renderedNodeDefs">TensorFlow.Build</a></td></tr><tr><td class="src">renderNodeName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:renderNodeName">TensorFlow.Build</a></td></tr><tr><td class="src">renderOutput</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:renderOutput">TensorFlow.Build</a></td></tr><tr><td class="src">reshape</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:reshape">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:reshape">TensorFlow.Ops</a></td></tr><tr><td class="src">resizeArea</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:resizeArea">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeBicubic</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:resizeBicubic">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeBilinear</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:resizeBilinear">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeBilinearGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:resizeBilinearGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeNearestNeighbor</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:resizeNearestNeighbor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeNearestNeighborGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:resizeNearestNeighborGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">ResourceHandle</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:ResourceHandle">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">resourceHandleVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:resourceHandleVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">restore</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:restore">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:restore">TensorFlow.Ops</a></td></tr><tr><td class="src">restoreFromName</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:restoreFromName">TensorFlow.Ops</a></td></tr><tr><td class="src">restoreSlice</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:restoreSlice">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reverse</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:reverse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reverseSequence</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:reverseSequence">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">rGBToHSV</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:rGBToHSV">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">rsqrt</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:rsqrt">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">rsqrtGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:rsqrtGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">run</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:run">TensorFlow.Session</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:run">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">runBuildT</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:runBuildT">TensorFlow.Build</a></td></tr><tr><td class="src">RunMetadata</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:RunMetadata">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'FULL_TRACE</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions-39-FULL_TRACE">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'HARDWARE_TRACE</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions-39-HARDWARE_TRACE">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'NO_TRACE</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions-39-NO_TRACE">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'SOFTWARE_TRACE</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions-39-SOFTWARE_TRACE">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'TraceLevel</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions-39-TraceLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">runSession</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:runSession">TensorFlow.Session</a></td></tr><tr><td class="src">runSessionWithOptions</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:runSessionWithOptions">TensorFlow.Session</a></td></tr><tr><td class="src">runWithFeeds</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:runWithFeeds">TensorFlow.Session</a></td></tr><tr><td class="src">runWithFeeds_</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:runWithFeeds_">TensorFlow.Session</a></td></tr><tr><td class="src">run_</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:run_">TensorFlow.Session</a></td></tr><tr><td class="src">s</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:s">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">sampleDistortedBoundingBox</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sampleDistortedBoundingBox">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">save</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:save">TensorFlow.Ops</a></td></tr><tr><td class="src">Scalar</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:Scalar">TensorFlow.Nodes</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#t:Scalar">TensorFlow.Nodes</a></td></tr><tr><td class="src">scalar</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:scalar">TensorFlow.Ops</a></td></tr><tr><td class="src">scalarSummary</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:scalarSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterAdd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:scatterAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterDiv</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:scatterDiv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterMul</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:scatterMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterSub</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:scatterSub">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterUpdate</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:scatterUpdate">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scomplexVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:scomplexVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">segmentMax</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:segmentMax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentMean</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:segmentMean">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentMin</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:segmentMin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentProd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:segmentProd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentSum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:segmentSum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">select</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:select">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">selfAdjointEig</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:selfAdjointEig">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">selfAdjointEigV2</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:selfAdjointEigV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">serializeManySparse</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:serializeManySparse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">serializeSparse</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:serializeSparse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Session</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#t:Session">TensorFlow.Session</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#t:Session">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">sessionConfig</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:sessionConfig">TensorFlow.Session</a></td></tr><tr><td class="src">sessionInterOpThreadPool</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:sessionInterOpThreadPool">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">SessionOption</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#t:SessionOption">TensorFlow.Session</a></td></tr><tr><td class="src">sessionTarget</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:sessionTarget">TensorFlow.Session</a></td></tr><tr><td class="src">setSessionConfig</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:setSessionConfig">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">setSessionTarget</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:setSessionTarget">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">Shape</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:Shape">TensorFlow.Types</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">TensorFlow.Types</a></td></tr><tr><td class="src">shape</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shape">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:shape">TensorFlow.Ops</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:shape">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">shapeN</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shapeN">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">shardedFilename</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shardedFilename">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">shardedFilespec</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shardedFilespec">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sigmoid</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sigmoid">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sigmoidGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sigmoidGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sign</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sign">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:sign">TensorFlow.Ops</a></td></tr><tr><td class="src">sin</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">size</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:size">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:size">TensorFlow.Ops</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:size">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">slice</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:slice">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softmax</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:softmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:softmax">TensorFlow.Ops</a></td></tr><tr><td class="src">softmaxCrossEntropyWithLogits</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:softmaxCrossEntropyWithLogits">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:softmaxCrossEntropyWithLogits">TensorFlow.Ops</a></td></tr><tr><td class="src">softplus</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:softplus">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softplusGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:softplusGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softsign</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:softsign">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softsignGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:softsignGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">spaceToBatch</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:spaceToBatch">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">spaceToBatchND</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:spaceToBatchND">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">spaceToDepth</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:spaceToDepth">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseAdd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseAddGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseAddGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyAdadelta</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseApplyAdadelta">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyAdagrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseApplyAdagrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyAdagradDA</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseApplyAdagradDA">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyFtrl</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseApplyFtrl">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyMomentum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseApplyMomentum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyProximalAdagrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseApplyProximalAdagrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyProximalGradientDescent</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseApplyProximalGradientDescent">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyRMSProp</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseApplyRMSProp">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseConcat</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseConcat">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseDenseCwiseAdd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseDenseCwiseAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseDenseCwiseDiv</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseDenseCwiseDiv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseDenseCwiseMul</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseDenseCwiseMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseMatMul</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseMatMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseReduceSum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseReduceSum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseReduceSumSparse</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseReduceSumSparse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseReorder</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseReorder">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseReshape</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseReshape">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentMean</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSegmentMean">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentMeanGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSegmentMeanGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentSqrtN</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSegmentSqrtN">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentSqrtNGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSegmentSqrtNGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentSum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSegmentSum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSoftmax</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSoftmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSoftmaxCrossEntropyWithLogits</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSoftmaxCrossEntropyWithLogits">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSparseMaximum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSparseMaximum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSparseMinimum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSparseMinimum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSplit</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSplit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseTensorDenseAdd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseTensorDenseAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseTensorDenseMatMul</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseTensorDenseMatMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseToDense</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseToDense">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:sparseToDense">TensorFlow.Ops</a></td></tr><tr><td class="src">split</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:split">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sqrt</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sqrt">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sqrtGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sqrtGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">square</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:square">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">squaredDifference</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:squaredDifference">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">squeeze</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:squeeze">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stackClose</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stackClose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stackPop</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stackPop">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stackPush</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stackPush">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stepStats</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:stepStats">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">stopGradient</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stopGradient">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stridedSlice</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stridedSlice">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stridedSliceAssign</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stridedSliceAssign">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stridedSliceGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stridedSliceGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringJoin</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stringJoin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringSplit</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stringSplit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringToHashBucket</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stringToHashBucket">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringToHashBucketFast</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stringToHashBucketFast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringToHashBucketStrong</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stringToHashBucketStrong">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringToNumber</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stringToNumber">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:stringVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">sub</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sub">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:sub">TensorFlow.Ops</a></td></tr><tr><td class="src">sum</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:sum">TensorFlow.Ops</a></td></tr><tr><td class="src">summary</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:summary">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">SummaryTensor</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#t:SummaryTensor">TensorFlow.Build</a></td></tr><tr><td class="src">svd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:svd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">switch</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:switch">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tan</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tan">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tanh</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tanh">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tanhGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tanhGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Template</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#t:Template">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">templateDefault</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:templateDefault">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">templateRestrictions</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:templateRestrictions">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">temporaryVariable</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:temporaryVariable">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Tensor</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:Tensor">TensorFlow.Tensor</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">TensorFlow.Tensor</a></td></tr><tr><td class="src">tensor</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:tensor">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">tensorArrayClose</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayClose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayConcat</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayConcat">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayGather</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayGather">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayPack</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayPack">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayRead</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayRead">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayScatter</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayScatter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArraySize</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArraySize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArraySplit</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArraySplit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayUnpack</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayUnpack">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayWrite</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayWrite">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorAttr</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:tensorAttr">TensorFlow.Tensor</a></td></tr><tr><td class="src">tensorContent</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:tensorContent">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">TensorData</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:TensorData">TensorFlow.Types</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorData">TensorFlow.Types</a></td></tr><tr><td class="alt">3 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:TensorData">TensorFlow.Internal.FFI</a></td></tr><tr><td class="alt">4 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#t:TensorData">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">tensorDataBytes</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:tensorDataBytes">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">tensorDataDimensions</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:tensorDataDimensions">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">tensorDataType</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:tensorDataType">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">TensorFlowException</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:TensorFlowException">TensorFlow.Internal.FFI</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#t:TensorFlowException">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">tensorFromName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:tensorFromName">TensorFlow.Tensor</a></td></tr><tr><td class="src">TensorKind</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:TensorKind">TensorFlow.Tensor</a></td></tr><tr><td class="src">tensorKind</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:tensorKind">TensorFlow.Tensor</a></td></tr><tr><td class="src">tensorOutput</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:tensorOutput">TensorFlow.Tensor</a></td></tr><tr><td class="src">TensorProto</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:TensorProto">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">tensorRefType</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:tensorRefType">TensorFlow.Types</a></td></tr><tr><td class="src">tensorShape</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:tensorShape">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">TensorShapeProto</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:TensorShapeProto">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">TensorShapeProto'Dim</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:TensorShapeProto-39-Dim">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">tensorSummary</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">TensorType</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorFlow.Types</a></td></tr><tr><td class="src">tensorType</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:tensorType">TensorFlow.Types</a></td></tr><tr><td class="src">TensorTypes</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorTypes">TensorFlow.Types</a></td></tr><tr><td class="src">tensorVal</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:tensorVal">TensorFlow.Types</a></td></tr><tr><td class="src">testImageData</td><td class="module"><a href="./tensorflow-mnist-input-data-0.1.0.0/TensorFlow-Examples-MNIST-InputData.html#v:testImageData">TensorFlow.Examples.MNIST.InputData</a></td></tr><tr><td class="src">testLabelData</td><td class="module"><a href="./tensorflow-mnist-input-data-0.1.0.0/TensorFlow-Examples-MNIST-InputData.html#v:testLabelData">TensorFlow.Examples.MNIST.InputData</a></td></tr><tr><td class="src">textLineReader</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:textLineReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tFRecordReader</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tFRecordReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">ThreadPoolOptionProto</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:ThreadPoolOptionProto">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:ThreadPoolOptionProto">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">threadUnsafeUnigramCandidateSampler</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:threadUnsafeUnigramCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tile</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tile">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tileGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tileGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">timelineStep</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:timelineStep">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">timeoutInMs</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:timeoutInMs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">topK</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:topK">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:topK">TensorFlow.Ops</a></td></tr><tr><td class="src">topKV2</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:topKV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">traceLevel</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:traceLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">trainingImageData</td><td class="module"><a href="./tensorflow-mnist-input-data-0.1.0.0/TensorFlow-Examples-MNIST-InputData.html#v:trainingImageData">TensorFlow.Examples.MNIST.InputData</a></td></tr><tr><td class="src">trainingLabelData</td><td class="module"><a href="./tensorflow-mnist-input-data-0.1.0.0/TensorFlow-Examples-MNIST-InputData.html#v:trainingLabelData">TensorFlow.Examples.MNIST.InputData</a></td></tr><tr><td class="src">transpose</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:transpose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:transpose">TensorFlow.Ops</a></td></tr><tr><td class="src">truncatedNormal</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:truncatedNormal">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:truncatedNormal">TensorFlow.Ops</a></td></tr><tr><td class="src">type'</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:type-39-">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:type-39-">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">typeAttr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:typeAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">TypeError</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:TypeError">TensorFlow.Types</a></td></tr><tr><td class="src">typeListAttr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:typeListAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">unControlNode</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:unControlNode">TensorFlow.Output</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:unControlNode">TensorFlow.Build</a></td></tr><tr><td class="src">uniformCandidateSampler</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:uniformCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Unique</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#t:Unique">TensorFlow.Build</a></td></tr><tr><td class="src">unique</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:unique">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">uniqueWithCounts</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:uniqueWithCounts">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">unknownRank</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:unknownRank">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">unNodeName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:unNodeName">TensorFlow.Output</a></td></tr><tr><td class="src">unOpType</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:unOpType">TensorFlow.Output</a></td></tr><tr><td class="src">unOutputIx</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:unOutputIx">TensorFlow.Output</a></td></tr><tr><td class="src">unpack</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:unpack">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Unrendered</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:Unrendered">TensorFlow.Output</a></td></tr><tr><td class="src">unScalar</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:unScalar">TensorFlow.Nodes</a></td></tr><tr><td class="src">unsortedSegmentSum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:unsortedSegmentSum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">unTensorData</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:unTensorData">TensorFlow.Types</a></td></tr><tr><td class="src">usePerSessionThreads</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:usePerSessionThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">useProtoAsVoidPtrLen</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:useProtoAsVoidPtrLen">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">Value</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">TensorFlow.Tensor</a></td></tr><tr><td class="src">value</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">TensorFlow.Tensor</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:value">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:value">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">4 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:value">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">ValueKind</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:ValueKind">TensorFlow.Tensor</a></td></tr><tr><td class="src">variable</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:variable">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:variable">TensorFlow.Ops</a></td></tr><tr><td class="src">vector</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:vector">TensorFlow.Ops</a></td></tr><tr><td class="src">version</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:version">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:version">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">versionNumber</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:versionNumber">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">versions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:versions">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">visibleDeviceList</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:visibleDeviceList">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">where'</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:where-39-">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">wholeFileReader</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:wholeFileReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">withControlDependencies</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-ControlFlow.html#v:withControlDependencies">TensorFlow.ControlFlow</a></td></tr><tr><td class="src">withDevice</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:withDevice">TensorFlow.Build</a></td></tr><tr><td class="src">withNameScope</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:withNameScope">TensorFlow.Build</a></td></tr><tr><td class="src">withNodeDependencies</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:withNodeDependencies">TensorFlow.Build</a></td></tr><tr><td class="src">withSession</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:withSession">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">withStateLens</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:withStateLens">TensorFlow.Build</a></td></tr><tr><td class="src">wtsCkpt</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-TrainedGraph.html#v:wtsCkpt">TensorFlow.Examples.MNIST.TrainedGraph</a></td></tr><tr><td class="src">zeroInitializedVariable</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:zeroInitializedVariable">TensorFlow.Ops</a></td></tr><tr><td class="src">zeros</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:zeros">TensorFlow.Ops</a></td></tr><tr><td class="src">zerosLike</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:zerosLike">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:zerosLike">TensorFlow.Ops</a></td></tr><tr><td class="src">zeta</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:zeta">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">\\</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:-92--92-">TensorFlow.Types</a></td></tr><tr><td class="src">_Arg</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:_Arg">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_AttrValue'b</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-b">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'f</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-f">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'func</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-func">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'i</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-i">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'list</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-list">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'b</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-b">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'f</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-f">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'i</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-i">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue's</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-s">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'shape</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-shape">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'tensor</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-tensor">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'type'</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-type-39-">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'placeholder</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-placeholder">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue's</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-s">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'shape</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-shape">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'tensor</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-tensor">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'type'</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-type-39-">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_ConfigProto'allowSoftPlacement</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-allowSoftPlacement">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'deviceCount</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-deviceCount">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'DeviceCountEntry'key</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-DeviceCountEntry-39-key">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'DeviceCountEntry'value</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-DeviceCountEntry-39-value">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'deviceFilters</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-deviceFilters">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'gpuOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-gpuOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'graphOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-graphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'interOpParallelismThreads</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-interOpParallelismThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'intraOpParallelismThreads</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-intraOpParallelismThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'logDevicePlacement</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-logDevicePlacement">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'operationTimeoutInMs</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-operationTimeoutInMs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'placementPeriod</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-placementPeriod">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'sessionInterOpThreadPool</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-sessionInterOpThreadPool">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'usePerSessionThreads</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-usePerSessionThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_DebugTensorWatch'debugOps</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_DebugTensorWatch-39-debugOps">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_DebugTensorWatch'debugUrls</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_DebugTensorWatch-39-debugUrls">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_DebugTensorWatch'nodeName</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_DebugTensorWatch-39-nodeName">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_DebugTensorWatch'outputSlot</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_DebugTensorWatch-39-outputSlot">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'allocatorType</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-allocatorType">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'allowGrowth</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-allowGrowth">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'deferredDeletionBytes</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-deferredDeletionBytes">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'perProcessGpuMemoryFraction</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-perProcessGpuMemoryFraction">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'visibleDeviceList</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-visibleDeviceList">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphDef'library</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:_GraphDef-39-library">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">_GraphDef'node</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:_GraphDef-39-node">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">_GraphDef'version</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:_GraphDef-39-version">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">_GraphDef'versions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:_GraphDef-39-versions">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">_GraphOptions'buildCostModel</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-buildCostModel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'buildCostModelAfter</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-buildCostModelAfter">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'enableBfloat16Sendrecv</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-enableBfloat16Sendrecv">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'enableRecvScheduling</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-enableRecvScheduling">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'inferShapes</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-inferShapes">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'optimizerOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-optimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'placePrunedGraph</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-placePrunedGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'timelineStep</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-timelineStep">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_HostCast</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:_HostCast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_HostRecv</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:_HostRecv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_HostSend</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:_HostSend">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_NameAttrList'attr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_NameAttrList-39-attr">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_NameAttrList'AttrEntry'key</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_NameAttrList-39-AttrEntry-39-key">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_NameAttrList'AttrEntry'value</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_NameAttrList-39-AttrEntry-39-value">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_NameAttrList'name</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:_NameAttrList-39-name">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_NodeDef'attr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-attr">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'AttrEntry'key</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-AttrEntry-39-key">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'AttrEntry'value</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-AttrEntry-39-value">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'device</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-device">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'input</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-input">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'name</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-name">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'op</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-op">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_opAttrs</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:_opAttrs">TensorFlow.Output</a></td></tr><tr><td class="src">_opControlInputs</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:_opControlInputs">TensorFlow.Output</a></td></tr><tr><td class="src">_OpDef'allowsUninitializedInput</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-allowsUninitializedInput">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'description</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-description">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'isRef</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-isRef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'name</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-name">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'numberAttr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-numberAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'type'</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-type-39-">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'typeAttr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-typeAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'typeListAttr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-typeListAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'attr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-attr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'allowedValues</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-allowedValues">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'defaultValue</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-defaultValue">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'description</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-description">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'hasMinimum</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-hasMinimum">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'minimum</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-minimum">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'name</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-name">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'type'</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-type-39-">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'deprecation</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-deprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'description</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-description">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'inputArg</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-inputArg">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'isAggregate</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-isAggregate">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'isCommutative</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-isCommutative">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'isStateful</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-isStateful">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'name</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-name">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'outputArg</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-outputArg">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'summary</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-summary">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDeprecation'explanation</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDeprecation-39-explanation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDeprecation'version</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDeprecation-39-version">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_opInputs</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:_opInputs">TensorFlow.Output</a></td></tr><tr><td class="src">_OpList'op</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpList-39-op">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_opName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:_opName">TensorFlow.Output</a></td></tr><tr><td class="src">_OptimizerOptions'doCommonSubexpressionElimination</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_OptimizerOptions-39-doCommonSubexpressionElimination">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_OptimizerOptions'doConstantFolding</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_OptimizerOptions-39-doConstantFolding">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_OptimizerOptions'doFunctionInlining</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_OptimizerOptions-39-doFunctionInlining">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_OptimizerOptions'optLevel</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_OptimizerOptions-39-optLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_opType</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:_opType">TensorFlow.Output</a></td></tr><tr><td class="src">_Recv</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:_Recv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_ResourceHandle'container</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-container">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_ResourceHandle'device</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-device">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_ResourceHandle'hashCode</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-hashCode">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_ResourceHandle'maybeTypeName</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-maybeTypeName">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_ResourceHandle'name</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-name">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_Retval</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:_Retval">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_RunMetadata'costGraph</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunMetadata-39-costGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunMetadata'partitionGraphs</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunMetadata-39-partitionGraphs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunMetadata'stepStats</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunMetadata-39-stepStats">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'debugTensorWatchOpts</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-debugTensorWatchOpts">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'interOpThreadPool</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-interOpThreadPool">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'outputPartitionGraphs</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-outputPartitionGraphs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'timeoutInMs</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-timeoutInMs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'traceLevel</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-traceLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_Send</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:_Send">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_TensorProto'boolVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-boolVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'dcomplexVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-dcomplexVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'doubleVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-doubleVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'dtype</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-dtype">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'floatVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-floatVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'halfVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-halfVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'int64Val</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-int64Val">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'intVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-intVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'resourceHandleVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-resourceHandleVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'scomplexVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-scomplexVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'stringVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-stringVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'tensorContent</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-tensorContent">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'tensorShape</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-tensorShape">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'versionNumber</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-versionNumber">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorShapeProto'dim</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:_TensorShapeProto-39-dim">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">_TensorShapeProto'Dim'name</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:_TensorShapeProto-39-Dim-39-name">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">_TensorShapeProto'Dim'size</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:_TensorShapeProto-39-Dim-39-size">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">_TensorShapeProto'unknownRank</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:_TensorShapeProto-39-unknownRank">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">_ThreadPoolOptionProto'numThreads</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:_ThreadPoolOptionProto-39-numThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-B.html b/docs/haddock/doc-index-B.html
new file mode 100644
index 0000000..417692a
--- /dev/null
+++ b/docs/haddock/doc-index-B.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - B)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - B</p><table><tr><td class="src">b</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:b">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">barrier</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:barrier">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrierClose</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:barrierClose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrierIncompleteSize</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:barrierIncompleteSize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrierInsertMany</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:barrierInsertMany">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrierReadySize</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:barrierReadySize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchCholesky</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchCholesky">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchCholeskyGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchCholeskyGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchFFT</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchFFT">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchFFT2D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchFFT2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchFFT3D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchFFT3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchIFFT</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchIFFT">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchIFFT2D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchIFFT2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchIFFT3D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchIFFT3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatMul</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixBandPart</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixBandPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixDeterminant</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixDeterminant">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixDiag</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixDiag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixDiagPart</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixDiagPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixInverse</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixInverse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixSetDiag</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixSetDiag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixSolve</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixSolve">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixSolveLs</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixSolveLs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixTriangularSolve</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchMatrixTriangularSolve">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchNormWithGlobalNormalization</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchNormWithGlobalNormalization">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchNormWithGlobalNormalizationGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchNormWithGlobalNormalizationGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchSelfAdjointEig</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchSelfAdjointEig">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchSelfAdjointEigV2</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchSelfAdjointEigV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchSvd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchSvd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchToSpace</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchToSpace">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchToSpaceND</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:batchToSpaceND">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">betainc</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:betainc">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">biasAdd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:biasAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">biasAddGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:biasAddGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">biasAddV1</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:biasAddV1">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">biasCkpt</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-TrainedGraph.html#v:biasCkpt">TensorFlow.Examples.MNIST.TrainedGraph</a></td></tr><tr><td class="src">bitcast</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:bitcast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">boolVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:boolVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">broadcastGradientArgs</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:broadcastGradientArgs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:broadcastGradientArgs">TensorFlow.Ops</a></td></tr><tr><td class="src">Build</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">TensorFlow.Build</a></td></tr><tr><td class="src">build</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:build">TensorFlow.Session</a></td></tr><tr><td class="src">buildAnd</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:buildAnd">TensorFlow.Session</a></td></tr><tr><td class="src">buildCostModel</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:buildCostModel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">buildCostModelAfter</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:buildCostModelAfter">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">buildListOp</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-BuildOp.html#v:buildListOp">TensorFlow.BuildOp</a></td></tr><tr><td class="src">BuildOp</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-BuildOp.html#t:BuildOp">TensorFlow.BuildOp</a></td></tr><tr><td class="src">buildOp</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-BuildOp.html#v:buildOp">TensorFlow.BuildOp</a></td></tr><tr><td class="src">BuildT</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#t:BuildT">TensorFlow.Build</a></td></tr><tr><td class="src">buildWithSummary</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:buildWithSummary">TensorFlow.Session</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-C.html b/docs/haddock/doc-index-C.html
new file mode 100644
index 0000000..e970f38
--- /dev/null
+++ b/docs/haddock/doc-index-C.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - C)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - C</p><table><tr><td class="src">cast</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:cast">TensorFlow.Ops</a></td></tr><tr><td class="src">ceil</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:ceil">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">checkEndian</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html#v:checkEndian">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">checkNumerics</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:checkNumerics">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cholesky</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cholesky">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">choleskyGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:choleskyGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">collectAllSummaries</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:collectAllSummaries">TensorFlow.Build</a></td></tr><tr><td class="src">colocateWith</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:colocateWith">TensorFlow.Build</a></td></tr><tr><td class="src">complex</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:complex">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">complexAbs</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:complexAbs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">computeAccidentalHits</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:computeAccidentalHits">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">concat</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:concat">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:concat">TensorFlow.Ops</a></td></tr><tr><td class="src">concatOffset</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:concatOffset">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">ConfigProto</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:ConfigProto">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">ConfigProto'DeviceCountEntry</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:ConfigProto-39-DeviceCountEntry">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto-39-DeviceCountEntry">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">conj</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conj">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">const</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:const">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">constant</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:constant">TensorFlow.Ops</a></td></tr><tr><td class="src">container</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:container">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">ControlNode</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:ControlNode">TensorFlow.Output</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:ControlNode">TensorFlow.Build</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">TensorFlow.Output</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#t:ControlNode">TensorFlow.Build</a></td></tr><tr><td class="src">controlTrigger</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:controlTrigger">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv2D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conv2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv2DBackpropFilter</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conv2DBackpropFilter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv2DBackpropInput</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conv2DBackpropInput">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conv3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3DBackpropFilter</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conv3DBackpropFilter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3DBackpropFilterV2</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conv3DBackpropFilterV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3DBackpropInput</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conv3DBackpropInput">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3DBackpropInputV2</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:conv3DBackpropInputV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">copy</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:copy">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">copyHost</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:copyHost">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cos</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cos">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">costGraph</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:costGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">countUpTo</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:countUpTo">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cropAndResize</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cropAndResize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cropAndResizeGradBoxes</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cropAndResizeGradBoxes">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cropAndResizeGradImage</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cropAndResizeGradImage">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cross</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cross">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cTCBeamSearchDecoder</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cTCBeamSearchDecoder">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cTCGreedyDecoder</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cTCGreedyDecoder">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cTCLoss</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cTCLoss">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cumprod</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cumprod">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cumsum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:cumsum">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-D.html b/docs/haddock/doc-index-D.html
new file mode 100644
index 0000000..567d5a8
--- /dev/null
+++ b/docs/haddock/doc-index-D.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - D)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - D</p><table><tr><td class="src">DataType</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#t:DataType">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">dcomplexVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:dcomplexVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">debugIdentity</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:debugIdentity">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">debugNanCount</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:debugNanCount">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">debugOps</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:debugOps">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">DebugTensorWatch</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:DebugTensorWatch">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">debugTensorWatchOpts</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:debugTensorWatchOpts">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">debugUrls</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:debugUrls">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">decodeBase64</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:decodeBase64">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeGif</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:decodeGif">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeJpeg</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:decodeJpeg">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeJSONExample</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:decodeJSONExample">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodePng</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:decodePng">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeRaw</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:decodeRaw">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeTensorData</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:decodeTensorData">TensorFlow.Types</a></td></tr><tr><td class="src">defaultValue</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:defaultValue">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">deferredDeletionBytes</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:deferredDeletionBytes">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">Delete</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:Delete">TensorFlow.Types</a></td></tr><tr><td class="src">deleteSessionTensor</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:deleteSessionTensor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">deprecation</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:deprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">depthToSpace</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:depthToSpace">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">depthwiseConv2dNative</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:depthwiseConv2dNative">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">depthwiseConv2dNativeBackpropFilter</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:depthwiseConv2dNativeBackpropFilter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">depthwiseConv2dNativeBackpropInput</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:depthwiseConv2dNativeBackpropInput">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dequeue</td><td class="module"><a href="./tensorflow-queue-0.1.0.0/TensorFlow-Queue.html#v:dequeue">TensorFlow.Queue</a></td></tr><tr><td class="src">description</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:description">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">deserializeManySparse</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:deserializeManySparse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">destroyTemporaryVariable</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:destroyTemporaryVariable">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Device</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:Device">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:Device">TensorFlow.Output</a></td></tr><tr><td class="src">device</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:device">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:device">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">deviceCount</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:deviceCount">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">deviceFilters</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:deviceFilters">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">deviceName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:deviceName">TensorFlow.Output</a></td></tr><tr><td class="src">diag</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:diag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">diagPart</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:diagPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">digamma</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:digamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dilation2D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:dilation2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dilation2DBackpropFilter</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:dilation2DBackpropFilter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dilation2DBackpropInput</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:dilation2DBackpropInput">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dim</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:dim">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">div</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:div">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">doCommonSubexpressionElimination</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:doCommonSubexpressionElimination">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">doConstantFolding</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:doConstantFolding">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">docOpList</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html#v:docOpList">TensorFlow.OpGen</a></td></tr><tr><td class="src">doFunctionInlining</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:doFunctionInlining">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">doubleVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:doubleVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">drawBoundingBoxes</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:drawBoundingBoxes">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">drawMNIST</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html#v:drawMNIST">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">dtype</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:dtype">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">DT_BFLOAT16</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_BFLOAT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_BFLOAT16_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_BFLOAT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_BOOL</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_BOOL">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_BOOL_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_BOOL_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_COMPLEX128</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_COMPLEX128">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_COMPLEX128_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_COMPLEX128_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_COMPLEX64</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_COMPLEX64">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_COMPLEX64_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_COMPLEX64_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_DOUBLE</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_DOUBLE">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_DOUBLE_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_DOUBLE_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_FLOAT</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_FLOAT">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_FLOAT_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_FLOAT_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_HALF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_HALF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_HALF_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_HALF_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT16</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT16_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT32</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT32">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT32_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT32_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT64</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT64">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT64_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT64_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT8</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT8">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT8_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT8_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INVALID</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_INVALID">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT16</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT16_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT32</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT32">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT32_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT32_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT8</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT8">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT8_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT8_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QUINT16</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QUINT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QUINT16_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QUINT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QUINT8</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QUINT8">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QUINT8_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_QUINT8_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_RESOURCE</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_RESOURCE">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_RESOURCE_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_RESOURCE_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_STRING</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_STRING">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_STRING_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_STRING_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_UINT16</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_UINT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_UINT16_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_UINT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_UINT8</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_UINT8">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_UINT8_REF</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#v:DT_UINT8_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">dynamicPartition</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:dynamicPartition">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dynamicStitch</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:dynamicStitch">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-E.html b/docs/haddock/doc-index-E.html
new file mode 100644
index 0000000..2ce0f45
--- /dev/null
+++ b/docs/haddock/doc-index-E.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - E)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - E</p><table><tr><td class="src">editDistance</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:editDistance">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">elu</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:elu">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">eluGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:eluGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">embeddingLookup</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-EmbeddingOps.html#v:embeddingLookup">TensorFlow.EmbeddingOps</a></td></tr><tr><td class="src">enableBfloat16Sendrecv</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:enableBfloat16Sendrecv">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">enableRecvScheduling</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:enableRecvScheduling">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">encodeBase64</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:encodeBase64">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">encodeJpeg</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:encodeJpeg">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">encodePng</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:encodePng">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">encodeTensorData</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:encodeTensorData">TensorFlow.Types</a></td></tr><tr><td class="src">enqueue</td><td class="module"><a href="./tensorflow-queue-0.1.0.0/TensorFlow-Queue.html#v:enqueue">TensorFlow.Queue</a></td></tr><tr><td class="src">enter</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:enter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">eqLengthGuard</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-BuildOp.html#v:eqLengthGuard">TensorFlow.BuildOp</a></td></tr><tr><td class="src">equal</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:equal">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">erf</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:erf">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">erfc</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:erfc">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">evalBuildT</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:evalBuildT">TensorFlow.Build</a></td></tr><tr><td class="src">ExcludedCase</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:ExcludedCase">TensorFlow.Types</a></td></tr><tr><td class="src">excludeList</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html#v:excludeList">TensorFlow.OpGen</a></td></tr><tr><td class="src">exit</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:exit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">exp</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:exp">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">expandDims</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:expandDims">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:expandDims">TensorFlow.Ops</a></td></tr><tr><td class="src">explanation</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:explanation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">ExplicitName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:ExplicitName">TensorFlow.Output</a></td></tr><tr><td class="src">explicitName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:explicitName">TensorFlow.Build</a></td></tr><tr><td class="src">extend</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:extend">TensorFlow.Session</a></td></tr><tr><td class="src">extendGraph</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:extendGraph">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">extractGlimpse</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:extractGlimpse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">extractImagePatches</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:extractImagePatches">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-F.html b/docs/haddock/doc-index-F.html
new file mode 100644
index 0000000..8e297d2
--- /dev/null
+++ b/docs/haddock/doc-index-F.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - F)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - F</p><table><tr><td class="src">f</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:f">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">fact</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fact">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Feed</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:Feed">TensorFlow.Tensor</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Feed">TensorFlow.Tensor</a></td></tr><tr><td class="src">feed</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:feed">TensorFlow.Tensor</a></td></tr><tr><td class="src">Fetch</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:Fetch">TensorFlow.Nodes</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#t:Fetch">TensorFlow.Nodes</a></td></tr><tr><td class="src">Fetchable</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#t:Fetchable">TensorFlow.Nodes</a></td></tr><tr><td class="src">fetches</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:fetches">TensorFlow.Nodes</a></td></tr><tr><td class="src">fetchRestore</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:fetchRestore">TensorFlow.Nodes</a></td></tr><tr><td class="src">fetchTensorList</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:fetchTensorList">TensorFlow.Nodes</a></td></tr><tr><td class="src">fetchTensorVector</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:fetchTensorVector">TensorFlow.Nodes</a></td></tr><tr><td class="src">fFT</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fFT">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fFT2D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fFT2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fFT3D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fFT3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fIFOQueue</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fIFOQueue">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fill</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fill">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:fill">TensorFlow.Ops</a></td></tr><tr><td class="src">fixedLengthRecordReader</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fixedLengthRecordReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fixedUnigramCandidateSampler</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fixedUnigramCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">flagParser</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html#v:flagParser">TensorFlow.OpGen</a></td></tr><tr><td class="src">floatVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:floatVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">floor</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:floor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">flushInitializers</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:flushInitializers">TensorFlow.Build</a></td></tr><tr><td class="src">flushNodeBuffer</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:flushNodeBuffer">TensorFlow.Build</a></td></tr><tr><td class="src">fractionalAvgPool</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fractionalAvgPool">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fractionalAvgPoolGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fractionalAvgPoolGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fractionalMaxPool</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fractionalMaxPool">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fractionalMaxPoolGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fractionalMaxPoolGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">func</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:func">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">fusedResizeAndPadConv2D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:fusedResizeAndPadConv2D">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-G.html b/docs/haddock/doc-index-G.html
new file mode 100644
index 0000000..03eeac6
--- /dev/null
+++ b/docs/haddock/doc-index-G.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - G)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - G</p><table><tr><td class="src">gather</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:gather">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">gatherNd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:gatherNd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">getAllOpList</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:getAllOpList">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">getFetch</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:getFetch">TensorFlow.Nodes</a></td></tr><tr><td class="src">getNodes</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:getNodes">TensorFlow.Nodes</a></td></tr><tr><td class="src">getOrAddOp</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:getOrAddOp">TensorFlow.Build</a></td></tr><tr><td class="src">getSessionHandle</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:getSessionHandle">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">getSessionTensor</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:getSessionTensor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">getVarInt</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-VarInt.html#v:getVarInt">TensorFlow.Internal.VarInt</a></td></tr><tr><td class="src">GPUOptions</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:GPUOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">gpuOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:gpuOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">gradients</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Gradient.html#v:gradients">TensorFlow.Gradient</a></td></tr><tr><td class="src">GraphDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:GraphDef">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">GraphOptions</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:GraphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">graphOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:graphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">GraphState</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#t:GraphState">TensorFlow.Build</a></td></tr><tr><td class="src">greater</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:greater">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">greaterEqual</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:greaterEqual">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">group</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-ControlFlow.html#v:group">TensorFlow.ControlFlow</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-H.html b/docs/haddock/doc-index-H.html
new file mode 100644
index 0000000..34f5682
--- /dev/null
+++ b/docs/haddock/doc-index-H.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - H)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - H</p><table><tr><td class="src">halfVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:halfVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">hashCode</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:hashCode">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">hasMinimum</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:hasMinimum">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">histogramSummary</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:histogramSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">hoistBuildT</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:hoistBuildT">TensorFlow.Build</a></td></tr><tr><td class="src">hSVToRGB</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:hSVToRGB">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-I.html b/docs/haddock/doc-index-I.html
new file mode 100644
index 0000000..81c2dfc
--- /dev/null
+++ b/docs/haddock/doc-index-I.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - I)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - I</p><table><tr><td class="src">i</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:i">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">identity</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-ControlFlow.html#v:identity">TensorFlow.ControlFlow</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:identity">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">identityReader</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:identityReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">iFFT</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:iFFT">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">iFFT2D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:iFFT2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">iFFT3D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:iFFT3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">igamma</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:igamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">igammac</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:igammac">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">imag</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:imag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">imageSummary</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:imageSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">immutableConst</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:immutableConst">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">ImplicitName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:ImplicitName">TensorFlow.Output</a></td></tr><tr><td class="src">implicitName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:implicitName">TensorFlow.Build</a></td></tr><tr><td class="src">inferShapes</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:inferShapes">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">initializedVariable</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:initializedVariable">TensorFlow.Ops</a></td></tr><tr><td class="src">initializeTable</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:initializeTable">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">initializeTableFromTextFile</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:initializeTableFromTextFile">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">input</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:input">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">inputArg</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:inputArg">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">int64Val</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:int64Val">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">interOpParallelismThreads</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:interOpParallelismThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">interOpThreadPool</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:interOpThreadPool">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">inTopK</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:inTopK">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">intraOpParallelismThreads</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:intraOpParallelismThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">intVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:intVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">inv</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:inv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">invertPermutation</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:invertPermutation">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">invGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:invGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">isAggregate</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:isAggregate">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">isCommutative</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:isCommutative">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">isFinite</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:isFinite">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">isInf</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:isInf">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">isNan</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:isNan">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">isRef</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:isRef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">isStateful</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:isStateful">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">isVariableInitialized</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:isVariableInitialized">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-K.html b/docs/haddock/doc-index-K.html
new file mode 100644
index 0000000..d84e5d6
--- /dev/null
+++ b/docs/haddock/doc-index-K.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - K)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - K</p><table><tr><td class="src">key</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:key">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:key">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:key">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-L.html b/docs/haddock/doc-index-L.html
new file mode 100644
index 0000000..28d02bc
--- /dev/null
+++ b/docs/haddock/doc-index-L.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - L)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - L</p><table><tr><td class="src">l2Loss</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:l2Loss">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">learnedUnigramCandidateSampler</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:learnedUnigramCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">less</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:less">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lessEqual</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lessEqual">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lgamma</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lgamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">library</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:library">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">linSpace</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:linSpace">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">list</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:list">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">listDiff</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:listDiff">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">log</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:log">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logDevicePlacement</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:logDevicePlacement">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">logicalAnd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:logicalAnd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logicalNot</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:logicalNot">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logicalOr</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:logicalOr">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logSoftmax</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:logSoftmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logUniformCandidateSampler</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:logUniformCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableExport</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lookupTableExport">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableFind</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lookupTableFind">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableImport</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lookupTableImport">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableInsert</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lookupTableInsert">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableSize</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lookupTableSize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">loopCond</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:loopCond">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lRN</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lRN">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lRNGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:lRNGrad">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-M.html b/docs/haddock/doc-index-M.html
new file mode 100644
index 0000000..41df9e3
--- /dev/null
+++ b/docs/haddock/doc-index-M.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - M)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - M</p><table><tr><td class="src">makeQueue2</td><td class="module"><a href="./tensorflow-queue-0.1.0.0/TensorFlow-Queue.html#v:makeQueue2">TensorFlow.Queue</a></td></tr><tr><td class="src">matchingFiles</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matchingFiles">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matMul</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:matMul">TensorFlow.Ops</a></td></tr><tr><td class="src">matrixBandPart</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixBandPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixDeterminant</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixDeterminant">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixDiag</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixDiag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixDiagPart</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixDiagPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixInverse</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixInverse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixSetDiag</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixSetDiag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixSolve</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixSolve">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixSolveLs</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixSolveLs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixTriangularSolve</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:matrixTriangularSolve">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matTranspose</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:matTranspose">TensorFlow.Ops</a></td></tr><tr><td class="src">max</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:max">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maximum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:maximum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPool</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:maxPool">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPool3D</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:maxPool3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPool3DGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:maxPool3DGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPoolGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:maxPoolGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPoolGradWithArgmax</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:maxPoolGradWithArgmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPoolWithArgmax</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:maxPoolWithArgmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maybe'allowedValues</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:maybe-39-allowedValues">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">maybe'b</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-b">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'costGraph</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-costGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'defaultValue</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:maybe-39-defaultValue">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">maybe'deprecation</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:maybe-39-deprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">maybe'f</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-f">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'func</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-func">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'gpuOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-gpuOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'graphOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-graphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'i</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-i">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'library</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:maybe-39-library">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">maybe'list</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-list">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'optimizerOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-optimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'placeholder</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-placeholder">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe's</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-s">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'shape</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-shape">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'stepStats</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-stepStats">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'tensor</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-tensor">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'tensorShape</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:maybe-39-tensorShape">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">maybe'type'</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-type-39-">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'value</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:maybe-39-value">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-value">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'versions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:maybe-39-versions">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">maybeTypeName</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:maybeTypeName">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">mean</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:mean">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">merge</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:merge">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mergeSummary</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:mergeSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">min</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:min">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">minimum</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:minimum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:minimum">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">mirrorPad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:mirrorPad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mirrorPadGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:mirrorPadGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">MNIST</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html#t:MNIST">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">mnistPb</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-TrainedGraph.html#v:mnistPb">TensorFlow.Examples.MNIST.TrainedGraph</a></td></tr><tr><td class="src">mod</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:mod">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mul</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:mul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:mul">TensorFlow.Ops</a></td></tr><tr><td class="src">multinomial</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:multinomial">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-N.html b/docs/haddock/doc-index-N.html
new file mode 100644
index 0000000..38cdec6
--- /dev/null
+++ b/docs/haddock/doc-index-N.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - N)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - N</p><table><tr><td class="src">name</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:name">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:name">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:name">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">4 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:name">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="alt">5 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:name">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">NameAttrList</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:NameAttrList">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">NameAttrList'AttrEntry</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:NameAttrList-39-AttrEntry">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">named</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-ControlFlow.html#v:named">TensorFlow.ControlFlow</a></td></tr><tr><td class="src">neg</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:neg">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:neg">TensorFlow.Ops</a></td></tr><tr><td class="src">negTrain</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:negTrain">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">nextIteration</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:nextIteration">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">node</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:node">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">NodeDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:NodeDef">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">NodeDef'AttrEntry</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:NodeDef-39-AttrEntry">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">NodeName</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:NodeName">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:NodeName">TensorFlow.Output</a></td></tr><tr><td class="src">nodeName</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:nodeName">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">Nodes</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#t:Nodes">TensorFlow.Nodes</a></td></tr><tr><td class="src">nodesUnion</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:nodesUnion">TensorFlow.Nodes</a></td></tr><tr><td class="src">NoneOf</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:NoneOf">TensorFlow.Types</a></td></tr><tr><td class="src">nonMaxSuppression</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:nonMaxSuppression">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">noOp</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-ControlFlow.html#v:noOp">TensorFlow.ControlFlow</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:noOp">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">notEqual</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:notEqual">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">numberAttr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:numberAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">numThreads</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:numThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-O.html b/docs/haddock/doc-index-O.html
new file mode 100644
index 0000000..fb6acda
--- /dev/null
+++ b/docs/haddock/doc-index-O.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - O)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - O</p><table><tr><td class="src">oneHot</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:oneHot">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">OneOf</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">TensorFlow.Types</a></td></tr><tr><td class="src">Op</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:Op">TensorFlow.Output</a></td></tr><tr><td class="src">op</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:op">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:op">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">opAttr</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:opAttr">TensorFlow.Output</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:opAttr">TensorFlow.Build</a></td></tr><tr><td class="src">opControlInputs</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:opControlInputs">TensorFlow.Output</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:opControlInputs">TensorFlow.Build</a></td></tr><tr><td class="src">OpDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:OpDef">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:OpDef">TensorFlow.Output</a></td></tr><tr><td class="alt">3 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:OpDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">4 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">opDef</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:opDef">TensorFlow.Build</a></td></tr><tr><td class="src">OpDef'ArgDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:OpDef-39-ArgDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">OpDef'AttrDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:OpDef-39-AttrDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">opDefWithName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:opDefWithName">TensorFlow.Build</a></td></tr><tr><td class="src">OpDeprecation</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:OpDeprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">operationTimeoutInMs</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:operationTimeoutInMs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OpGenFlags</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html#v:OpGenFlags">TensorFlow.OpGen</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html#t:OpGenFlags">TensorFlow.OpGen</a></td></tr><tr><td class="src">opInputs</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:opInputs">TensorFlow.Output</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:opInputs">TensorFlow.Build</a></td></tr><tr><td class="src">OpList</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:OpList">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#t:OpList">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">opName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:opName">TensorFlow.Output</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:opName">TensorFlow.Build</a></td></tr><tr><td class="src">OpResult</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-BuildOp.html#t:OpResult">TensorFlow.BuildOp</a></td></tr><tr><td class="src">OptimizerOptions</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:OptimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">optimizerOptions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:optimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OptimizerOptions'L0</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:OptimizerOptions-39-L0">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OptimizerOptions'L1</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:OptimizerOptions-39-L1">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OptimizerOptions'Level</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions-39-Level">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">optLevel</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:optLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OpType</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:OpType">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:OpType">TensorFlow.Output</a></td></tr><tr><td class="src">opType</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:opType">TensorFlow.Output</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:opType">TensorFlow.Build</a></td></tr><tr><td class="src">opUnrendered</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:opUnrendered">TensorFlow.Output</a></td></tr><tr><td class="src">Output</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:Output">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:Output">TensorFlow.Output</a></td></tr><tr><td class="src">output</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:output">TensorFlow.Output</a></td></tr><tr><td class="src">outputArg</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:outputArg">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">outputFile</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html#v:outputFile">TensorFlow.OpGen</a></td></tr><tr><td class="src">outputIndex</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:outputIndex">TensorFlow.Output</a></td></tr><tr><td class="src">OutputIx</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:OutputIx">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:OutputIx">TensorFlow.Output</a></td></tr><tr><td class="src">outputOp</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:outputOp">TensorFlow.Output</a></td></tr><tr><td class="src">outputPartitionGraphs</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:outputPartitionGraphs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">outputSlot</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:outputSlot">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-P.html b/docs/haddock/doc-index-P.html
new file mode 100644
index 0000000..f1ecc52
--- /dev/null
+++ b/docs/haddock/doc-index-P.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - P)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - P</p><table><tr><td class="src">pack</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:pack">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:pack">TensorFlow.Ops</a></td></tr><tr><td class="src">pad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:pad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">paddingFIFOQueue</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:paddingFIFOQueue">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">parameterizedTruncatedNormal</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:parameterizedTruncatedNormal">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">parseTensor</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:parseTensor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">partitionGraphs</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:partitionGraphs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">PendingNodeName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#t:PendingNodeName">TensorFlow.Output</a></td></tr><tr><td class="src">perProcessGpuMemoryFraction</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:perProcessGpuMemoryFraction">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">placeholder</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:placeholder">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:placeholder">TensorFlow.Ops</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:placeholder">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">placeholderWithDefault</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:placeholderWithDefault">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">placementPeriod</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:placementPeriod">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">placePrunedGraph</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:placePrunedGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">polygamma</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:polygamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">pow</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:pow">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">prefix</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html#v:prefix">TensorFlow.OpGen</a></td></tr><tr><td class="src">priorityQueue</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:priorityQueue">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">prod</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:prod">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">protoShape</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:protoShape">TensorFlow.Types</a></td></tr><tr><td class="src">putVarInt</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-VarInt.html#v:putVarInt">TensorFlow.Internal.VarInt</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-Q.html b/docs/haddock/doc-index-Q.html
new file mode 100644
index 0000000..b605f6b
--- /dev/null
+++ b/docs/haddock/doc-index-Q.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - Q)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - Q</p><table><tr><td class="src">quantizeAndDequantize</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:quantizeAndDequantize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Queue2</td><td class="module"><a href="./tensorflow-queue-0.1.0.0/TensorFlow-Queue.html#t:Queue2">TensorFlow.Queue</a></td></tr><tr><td class="src">queueClose</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:queueClose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">queueSize</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:queueSize">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-R.html b/docs/haddock/doc-index-R.html
new file mode 100644
index 0000000..0d449a0
--- /dev/null
+++ b/docs/haddock/doc-index-R.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - R)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - R</p><table><tr><td class="src">randomCrop</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:randomCrop">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomGamma</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:randomGamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomShuffle</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:randomShuffle">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomShuffleQueue</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:randomShuffleQueue">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomStandardNormal</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:randomStandardNormal">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomUniform</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:randomUniform">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomUniformInt</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:randomUniformInt">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">range</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:range">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:range">TensorFlow.Ops</a></td></tr><tr><td class="src">rank</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:rank">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerNumRecordsProduced</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:readerNumRecordsProduced">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerNumWorkUnitsCompleted</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:readerNumWorkUnitsCompleted">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerRead</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:readerRead">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerReadUpTo</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:readerReadUpTo">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerReset</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:readerReset">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerRestoreState</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:readerRestoreState">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerSerializeState</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:readerSerializeState">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readFile</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:readFile">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readMessageFromFileOrDie</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html#v:readMessageFromFileOrDie">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">readMNISTLabels</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html#v:readMNISTLabels">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">readMNISTSamples</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html#v:readMNISTSamples">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">real</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:real">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reducedShape</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:reducedShape">TensorFlow.Ops</a></td></tr><tr><td class="src">reduceJoin</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:reduceJoin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Ref</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">TensorFlow.Tensor</a></td></tr><tr><td class="src">refEnter</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:refEnter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refExit</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:refExit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refIdentity</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:refIdentity">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">RefKind</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:RefKind">TensorFlow.Tensor</a></td></tr><tr><td class="src">refMerge</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:refMerge">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refNextIteration</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:refNextIteration">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refSelect</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:refSelect">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refSwitch</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:refSwitch">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">relu</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:relu">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:relu">TensorFlow.Ops</a></td></tr><tr><td class="src">relu6</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:relu6">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">relu6Grad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:relu6Grad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reluGrad</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:reluGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:reluGrad">TensorFlow.Ops</a></td></tr><tr><td class="src">render</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:render">TensorFlow.Build</a></td></tr><tr><td class="src">Rendered</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:Rendered">TensorFlow.Output</a></td></tr><tr><td class="src">renderedNodeDefs</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:renderedNodeDefs">TensorFlow.Build</a></td></tr><tr><td class="src">renderNodeName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:renderNodeName">TensorFlow.Build</a></td></tr><tr><td class="src">renderOutput</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:renderOutput">TensorFlow.Build</a></td></tr><tr><td class="src">reshape</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:reshape">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:reshape">TensorFlow.Ops</a></td></tr><tr><td class="src">resizeArea</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:resizeArea">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeBicubic</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:resizeBicubic">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeBilinear</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:resizeBilinear">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeBilinearGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:resizeBilinearGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeNearestNeighbor</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:resizeNearestNeighbor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeNearestNeighborGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:resizeNearestNeighborGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">ResourceHandle</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:ResourceHandle">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">resourceHandleVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:resourceHandleVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">restore</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:restore">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:restore">TensorFlow.Ops</a></td></tr><tr><td class="src">restoreFromName</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:restoreFromName">TensorFlow.Ops</a></td></tr><tr><td class="src">restoreSlice</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:restoreSlice">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reverse</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:reverse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reverseSequence</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:reverseSequence">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">rGBToHSV</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:rGBToHSV">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">rsqrt</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:rsqrt">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">rsqrtGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:rsqrtGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">run</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:run">TensorFlow.Session</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:run">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">runBuildT</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:runBuildT">TensorFlow.Build</a></td></tr><tr><td class="src">RunMetadata</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:RunMetadata">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'FULL_TRACE</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions-39-FULL_TRACE">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'HARDWARE_TRACE</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions-39-HARDWARE_TRACE">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'NO_TRACE</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions-39-NO_TRACE">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'SOFTWARE_TRACE</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions-39-SOFTWARE_TRACE">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'TraceLevel</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions-39-TraceLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">runSession</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:runSession">TensorFlow.Session</a></td></tr><tr><td class="src">runSessionWithOptions</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:runSessionWithOptions">TensorFlow.Session</a></td></tr><tr><td class="src">runWithFeeds</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:runWithFeeds">TensorFlow.Session</a></td></tr><tr><td class="src">runWithFeeds_</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:runWithFeeds_">TensorFlow.Session</a></td></tr><tr><td class="src">run_</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:run_">TensorFlow.Session</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-S.html b/docs/haddock/doc-index-S.html
new file mode 100644
index 0000000..505dbb5
--- /dev/null
+++ b/docs/haddock/doc-index-S.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - S)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - S</p><table><tr><td class="src">s</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:s">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">sampleDistortedBoundingBox</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sampleDistortedBoundingBox">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">save</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:save">TensorFlow.Ops</a></td></tr><tr><td class="src">Scalar</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:Scalar">TensorFlow.Nodes</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#t:Scalar">TensorFlow.Nodes</a></td></tr><tr><td class="src">scalar</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:scalar">TensorFlow.Ops</a></td></tr><tr><td class="src">scalarSummary</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:scalarSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterAdd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:scatterAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterDiv</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:scatterDiv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterMul</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:scatterMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterSub</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:scatterSub">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterUpdate</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:scatterUpdate">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scomplexVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:scomplexVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">segmentMax</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:segmentMax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentMean</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:segmentMean">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentMin</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:segmentMin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentProd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:segmentProd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentSum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:segmentSum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">select</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:select">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">selfAdjointEig</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:selfAdjointEig">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">selfAdjointEigV2</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:selfAdjointEigV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">serializeManySparse</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:serializeManySparse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">serializeSparse</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:serializeSparse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Session</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#t:Session">TensorFlow.Session</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#t:Session">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">sessionConfig</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:sessionConfig">TensorFlow.Session</a></td></tr><tr><td class="src">sessionInterOpThreadPool</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:sessionInterOpThreadPool">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">SessionOption</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#t:SessionOption">TensorFlow.Session</a></td></tr><tr><td class="src">sessionTarget</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html#v:sessionTarget">TensorFlow.Session</a></td></tr><tr><td class="src">setSessionConfig</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:setSessionConfig">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">setSessionTarget</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:setSessionTarget">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">Shape</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:Shape">TensorFlow.Types</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">TensorFlow.Types</a></td></tr><tr><td class="src">shape</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shape">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:shape">TensorFlow.Ops</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:shape">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">shapeN</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shapeN">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">shardedFilename</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shardedFilename">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">shardedFilespec</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shardedFilespec">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sigmoid</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sigmoid">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sigmoidGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sigmoidGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sign</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sign">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:sign">TensorFlow.Ops</a></td></tr><tr><td class="src">sin</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">size</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:size">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:size">TensorFlow.Ops</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:size">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">slice</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:slice">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softmax</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:softmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:softmax">TensorFlow.Ops</a></td></tr><tr><td class="src">softmaxCrossEntropyWithLogits</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:softmaxCrossEntropyWithLogits">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:softmaxCrossEntropyWithLogits">TensorFlow.Ops</a></td></tr><tr><td class="src">softplus</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:softplus">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softplusGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:softplusGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softsign</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:softsign">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softsignGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:softsignGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">spaceToBatch</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:spaceToBatch">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">spaceToBatchND</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:spaceToBatchND">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">spaceToDepth</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:spaceToDepth">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseAdd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseAddGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseAddGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyAdadelta</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseApplyAdadelta">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyAdagrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseApplyAdagrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyAdagradDA</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseApplyAdagradDA">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyFtrl</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseApplyFtrl">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyMomentum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseApplyMomentum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyProximalAdagrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseApplyProximalAdagrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyProximalGradientDescent</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseApplyProximalGradientDescent">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyRMSProp</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseApplyRMSProp">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseConcat</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseConcat">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseDenseCwiseAdd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseDenseCwiseAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseDenseCwiseDiv</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseDenseCwiseDiv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseDenseCwiseMul</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseDenseCwiseMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseMatMul</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseMatMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseReduceSum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseReduceSum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseReduceSumSparse</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseReduceSumSparse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseReorder</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseReorder">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseReshape</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseReshape">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentMean</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSegmentMean">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentMeanGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSegmentMeanGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentSqrtN</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSegmentSqrtN">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentSqrtNGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSegmentSqrtNGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentSum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSegmentSum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSoftmax</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSoftmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSoftmaxCrossEntropyWithLogits</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSoftmaxCrossEntropyWithLogits">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSparseMaximum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSparseMaximum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSparseMinimum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSparseMinimum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSplit</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseSplit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseTensorDenseAdd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseTensorDenseAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseTensorDenseMatMul</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseTensorDenseMatMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseToDense</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sparseToDense">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:sparseToDense">TensorFlow.Ops</a></td></tr><tr><td class="src">split</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:split">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sqrt</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sqrt">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sqrtGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sqrtGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">square</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:square">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">squaredDifference</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:squaredDifference">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">squeeze</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:squeeze">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stackClose</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stackClose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stackPop</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stackPop">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stackPush</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stackPush">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stepStats</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:stepStats">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">stopGradient</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stopGradient">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stridedSlice</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stridedSlice">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stridedSliceAssign</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stridedSliceAssign">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stridedSliceGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stridedSliceGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringJoin</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stringJoin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringSplit</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stringSplit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringToHashBucket</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stringToHashBucket">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringToHashBucketFast</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stringToHashBucketFast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringToHashBucketStrong</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stringToHashBucketStrong">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringToNumber</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:stringToNumber">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringVal</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:stringVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">sub</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sub">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:sub">TensorFlow.Ops</a></td></tr><tr><td class="src">sum</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:sum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:sum">TensorFlow.Ops</a></td></tr><tr><td class="src">summary</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:summary">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">SummaryTensor</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#t:SummaryTensor">TensorFlow.Build</a></td></tr><tr><td class="src">svd</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:svd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">switch</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:switch">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-T.html b/docs/haddock/doc-index-T.html
new file mode 100644
index 0000000..2eb505c
--- /dev/null
+++ b/docs/haddock/doc-index-T.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - T)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - T</p><table><tr><td class="src">tan</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tan">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tanh</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tanh">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tanhGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tanhGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Template</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#t:Template">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">templateDefault</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:templateDefault">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">templateRestrictions</td><td class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html#v:templateRestrictions">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">temporaryVariable</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:temporaryVariable">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Tensor</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:Tensor">TensorFlow.Tensor</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">TensorFlow.Tensor</a></td></tr><tr><td class="src">tensor</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:tensor">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">tensorArrayClose</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayClose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayConcat</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayConcat">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayGather</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayGather">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayPack</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayPack">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayRead</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayRead">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayScatter</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayScatter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArraySize</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArraySize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArraySplit</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArraySplit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayUnpack</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayUnpack">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayWrite</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorArrayWrite">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorAttr</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:tensorAttr">TensorFlow.Tensor</a></td></tr><tr><td class="src">tensorContent</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:tensorContent">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">TensorData</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:TensorData">TensorFlow.Types</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorData">TensorFlow.Types</a></td></tr><tr><td class="alt">3 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:TensorData">TensorFlow.Internal.FFI</a></td></tr><tr><td class="alt">4 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#t:TensorData">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">tensorDataBytes</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:tensorDataBytes">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">tensorDataDimensions</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:tensorDataDimensions">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">tensorDataType</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:tensorDataType">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">TensorFlowException</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:TensorFlowException">TensorFlow.Internal.FFI</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#t:TensorFlowException">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">tensorFromName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:tensorFromName">TensorFlow.Tensor</a></td></tr><tr><td class="src">TensorKind</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:TensorKind">TensorFlow.Tensor</a></td></tr><tr><td class="src">tensorKind</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:tensorKind">TensorFlow.Tensor</a></td></tr><tr><td class="src">tensorOutput</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:tensorOutput">TensorFlow.Tensor</a></td></tr><tr><td class="src">TensorProto</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:TensorProto">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">tensorRefType</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:tensorRefType">TensorFlow.Types</a></td></tr><tr><td class="src">tensorShape</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:tensorShape">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">TensorShapeProto</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:TensorShapeProto">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">TensorShapeProto'Dim</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:TensorShapeProto-39-Dim">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">tensorSummary</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tensorSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">TensorType</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorFlow.Types</a></td></tr><tr><td class="src">tensorType</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:tensorType">TensorFlow.Types</a></td></tr><tr><td class="src">TensorTypes</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorTypes">TensorFlow.Types</a></td></tr><tr><td class="src">tensorVal</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:tensorVal">TensorFlow.Types</a></td></tr><tr><td class="src">testImageData</td><td class="module"><a href="./tensorflow-mnist-input-data-0.1.0.0/TensorFlow-Examples-MNIST-InputData.html#v:testImageData">TensorFlow.Examples.MNIST.InputData</a></td></tr><tr><td class="src">testLabelData</td><td class="module"><a href="./tensorflow-mnist-input-data-0.1.0.0/TensorFlow-Examples-MNIST-InputData.html#v:testLabelData">TensorFlow.Examples.MNIST.InputData</a></td></tr><tr><td class="src">textLineReader</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:textLineReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tFRecordReader</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tFRecordReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">ThreadPoolOptionProto</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Data Constructor)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:ThreadPoolOptionProto">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:ThreadPoolOptionProto">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">threadUnsafeUnigramCandidateSampler</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:threadUnsafeUnigramCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tile</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tile">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tileGrad</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:tileGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">timelineStep</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:timelineStep">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">timeoutInMs</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:timeoutInMs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">topK</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:topK">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:topK">TensorFlow.Ops</a></td></tr><tr><td class="src">topKV2</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:topKV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">traceLevel</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:traceLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">trainingImageData</td><td class="module"><a href="./tensorflow-mnist-input-data-0.1.0.0/TensorFlow-Examples-MNIST-InputData.html#v:trainingImageData">TensorFlow.Examples.MNIST.InputData</a></td></tr><tr><td class="src">trainingLabelData</td><td class="module"><a href="./tensorflow-mnist-input-data-0.1.0.0/TensorFlow-Examples-MNIST-InputData.html#v:trainingLabelData">TensorFlow.Examples.MNIST.InputData</a></td></tr><tr><td class="src">transpose</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:transpose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:transpose">TensorFlow.Ops</a></td></tr><tr><td class="src">truncatedNormal</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:truncatedNormal">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:truncatedNormal">TensorFlow.Ops</a></td></tr><tr><td class="src">type'</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:type-39-">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:type-39-">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">typeAttr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:typeAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">TypeError</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#t:TypeError">TensorFlow.Types</a></td></tr><tr><td class="src">typeListAttr</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:typeListAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-U.html b/docs/haddock/doc-index-U.html
new file mode 100644
index 0000000..e9ab94a
--- /dev/null
+++ b/docs/haddock/doc-index-U.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - U)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - U</p><table><tr><td class="src">unControlNode</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:unControlNode">TensorFlow.Output</a>, <a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:unControlNode">TensorFlow.Build</a></td></tr><tr><td class="src">uniformCandidateSampler</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:uniformCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Unique</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#t:Unique">TensorFlow.Build</a></td></tr><tr><td class="src">unique</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:unique">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">uniqueWithCounts</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:uniqueWithCounts">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">unknownRank</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#v:unknownRank">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">unNodeName</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:unNodeName">TensorFlow.Output</a></td></tr><tr><td class="src">unOpType</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:unOpType">TensorFlow.Output</a></td></tr><tr><td class="src">unOutputIx</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:unOutputIx">TensorFlow.Output</a></td></tr><tr><td class="src">unpack</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:unpack">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">Unrendered</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html#v:Unrendered">TensorFlow.Output</a></td></tr><tr><td class="src">unScalar</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html#v:unScalar">TensorFlow.Nodes</a></td></tr><tr><td class="src">unsortedSegmentSum</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:unsortedSegmentSum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">unTensorData</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html#v:unTensorData">TensorFlow.Types</a></td></tr><tr><td class="src">usePerSessionThreads</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:usePerSessionThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">useProtoAsVoidPtrLen</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:useProtoAsVoidPtrLen">TensorFlow.Internal.FFI</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-V.html b/docs/haddock/doc-index-V.html
new file mode 100644
index 0000000..70ac6d0
--- /dev/null
+++ b/docs/haddock/doc-index-V.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - V)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - V</p><table><tr><td class="src">Value</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">TensorFlow.Tensor</a></td></tr><tr><td class="src">value</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">TensorFlow.Tensor</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:value">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#v:value">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">4 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#v:value">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">ValueKind</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:ValueKind">TensorFlow.Tensor</a></td></tr><tr><td class="src">variable</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:variable">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:variable">TensorFlow.Ops</a></td></tr><tr><td class="src">vector</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:vector">TensorFlow.Ops</a></td></tr><tr><td class="src">version</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:version">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#v:version">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">versionNumber</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#v:versionNumber">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">versions</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#v:versions">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">visibleDeviceList</td><td class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#v:visibleDeviceList">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-W.html b/docs/haddock/doc-index-W.html
new file mode 100644
index 0000000..1b704d9
--- /dev/null
+++ b/docs/haddock/doc-index-W.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - W)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - W</p><table><tr><td class="src">where'</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:where-39-">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">wholeFileReader</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:wholeFileReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">withControlDependencies</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-ControlFlow.html#v:withControlDependencies">TensorFlow.ControlFlow</a></td></tr><tr><td class="src">withDevice</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:withDevice">TensorFlow.Build</a></td></tr><tr><td class="src">withNameScope</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:withNameScope">TensorFlow.Build</a></td></tr><tr><td class="src">withNodeDependencies</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:withNodeDependencies">TensorFlow.Build</a></td></tr><tr><td class="src">withSession</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html#v:withSession">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">withStateLens</td><td class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html#v:withStateLens">TensorFlow.Build</a></td></tr><tr><td class="src">wtsCkpt</td><td class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-TrainedGraph.html#v:wtsCkpt">TensorFlow.Examples.MNIST.TrainedGraph</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index-Z.html b/docs/haddock/doc-index-Z.html
new file mode 100644
index 0000000..473d679
--- /dev/null
+++ b/docs/haddock/doc-index-Z.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index - Z)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - Z</p><table><tr><td class="src">zeroInitializedVariable</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:zeroInitializedVariable">TensorFlow.Ops</a></td></tr><tr><td class="src">zeros</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:zeros">TensorFlow.Ops</a></td></tr><tr><td class="src">zerosLike</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:zerosLike">TensorFlow.GenOps.Core</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html#v:zerosLike">TensorFlow.Ops</a></td></tr><tr><td class="src">zeta</td><td class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:zeta">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/doc-index.html b/docs/haddock/doc-index.html
new file mode 100644
index 0000000..07ad04d
--- /dev/null
+++ b/docs/haddock/doc-index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title> (Index)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-47.html">/</a></li><li><a href="doc-index-92.html">\</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/frames.html b/docs/haddock/frames.html
new file mode 100644
index 0000000..1b4e38d
--- /dev/null
+++ b/docs/haddock/frames.html
@@ -0,0 +1,30 @@
+<!DOCTYPE html 
+     PUBLIC "-//W3C//DTD XHTML 1.0 Frameset//EN"
+     "http://www.w3.org/TR/xhtml1/DTD/xhtml1-frameset.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<title></title>
+<script src="haddock-util.js" type="text/javascript"></script>
+<script type="text/javascript"><!--
+/*
+
+  The synopsis frame needs to be updated using javascript, so we hide
+  it by default and only show it if javascript is enabled.
+
+  TODO: provide some means to disable it.
+*/
+function load() {
+  var d = document.getElementById("inner-fs");
+  d.rows = "50%,50%";
+  postReframe();
+}
+--></script>
+</head>
+<frameset id="outer-fs" cols="25%,75%" onload="load()">
+  <frameset id="inner-fs" rows="100%,0%">
+    <frame src="index-frames.html" name="modules" />
+    <frame src="" name="synopsis" />
+  </frameset>
+  <frame src="index.html" name="main" />
+</frameset>
+</html>
diff --git a/docs/haddock/haddock-util.js b/docs/haddock/haddock-util.js
new file mode 100644
index 0000000..9a6fccf
--- /dev/null
+++ b/docs/haddock/haddock-util.js
@@ -0,0 +1,344 @@
+// Haddock JavaScript utilities
+
+var rspace = /\s\s+/g,
+	  rtrim = /^\s+|\s+$/g;
+
+function spaced(s) { return (" " + s + " ").replace(rspace, " "); }
+function trim(s)   { return s.replace(rtrim, ""); }
+
+function hasClass(elem, value) {
+  var className = spaced(elem.className || "");
+  return className.indexOf( " " + value + " " ) >= 0;
+}
+
+function addClass(elem, value) {
+  var className = spaced(elem.className || "");
+  if ( className.indexOf( " " + value + " " ) < 0 ) {
+    elem.className = trim(className + " " + value);
+  }
+}
+
+function removeClass(elem, value) {
+  var className = spaced(elem.className || "");
+  className = className.replace(" " + value + " ", " ");
+  elem.className = trim(className);
+}
+
+function toggleClass(elem, valueOn, valueOff, bool) {
+  if (bool == null) { bool = ! hasClass(elem, valueOn); }
+  if (bool) {
+    removeClass(elem, valueOff);
+    addClass(elem, valueOn);
+  }
+  else {
+    removeClass(elem, valueOn);
+    addClass(elem, valueOff);
+  }
+  return bool;
+}
+
+
+function makeClassToggle(valueOn, valueOff)
+{
+  return function(elem, bool) {
+    return toggleClass(elem, valueOn, valueOff, bool);
+  }
+}
+
+toggleShow = makeClassToggle("show", "hide");
+toggleCollapser = makeClassToggle("collapser", "expander");
+
+function toggleSection(id)
+{
+  var b = toggleShow(document.getElementById("section." + id));
+  toggleCollapser(document.getElementById("control." + id), b);
+  rememberCollapsed(id, b);
+  return b;
+}
+
+var collapsed = {};
+function rememberCollapsed(id, b)
+{
+  if(b)
+    delete collapsed[id]
+  else
+    collapsed[id] = null;
+
+  var sections = [];
+  for(var i in collapsed)
+  {
+    if(collapsed.hasOwnProperty(i))
+      sections.push(i);
+  }
+  // cookie specific to this page; don't use setCookie which sets path=/
+  document.cookie = "collapsed=" + escape(sections.join('+'));
+}
+
+function restoreCollapsed()
+{
+  var cookie = getCookie("collapsed");
+  if(!cookie)
+    return;
+
+  var ids = cookie.split('+');
+  for(var i in ids)
+  {
+    if(document.getElementById("section." + ids[i]))
+      toggleSection(ids[i]);
+  }
+}
+
+function setCookie(name, value) {
+  document.cookie = name + "=" + escape(value) + ";path=/;";
+}
+
+function clearCookie(name) {
+  document.cookie = name + "=;path=/;expires=Thu, 01-Jan-1970 00:00:01 GMT;";
+}
+
+function getCookie(name) {
+  var nameEQ = name + "=";
+  var ca = document.cookie.split(';');
+  for(var i=0;i < ca.length;i++) {
+    var c = ca[i];
+    while (c.charAt(0)==' ') c = c.substring(1,c.length);
+    if (c.indexOf(nameEQ) == 0) {
+      return unescape(c.substring(nameEQ.length,c.length));
+    }
+  }
+  return null;
+}
+
+
+
+var max_results = 75; // 50 is not enough to search for map in the base libraries
+var shown_range = null;
+var last_search = null;
+
+function quick_search()
+{
+    perform_search(false);
+}
+
+function full_search()
+{
+    perform_search(true);
+}
+
+
+function perform_search(full)
+{
+    var text = document.getElementById("searchbox").value.toLowerCase();
+    if (text == last_search && !full) return;
+    last_search = text;
+    
+    var table = document.getElementById("indexlist");
+    var status = document.getElementById("searchmsg");
+    var children = table.firstChild.childNodes;
+    
+    // first figure out the first node with the prefix
+    var first = bisect(-1);
+    var last = (first == -1 ? -1 : bisect(1));
+
+    if (first == -1)
+    {
+        table.className = "";
+        status.innerHTML = "No results found, displaying all";
+    }
+    else if (first == 0 && last == children.length - 1)
+    {
+        table.className = "";
+        status.innerHTML = "";
+    }
+    else if (last - first >= max_results && !full)
+    {
+        table.className = "";
+        status.innerHTML = "More than " + max_results + ", press Search to display";
+    }
+    else
+    {
+        // decide what you need to clear/show
+        if (shown_range)
+            setclass(shown_range[0], shown_range[1], "indexrow");
+        setclass(first, last, "indexshow");
+        shown_range = [first, last];
+        table.className = "indexsearch";
+        status.innerHTML = "";
+    }
+
+    
+    function setclass(first, last, status)
+    {
+        for (var i = first; i <= last; i++)
+        {
+            children[i].className = status;
+        }
+    }
+    
+    
+    // do a binary search, treating 0 as ...
+    // return either -1 (no 0's found) or location of most far match
+    function bisect(dir)
+    {
+        var first = 0, finish = children.length - 1;
+        var mid, success = false;
+
+        while (finish - first > 3)
+        {
+            mid = Math.floor((finish + first) / 2);
+
+            var i = checkitem(mid);
+            if (i == 0) i = dir;
+            if (i == -1)
+                finish = mid;
+            else
+                first = mid;
+        }
+        var a = (dir == 1 ? first : finish);
+        var b = (dir == 1 ? finish : first);
+        for (var i = b; i != a - dir; i -= dir)
+        {
+            if (checkitem(i) == 0) return i;
+        }
+        return -1;
+    }    
+    
+    
+    // from an index, decide what the result is
+    // 0 = match, -1 is lower, 1 is higher
+    function checkitem(i)
+    {
+        var s = getitem(i).toLowerCase().substr(0, text.length);
+        if (s == text) return 0;
+        else return (s > text ? -1 : 1);
+    }
+    
+    
+    // from an index, get its string
+    // this abstracts over alternates
+    function getitem(i)
+    {
+        for ( ; i >= 0; i--)
+        {
+            var s = children[i].firstChild.firstChild.data;
+            if (s.indexOf(' ') == -1)
+                return s;
+        }
+        return ""; // should never be reached
+    }
+}
+
+function setSynopsis(filename) {
+    if (parent.window.synopsis) {
+        if (parent.window.synopsis.location.replace) {
+            // In Firefox this avoids adding the change to the history.
+            parent.window.synopsis.location.replace(filename);
+        } else {
+            parent.window.synopsis.location = filename;
+        }
+    }
+}
+
+function addMenuItem(html) {
+  var menu = document.getElementById("page-menu");
+  if (menu) {
+    var btn = menu.firstChild.cloneNode(false);
+    btn.innerHTML = html;
+    menu.appendChild(btn);
+  }
+}
+
+function adjustForFrames() {
+  var bodyCls;
+  
+  if (parent.location.href == window.location.href) {
+    // not in frames, so add Frames button
+    addMenuItem("<a href='#' onclick='reframe();return true;'>Frames</a>");
+    bodyCls = "no-frame";
+  }
+  else {
+    bodyCls = "in-frame";
+  }
+  addClass(document.body, bodyCls);
+}
+
+function reframe() {
+  setCookie("haddock-reframe", document.URL);
+  window.location = "frames.html";
+}
+
+function postReframe() {
+  var s = getCookie("haddock-reframe");
+  if (s) {
+    parent.window.main.location = s;
+    clearCookie("haddock-reframe");
+  }
+}
+
+function styles() {
+  var i, a, es = document.getElementsByTagName("link"), rs = [];
+  for (i = 0; a = es[i]; i++) {
+    if(a.rel.indexOf("style") != -1 && a.title) {
+      rs.push(a);
+    }
+  }
+  return rs;
+}
+
+function addStyleMenu() {
+  var as = styles();
+  var i, a, btns = "";
+  for(i=0; a = as[i]; i++) {
+    btns += "<li><a href='#' onclick=\"setActiveStyleSheet('"
+      + a.title + "'); return false;\">"
+      + a.title + "</a></li>"
+  }
+  if (as.length > 1) {
+    var h = "<div id='style-menu-holder'>"
+      + "<a href='#' onclick='styleMenu(); return false;'>Style &#9662;</a>"
+      + "<ul id='style-menu' class='hide'>" + btns + "</ul>"
+      + "</div>";
+    addMenuItem(h);
+  }
+}
+
+function setActiveStyleSheet(title) {
+  var as = styles();
+  var i, a, found;
+  for(i=0; a = as[i]; i++) {
+    a.disabled = true;
+          // need to do this always, some browsers are edge triggered
+    if(a.title == title) {
+      found = a;
+    }
+  }
+  if (found) {
+    found.disabled = false;
+    setCookie("haddock-style", title);
+  }
+  else {
+    as[0].disabled = false;
+    clearCookie("haddock-style");
+  }
+  styleMenu(false);
+}
+
+function resetStyle() {
+  var s = getCookie("haddock-style");
+  if (s) setActiveStyleSheet(s);
+}
+
+
+function styleMenu(show) {
+  var m = document.getElementById('style-menu');
+  if (m) toggleShow(m, show);
+}
+
+
+function pageLoad() {
+  addStyleMenu();
+  adjustForFrames();
+  resetStyle();
+  restoreCollapsed();
+}
+
diff --git a/docs/haddock/hslogo-16.png b/docs/haddock/hslogo-16.png
new file mode 100644
index 0000000000000000000000000000000000000000..0ff8579fbd897417b0d6dad6e920f8882138a7c0
GIT binary patch
literal 1684
zcmV;F25b3=P)<h;3K|Lk000e1NJLTq000*N000mO1^@s6{V(Jg0008+X+uL$Nkc;*
zP;zf(X>4Tx0C)j~RL^S@K@|QrZmG~B2wH0nvUrdpNm;9CMbtL^5n^i$+aI<?kYqDS
z)^0ZI?k2URh~(nYD&9ryp$Gqf9(wR(FG_`4MJV2@C}N=(E%D86Otz)!z|Ma2-Z$@k
zZ+71R4RX>n^?(HA4aZWV5ov6ELTdbo0FI&wK{O>*+w4vx20?>!`FrQsdJlnHR>OPy
zcd~b_n$otK2Za4V;76L-DzNVtaSB-y0*E}{p()372;bw_^6ZZ}PI-92wGS&j#91PI
zKs7DSe@(bk%_Y-7gGe}(^>I=@oY#w#*Bu9GZf3^F5WP>3rn}7Ut74&?PWBFvy`A)a
zPP5)V!Xd&78LdA?xQ(9mjMYElVd13a#D+Z_7&Y|xU=_C-srWU*6kiZcC!$nw*)9$7
zn6CX+@=AhmkT}X@VSsa5NKe;HZuq)~1$`#h6R+ZTR#D-3j}vF!)ZOnz+5)dI4jl{{
z44Mr{P!L4~VVJN`K!!XTF*LGrKO?IK8<Tr7btG!LbYeuYL3=jbJ-1P$-8}v%B5{;M
zwFr{@LH;VQ$xr2Z`O93e*jD$Ht(%&<^58qg<(at}9@o>z<8w`3e3jI8lUGNUta*C8
zn(P`s>{pjD=7Kek#B;Fw@hxAK%$F&Q6vg9J^Xf~4by_hu-=A!MJ3Znq&n~srbFGPs
zH&&aMXZ>nO`|hf|ljc?VPhR!${AbO?W8x_>CU%PFA&Hm8F7cAsOREdwU~R_;ot1_u
z(ruCYB-LPGn!NQdT|ZlRy+(fw^-+`=%+gee_kY4FWHg<*4sZI8+sFJD<oAl_pC|$^
zY~aY5x@}W&?+~G7rEYVs0vEs0eekw!YomTR`~+A$s}`+NHJ>270UUORdLHO0nA4V)
z%{fwsET5CQ>B?eK%uw4yQc~9?*JVo<vzb+5>2}ze(;aRcp*ceL#HUJSllrgm5wQKR
zQu+C;QrUh^8rFfA`ftFz{YAidi-`aL010qNS#tmY4c7nw4c7reD4Tcy00T@(L_t(I
z5sj2vNEA^R$7gqDc6T=2^@fUA2(c`MltuL5<|<Z1YDGjBEr^N`3VJpQ$`C4&9(*Z6
z=|*B$y;-1^F6vXRAd7+(;)77Qps2euv)AeWmj>KW>RWz$&YbU@|M|{$E*8Tu-Ux!w
z1Y*Dr&Ubfr&v-nZaaB{3ilRumrjPmk{sZvQEWlW+{o~IH|8)=s6c#X9S5s5d%<Agu
zgM$153Ttk++ZPA~ZqgqN1|^pNV-|?9-zo48n9+Yiy$*su5czgp*VFj6!1h&DRZR>J
z4@)&QH5|xQY-)^L1n<EM!01(Q6kM8}ojndgRD~2KVIa@>0pTRu0Lx9`08YTjTwn^6
z0;b1+aQ@)n;Em$q;=7BBi)v0zj&o^g>0Whp^_^5IbxIUP8C@y9;R?*Ouu}rmfxbU=
zwtWVNke-m!=`7bYEhWpcI5#)9qp`8E0lr6IQ)ARL3Ui}Af@grj8aN1=r>Cb+prlzO
zNfJs*N_tUm2ZL<f%*@E?H`Ak_J`Q#Qri<;K{{H@Bg@uKWS5{WE+}vCx91geD*4DNm
zJ?D^~dGz7lWtbWH`o<0O#|{R6h9M*GMpjnV?Y_RgatOb~1le#ikPKTNIn%<;=>%5*
zPmL2??da$TR904gL(VDAQ<BE%k@5ZA>-Fv_Dk}Pdw*4T(%*f4MKLRg=4ekMjhe2mW
zMFsBwg%ftWT}0kxRaIk1k7qJ8*#cKB;Ft{i`zVIs-Nqge;!!Ld7#O&Qqu7e0sJmP)
z$MW*>L$vSB&dxp@iA3U9fo)-7!Czlr{|o7<LYh`Fd@X2-b0fIa*Wdy$({Zt@t7`&L
zhoaGF!l5{1P1Qm;<GaSj#=Oa7atf1JZ8n=F9*@uC?b(X48O9kPoH~h;!o2=WM?ZLl
z$Lj;4K~)fAdW@^N%nzT>Hv{1oyg3xsu%gn@(b1>$;SM-ZaQ`HV=V0s;lr%d8bd;xY
zGw<i+<$17ay36gdrqk(Lxc1|{y}dSeM5OIV(FjhBg5#OM<!Z*8(@<7crlVw+B-8R1
zl{3J6w9fYQ^vojauaRgZ=~SE&CK!R3pCe!v3>Nvm3=Iu=tyXIgtJnf@A(2S@M140N
ew{UA~tMxaJq;$xaSSi*30000<MNUMnLSTa0p)&^n

literal 0
HcmV?d00001

diff --git a/docs/haddock/index-frames.html b/docs/haddock/index-frames.html
new file mode 100644
index 0000000..c0c27c9
--- /dev/null
+++ b/docs/haddock/index-frames.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title></title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-list"><p class="caption">Modules</p><ul><li class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html" target="main">Proto.Tensorflow.Core.Framework.AttrValue</a></li><li class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html" target="main">Proto.Tensorflow.Core.Framework.Graph</a></li><li class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html" target="main">Proto.Tensorflow.Core.Framework.NodeDef</a></li><li class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html" target="main">Proto.Tensorflow.Core.Framework.OpDef</a></li><li class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html" target="main">Proto.Tensorflow.Core.Framework.ResourceHandle</a></li><li class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html" target="main">Proto.Tensorflow.Core.Framework.Tensor</a></li><li class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html" target="main">Proto.Tensorflow.Core.Framework.TensorShape</a></li><li class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html" target="main">Proto.Tensorflow.Core.Framework.Types</a></li><li class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html" target="main">Proto.Tensorflow.Core.Protobuf.Config</a></li><li class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html" target="main">TensorFlow.Build</a></li><li class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-BuildOp.html" target="main">TensorFlow.BuildOp</a></li><li class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-ControlFlow.html" target="main">TensorFlow.ControlFlow</a></li><li class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-EmbeddingOps.html" target="main">TensorFlow.EmbeddingOps</a></li><li class="module"><a href="./tensorflow-mnist-input-data-0.1.0.0/TensorFlow-Examples-MNIST-InputData.html" target="main">TensorFlow.Examples.MNIST.InputData</a></li><li class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html" target="main">TensorFlow.Examples.MNIST.Parse</a></li><li class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-TrainedGraph.html" target="main">TensorFlow.Examples.MNIST.TrainedGraph</a></li><li class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html" target="main">TensorFlow.GenOps.Core</a></li><li class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Gradient.html" target="main">TensorFlow.Gradient</a></li><li class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html" target="main">TensorFlow.Internal.FFI</a></li><li class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-VarInt.html" target="main">TensorFlow.Internal.VarInt</a></li><li class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html" target="main">TensorFlow.Nodes</a></li><li class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html" target="main">TensorFlow.OpGen</a></li><li class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html" target="main">TensorFlow.OpGen.AttrVal</a></li><li class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html" target="main">TensorFlow.Ops</a></li><li class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html" target="main">TensorFlow.Output</a></li><li class="module"><a href="./tensorflow-queue-0.1.0.0/TensorFlow-Queue.html" target="main">TensorFlow.Queue</a></li><li class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html" target="main">TensorFlow.Session</a></li><li class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html" target="main">TensorFlow.Tensor</a></li><li class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html" target="main">TensorFlow.Types</a></li></ul></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/index.html b/docs/haddock/index.html
new file mode 100644
index 0000000..08c86c5
--- /dev/null
+++ b/docs/haddock/index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title></title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption empty">&nbsp;</p></div><div id="content"><div id="module-list"><p class="caption">Modules</p><ul><li><span id="control.n.1" class="module collapser" onclick="toggleSection('n.1')">Proto</span><ul id="section.n.1" class="show"><li><span id="control.n.1.1" class="module collapser" onclick="toggleSection('n.1.1')">Tensorflow</span><ul id="section.n.1.1" class="show"><li><span id="control.n.1.1.1" class="module collapser" onclick="toggleSection('n.1.1.1')">Core</span><ul id="section.n.1.1.1" class="show"><li><span id="control.n.1.1.1.1" class="module collapser" onclick="toggleSection('n.1.1.1.1')">Framework</span><ul id="section.n.1.1.1.1" class="show"><li><span class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html">Proto.Tensorflow.Core.Framework.AttrValue</a></span></li><li><span class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html">Proto.Tensorflow.Core.Framework.Graph</a></span></li><li><span class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html">Proto.Tensorflow.Core.Framework.NodeDef</a></span></li><li><span class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html">Proto.Tensorflow.Core.Framework.OpDef</a></span></li><li><span class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html">Proto.Tensorflow.Core.Framework.ResourceHandle</a></span></li><li><span class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html">Proto.Tensorflow.Core.Framework.Tensor</a></span></li><li><span class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html">Proto.Tensorflow.Core.Framework.TensorShape</a></span></li><li><span class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html">Proto.Tensorflow.Core.Framework.Types</a></span></li></ul></li><li><span id="control.n.1.1.1.2" class="module collapser" onclick="toggleSection('n.1.1.1.2')">Protobuf</span><ul id="section.n.1.1.1.2" class="show"><li><span class="module"><a href="./tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html">Proto.Tensorflow.Core.Protobuf.Config</a></span></li></ul></li></ul></li></ul></li></ul></li><li><span id="control.n.2" class="module collapser" onclick="toggleSection('n.2')">TensorFlow</span><ul id="section.n.2" class="show"><li><span class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Build.html">TensorFlow.Build</a></span></li><li><span class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-BuildOp.html">TensorFlow.BuildOp</a></span></li><li><span class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-ControlFlow.html">TensorFlow.ControlFlow</a></span></li><li><span class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-EmbeddingOps.html">TensorFlow.EmbeddingOps</a></span></li><li><span id="control.n.2.5" class="module collapser" onclick="toggleSection('n.2.5')">Examples</span><ul id="section.n.2.5" class="show"><li><span id="control.n.2.5.1" class="module collapser" onclick="toggleSection('n.2.5.1')">MNIST</span><ul id="section.n.2.5.1" class="show"><li><span class="module"><a href="./tensorflow-mnist-input-data-0.1.0.0/TensorFlow-Examples-MNIST-InputData.html">TensorFlow.Examples.MNIST.InputData</a></span></li><li><span class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html">TensorFlow.Examples.MNIST.Parse</a></span></li><li><span class="module"><a href="./tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-TrainedGraph.html">TensorFlow.Examples.MNIST.TrainedGraph</a></span></li></ul></li></ul></li><li><span id="control.n.2.6" class="module collapser" onclick="toggleSection('n.2.6')">GenOps</span><ul id="section.n.2.6" class="show"><li><span class="module"><a href="./tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html">TensorFlow.GenOps.Core</a></span></li></ul></li><li><span class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Gradient.html">TensorFlow.Gradient</a></span></li><li><span id="control.n.2.8" class="module collapser" onclick="toggleSection('n.2.8')">Internal</span><ul id="section.n.2.8" class="show"><li><span class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html">TensorFlow.Internal.FFI</a></span></li><li><span class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Internal-VarInt.html">TensorFlow.Internal.VarInt</a></span> Encoders and decoders for varint types.</li></ul></li><li><span class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Nodes.html">TensorFlow.Nodes</a></span></li><li><span class="module"><span id="control.n.2.10" class="collapser" onclick="toggleSection('n.2.10')">&nbsp;</span><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html">TensorFlow.OpGen</a></span><ul id="section.n.2.10" class="show"><li><span class="module"><a href="./tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html">TensorFlow.OpGen.AttrVal</a></span></li></ul></li><li><span class="module"><a href="./tensorflow-ops-0.1.0.0/TensorFlow-Ops.html">TensorFlow.Ops</a></span></li><li><span class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Output.html">TensorFlow.Output</a></span></li><li><span class="module"><a href="./tensorflow-queue-0.1.0.0/TensorFlow-Queue.html">TensorFlow.Queue</a></span></li><li><span class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Session.html">TensorFlow.Session</a></span></li><li><span class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Tensor.html">TensorFlow.Tensor</a></span></li><li><span class="module"><a href="./tensorflow-0.1.0.0/TensorFlow-Types.html">TensorFlow.Types</a></span></li></ul></li></ul></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/minus.gif b/docs/haddock/minus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..1deac2fe1a42e35b994f1b855488f392c50f6a89
GIT binary patch
literal 56
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS53HJ0a;6C;`j@u!^I(6E|
Jh4LZ{)&P#)4$%Mr

literal 0
HcmV?d00001

diff --git a/docs/haddock/ocean.css b/docs/haddock/ocean.css
new file mode 100644
index 0000000..1110b40
--- /dev/null
+++ b/docs/haddock/ocean.css
@@ -0,0 +1,600 @@
+/* @group Fundamentals */
+
+* { margin: 0; padding: 0 }
+
+/* Is this portable? */
+html {
+  background-color: white;
+  width: 100%;
+  height: 100%;
+}
+
+body {
+  background: white;
+  color: black;
+  text-align: left;
+  min-height: 100%;
+  position: relative;
+}
+
+p {
+  margin: 0.8em 0;
+}
+
+ul, ol {
+  margin: 0.8em 0 0.8em 2em;
+}
+
+dl {
+  margin: 0.8em 0;
+}
+
+dt {
+  font-weight: bold;
+}
+dd {
+  margin-left: 2em;
+}
+
+a { text-decoration: none; }
+a[href]:link { color: rgb(196,69,29); }
+a[href]:visited { color: rgb(171,105,84); }
+a[href]:hover { text-decoration:underline; }
+
+/* @end */
+
+/* @group Fonts & Sizes */
+
+/* Basic technique & IE workarounds from YUI 3
+   For reasons, see:
+      http://yui.yahooapis.com/3.1.1/build/cssfonts/fonts.css
+ */
+
+body {
+	font:13px/1.4 sans-serif;
+	*font-size:small; /* for IE */
+	*font:x-small; /* for IE in quirks mode */
+}
+
+h1 { font-size: 146.5%; /* 19pt */ }
+h2 { font-size: 131%;   /* 17pt */ }
+h3 { font-size: 116%;   /* 15pt */ }
+h4 { font-size: 100%;   /* 13pt */ }
+h5 { font-size: 100%;   /* 13pt */ }
+
+select, input, button, textarea {
+	font:99% sans-serif;
+}
+
+table {
+	font-size:inherit;
+	font:100%;
+}
+
+pre, code, kbd, samp, tt, .src {
+	font-family:monospace;
+	*font-size:108%;
+	line-height: 124%;
+}
+
+.links, .link {
+  font-size: 85%; /* 11pt */
+}
+
+#module-header .caption {
+  font-size: 182%; /* 24pt */
+}
+
+.info  {
+  font-size: 85%; /* 11pt */
+}
+
+#table-of-contents, #synopsis  {
+  /* font-size: 85%; /* 11pt */
+}
+
+
+/* @end */
+
+/* @group Common */
+
+.caption, h1, h2, h3, h4, h5, h6 {
+  font-weight: bold;
+  color: rgb(78,98,114);
+  margin: 0.8em 0 0.4em;
+}
+
+* + h1, * + h2, * + h3, * + h4, * + h5, * + h6 {
+  margin-top: 2em;
+}
+
+h1 + h2, h2 + h3, h3 + h4, h4 + h5, h5 + h6 {
+  margin-top: inherit;
+}
+
+ul.links {
+  list-style: none;
+  text-align: left;
+  float: right;
+  display: inline-table;
+  margin: 0 0 0 1em;
+}
+
+ul.links li {
+  display: inline;
+  border-left: 1px solid #d5d5d5;
+  white-space: nowrap;
+  padding: 0;
+}
+
+ul.links li a {
+  padding: 0.2em 0.5em;
+}
+
+.hide { display: none; }
+.show { display: inherit; }
+.clear { clear: both; }
+
+.collapser {
+  background-image: url(minus.gif);
+  background-repeat: no-repeat;
+}
+.expander {
+  background-image: url(plus.gif);
+  background-repeat: no-repeat;
+}
+p.caption.collapser,
+p.caption.expander {
+  background-position: 0 0.4em;
+}
+.collapser, .expander {
+  padding-left: 14px;
+  margin-left: -14px;
+  cursor: pointer;
+}
+
+pre {
+  padding: 0.25em;
+  margin: 0.8em 0;
+  background: rgb(229,237,244);
+  overflow: auto;
+  border-bottom: 0.25em solid white;
+  /* white border adds some space below the box to compensate
+     for visual extra space that paragraphs have between baseline
+     and the bounding box */
+}
+
+.src {
+  background: #f0f0f0;
+  padding: 0.2em 0.5em;
+}
+
+.keyword { font-weight: normal; }
+.def { font-weight: bold; }
+
+
+/* @end */
+
+/* @group Page Structure */
+
+#content {
+  margin: 0 auto;
+  padding: 0 2em 6em;
+}
+
+#package-header {
+  background: rgb(41,56,69);
+  border-top: 5px solid rgb(78,98,114);
+  color: #ddd;
+  padding: 0.2em;
+  position: relative;
+  text-align: left;
+}
+
+#package-header .caption {
+  background: url(hslogo-16.png) no-repeat 0em;
+  color: white;
+  margin: 0 2em;
+  font-weight: normal;
+  font-style: normal;
+  padding-left: 2em;
+}
+
+#package-header a:link, #package-header a:visited { color: white; }
+#package-header a:hover { background: rgb(78,98,114); }
+
+#module-header .caption {
+  color: rgb(78,98,114);
+  font-weight: bold;
+  border-bottom: 1px solid #ddd;
+}
+
+table.info {
+  float: right;
+  padding: 0.5em 1em;
+  border: 1px solid #ddd;
+  color: rgb(78,98,114);
+  background-color: #fff;
+  max-width: 40%;
+  border-spacing: 0;
+  position: relative;
+  top: -0.5em;
+  margin: 0 0 0 2em;
+}
+
+.info th {
+	padding: 0 1em 0 0;
+}
+
+div#style-menu-holder {
+  position: relative;
+  z-index: 2;
+  display: inline;
+}
+
+#style-menu {
+  position: absolute;
+  z-index: 1;
+  overflow: visible;
+  background: #374c5e;
+  margin: 0;
+  text-align: center;
+  right: 0;
+  padding: 0;
+  top: 1.25em;
+}
+
+#style-menu li {
+	display: list-item;
+	border-style: none;
+	margin: 0;
+	padding: 0;
+	color: #000;
+	list-style-type: none;
+}
+
+#style-menu li + li {
+	border-top: 1px solid #919191;
+}
+
+#style-menu a {
+  width: 6em;
+  padding: 3px;
+  display: block;
+}
+
+#footer {
+  background: #ddd;
+  border-top: 1px solid #aaa;
+  padding: 0.5em 0;
+  color: #666;
+  text-align: center;
+  position: absolute;
+  bottom: 0;
+  width: 100%;
+  height: 3em;
+}
+
+/* @end */
+
+/* @group Front Matter */
+
+#table-of-contents {
+  float: right;
+  clear: right;
+  background: #faf9dc;
+  border: 1px solid #d8d7ad;
+  padding: 0.5em 1em;
+  max-width: 20em;
+  margin: 0.5em 0 1em 1em;
+}
+
+#table-of-contents .caption {
+  text-align: center;
+  margin: 0;
+}
+
+#table-of-contents ul {
+  list-style: none;
+  margin: 0;
+}
+
+#table-of-contents ul ul {
+  margin-left: 2em;
+}
+
+#description .caption {
+  display: none;
+}
+
+#synopsis {
+  display: none;
+}
+
+.no-frame #synopsis {
+  display: block;
+  position: fixed;
+  right: 0;
+  height: 80%;
+  top: 10%;
+  padding: 0;
+  max-width: 75%;
+}
+
+#synopsis .caption {
+  float: left;
+  width: 29px;
+  color: rgba(255,255,255,0);
+  height: 110px;
+  margin: 0;
+  font-size: 1px;
+  padding: 0;
+}
+
+#synopsis p.caption.collapser {
+  background: url(synopsis.png) no-repeat -64px -8px;
+}
+
+#synopsis p.caption.expander {
+  background: url(synopsis.png) no-repeat 0px -8px;
+}
+
+#synopsis ul {
+  height: 100%;
+  overflow: auto;
+  padding: 0.5em;
+  margin: 0;
+}
+
+#synopsis ul ul {
+  overflow: hidden;
+}
+
+#synopsis ul,
+#synopsis ul li.src {
+  background-color: #faf9dc;
+  white-space: nowrap;
+  list-style: none;
+  margin-left: 0;
+}
+
+/* @end */
+
+/* @group Main Content */
+
+#interface div.top { margin: 2em 0; }
+#interface h1 + div.top,
+#interface h2 + div.top,
+#interface h3 + div.top,
+#interface h4 + div.top,
+#interface h5 + div.top {
+ 	margin-top: 1em;
+}
+#interface p.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface td.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface span.fixity {
+  color: #919191;
+  border-left: 1px solid #919191;
+  padding: 0.2em 0.5em 0.2em 0.5em;
+  margin: 0 -1em 0 1em;
+}
+
+#interface span.rightedge {
+  border-left: 1px solid #919191;
+  padding: 0.2em 0 0.2em 0;
+  margin: 0 0 0 1em;
+}
+
+#interface table { border-spacing: 2px; }
+#interface td {
+  vertical-align: top;
+  padding-left: 0.5em;
+}
+#interface td.src {
+  white-space: nowrap;
+}
+#interface td.doc p {
+  margin: 0;
+}
+#interface td.doc p + p {
+  margin-top: 0.8em;
+}
+
+.clearfix:after {
+  clear: both;
+  content: " ";
+  display: block;
+  height: 0;
+  visibility: hidden;
+}
+
+.subs dl {
+  margin: 0;
+}
+
+.subs dt {
+  float: left;
+  clear: left;
+  display: block;
+  margin: 1px 0;
+}
+
+.subs dd {
+  float: right;
+  width: 90%;
+  display: block;
+  padding-left: 0.5em;
+  margin-bottom: 0.5em;
+}
+
+.subs dd.empty {
+  display: none;
+}
+
+.subs dd p {
+  margin: 0;
+}
+
+/* Render short-style data instances */
+.inst ul {
+  height: 100%;
+  padding: 0.5em;
+  margin: 0;
+}
+
+.inst, .inst li {
+  list-style: none;
+  margin-left: 1em;
+}
+
+/* Workaround for bug in Firefox (issue #384) */
+.inst-left {
+  float: left;
+}
+
+.top p.src {
+  border-top: 1px solid #ccc;
+}
+
+.subs, .doc {
+  /* use this selector for one level of indent */
+  padding-left: 2em;
+}
+
+.warning {
+  color: red;
+}
+
+.arguments {
+  margin-top: -0.4em;
+}
+.arguments .caption {
+  display: none;
+}
+
+.fields { padding-left: 1em; }
+
+.fields .caption { display: none; }
+
+.fields p { margin: 0 0; }
+
+/* this seems bulky to me
+.methods, .constructors {
+  background: #f8f8f8;
+  border: 1px solid #eee;
+}
+*/
+
+/* @end */
+
+/* @group Auxillary Pages */
+
+
+.extension-list {
+    list-style-type: none;
+    margin-left: 0;
+}
+
+#mini {
+  margin: 0 auto;
+  padding: 0 1em 1em;
+}
+
+#mini > * {
+  font-size: 93%; /* 12pt */
+}
+
+#mini #module-list .caption,
+#mini #module-header .caption {
+  font-size: 125%; /* 15pt */
+}
+
+#mini #interface h1,
+#mini #interface h2,
+#mini #interface h3,
+#mini #interface h4 {
+  font-size: 109%; /* 13pt */
+  margin: 1em 0 0;
+}
+
+#mini #interface .top,
+#mini #interface .src {
+  margin: 0;
+}
+
+#mini #module-list ul {
+  list-style: none;
+  margin: 0;
+}
+
+#alphabet ul {
+	list-style: none;
+	padding: 0;
+	margin: 0.5em 0 0;
+	text-align: center;
+}
+
+#alphabet li {
+	display: inline;
+	margin: 0 0.25em;
+}
+
+#alphabet a {
+	font-weight: bold;
+}
+
+#index .caption,
+#module-list .caption { font-size: 131%; /* 17pt */ }
+
+#index table {
+  margin-left: 2em;
+}
+
+#index .src {
+  font-weight: bold;
+}
+#index .alt {
+  font-size: 77%; /* 10pt */
+  font-style: italic;
+  padding-left: 2em;
+}
+
+#index td + td {
+  padding-left: 1em;
+}
+
+#module-list ul {
+  list-style: none;
+  margin: 0 0 0 2em;
+}
+
+#module-list li {
+  clear: right;
+}
+
+#module-list span.collapser,
+#module-list span.expander {
+  background-position: 0 0.3em;
+}
+
+#module-list .package {
+  float: right;
+}
+
+/* @end */
diff --git a/docs/haddock/plus.gif b/docs/haddock/plus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..2d15c14173d23f664b955cd24f51c82f5f09d91d
GIT binary patch
literal 59
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS5iS+a@;NG@rPSdH*>gbBX
M^XE!9f*2UA0nx1yDgXcg

literal 0
HcmV?d00001

diff --git a/docs/haddock/synopsis.png b/docs/haddock/synopsis.png
new file mode 100644
index 0000000000000000000000000000000000000000..85fb86ec84907bcc86531dc82871948ff4d471fa
GIT binary patch
literal 11327
zcmV-FEWp!=P)<h;3K|Lk000e1NJLTq004jh003wR1^@s6IVt!g000+IX+uL$Nkc;*
zP;zf(X>4Tx0C)k_S!GyNTeqHT_l8Y(cXyX`gGi?cY`Qxn1VID|MJXwjPC)?)F$h6K
zMMOd+6hs7sqbPzXbr*U(-*=zy-hcPcUC*=TdiNM(jyd-lv&OpsU|J&v2m2!^0SE{T
z54F(O;<U4O;KWV=44?vZKnREv-NJ%3&CHI1|Ml9M1kj;!OzhCt|7!L>E2!K(!rTCW
z%wV;vdzf1QjBf#e&~gh74F>?Z4a=WLg$KhJ^$5nap>PLbJadS>e&h8+?D`9%QNL`g
zEVKbYGXj7k5Q(8)0Fd#*a?VIMFW3*64geVHKzE-&0BG!BtmfuTbO(T`0Jaeg2nagF
z{V*1E{Wm{e|AvV~*MEExiC+KU-~R=!2{)|c6Bg`GjQ;iG|FQ`1kAUCTuZtQk34#8{
z4r4<P-^=bq9b@>(3g7#|{=Z@d+d#}7f!3C=>=26vx*jwA8>@MS>RG@Tt_zt3hie^T
z_?<zX!Qp0Bf6L=zytR+Pm;lRjJ;HSV_$l=z8Xd-r9G16)Mp#(Gm>0%9VUd=)Fos7I
z^ghPh%Jy%YZ|)vCf6EaFPai$Q-!=$ppK!y&wrJs)bNdAuANB!m3n34Tfj{s75g-&U
z1A!Pg3bcXF-=!Gv1VmU93G2duANT;{0JugFTqg*|oPXPC|A$2HS3NJd-hcPV3EW`Y
zh=1Dr-5Mv{<{zIvz#Yba<V3iQL)~HT!km4Q{?+Cmi$j0EdL!8&xgxpyr%xz;lr%~X
zrG-*OsiBkrCyE}$48%~fC`FVeN*$#F>y&^Vcn^E_`qRfl{{bzYkp)4~$~NAx_VB;E
z{?P)PU)DbV{Qi#~0H0@T9czDj06@6MNq8OrpdAz(9qQxd9nPr<&s+~tPQySqaZyfb
zNh!%g_5YjeaLxMN*$s<dYY{|0@-La=c>v_p;d%b#U$Wpz0Geb0U>E+EOsEQ;I<Re%
zqFY30<Zpkd!v`9W019}7G68nL0|epsEDhv=3eW($zzCQED_{?t;PK-PkHJt71>!&=
zNC6q(BFF<p;2Nj^wV)BSfDUjM^n!=r33v`Bz%-Zxi(nP3gDtQFK@b)qgJ>Wohy&t-
zL?CHM5mJM6p`(xmWDmJOUQi$u0mVUQpbRJ*DuT+OI;a`C4fR4p&?xj8nuk`Puh35f
z5<x)FBG?f82nmEDLK9((utJ<bcp*X%F^FWu1w;X&9C01dj_5@^K};a#5UYqyBtViP
z>5*JWF{C0=8)=GkKzbrWk@3iMWInPS*@Wyu4kE{pbI3L14-^JPgW^Pq!Q<2bWsPz}
zg`nb5nW!REEvg;Wj~YYGqt;RTXfiY_S_G|(HbmQ@z0gtU6m&ki8r_B-Ku@3-(OVb{
zh8`n;QNS2r>@mKWSWG773g!l;2Q!LUz-(f%SSG9pRuyZCC1S&|DcC~nb!<2G1$Gg;
zjU&Zz;G}VSI0sxHE(w>9tH<5Py}&KucJP#VKD;vC6z`6Y#%JLx@m=^4{33pbgo;Ff
zM3uyf#Fr$Iq=2M}WPoIbWP_BHl$%tE)ST3Z^fYM!=}po{r1PXd2-E~&f;PdC5J9*=
zs3G(aUK2LR$jJD~G{_vt!pSa>)sa0QdqcKOPD3tEZb<G<o<Lqq-cJ64{1XM9f{#Lz
z!ige=BA=p_VuWIi5>LrbsZB|wjHfK7yiNI%a+8XNN{Y&qDu61Js-9|yYMB~K%}=dM
z?M|IcT|xbTdVvN>!$YG@<3@9arjllWW|0;{D?n>V>r0zK+erJ2cAbuzPL|Gw?j&6?
z-95TFdL%tRy&=6neHMKS{UrTQ1~vvw1`mcbh9-s=4Br`97&RC@7}FVVFitT3Wa4Df
zW%6UX#MHqw%Zy?cW;SPzV!p~ez`Vvn%c8>K#*)s`!ZO8*U=?PyV2x$1V13HE$;Qs6
z&lb#9$o7D3jh&udgWZ=sm;FBb3I`2`8ix-@E=M=VM@~9UO-_H#0?vNUbuLye1Fi_J
zGOlM_JKO@?*4#<lH@IhcNO%-^Jb7|?`gqoPIe1NY<9Hi*r}*%EihSODm-z<yzVi$4
z+w!OKcknL@FbWt7#0WG9%m|VRstbk)mJ5yxVT2Tg{DiIuJr_m@%L#i67YmPwAVlOv
zd_=B@jEbT~l|%zY%SB&`5yUjbBE=fS=Edp7kBTRWcZh$I;FWNYxFj(su`ekn=`UF+
zIW0vkWh9j()g|>+T3Fgmx>$N#hD=6JCPAiC=8LR|tcUDX*;jHjawc-Aa(!}p@(S{y
z@=fw93cLy~3MC3J6=@aC6f+ecDWR3LloFKgD*aHFR}NQhQU0tVrsAhkud;kZ;E2bO
z$|DP^+^R&?GSxXXPBj;`QnfjCE_I@Mx%xW|9u0SmYK<jLAx&S+>zbdmB(*}d+O)oF
zD{G(9?$JT&=<B5GJkzDtwbm`teWS;%=cU)6x2`XzAFJPEfHp8TxMc9skj>D|u+DJZ
zNWtioQNJ<4*wVPj_}x+AqoGH;Ob{kUCOIZE$M}u~9_ug#riP|Drn6=OW+7&G%rWL>
z=Ede8ETk;rECwxUES)XuEw`++tg@`8tp%+ktov*zY#eRsY`)v-*k;?#*-6-)vU_6B
zZ0}>=>40^xaj16KJg$2@@A#sloMVdP<OzWjCr=DHF+2G?-E*dJCOWsepj~WT>Ron;
zro?jMrmLZAiR-$Xw%cX5Rd)^dT=x|ZRgY|sB~Mk)Y|mvcRj(Yc6>oL#eD5_MZJ#2a
zFTMu8*L=VGnflfE9r)Y&-w413xCGn|qz?28><?lOIvF$?EE1d&JQt!Ik{7ZbY8+Y}
z28Epn>kOxb4~I`91S8Hy%txw47DsMJ*+jLTq&gXR@@ceibXxRMj9yGtEGpJ5wl9t=
zE-`NYl;)|jcqraAzAu3%Avt03wEpSZM3O|m#Ni~#r0k?`XKc@OC9@@;PF^^xf3_io
zJS8;cWvWW*wR5O*KIfjL$)pvg?Wen^KhBWM$j{i#bjy5vUg~_o`GX6d7oKIwXI;IB
zxfpnH@{;<c+H9)qnCyidlbrTk&fN6eZ+XsnL;14#*DjM>j<`HmaI~Pakhkz+;ck(4
z(L}LU@r@GJlC+ZVSKP0>xT<lr=^ER$jB7tjeM(=I>6f*a^OxsWU@9UjK2+LN4pu2v
z)m1ZBXH@Ui1lG*eTGaN}Db&@~v({%dAQ~bXR<1i<f8MCm*x4l3RCa^@M#fEWGvemQ
zW|!v47UPz_R>jt)TYR@l+GyI++oAU8_Vo_$j=4_z&e7XOxBI$Oy4voD->JFFb+`B)
z-My^)B=?i=A9TlbZ}tTDto3^JF7!F~O+T=EFy3$8|7^f`;L$_9hYtod2fH7sKDs-k
zJaqf9;^U4d@=w~I$~|oxmK$z+CjYE`L}8@!xzh8l(IcbxU#P$69n%?mIBq!pWa8Mw
z=%n@JtCx;1=U%zLT7K>S`pZ=0)Xwzj8T3s0Eahze8`d}FZ-w68n3JEoH?K4Q^qu9q
z=>@li)%RiVcNddCkbTHs;#jI%mR`QQqPOz=<MEG+tHG-~Ybk$G{Z;Zw@YAi&8lRu8
z+pRBt3H@^LHFJY$qvo6Jx4uoY%{N;CTRY#=x0$wUe<=PK{Av4hX(wtIyIZg)wAZz7
zygzdgaBvVz3?=?L3n1YV;pqv$*BUs7u!nPp=KzrO{LU$US-|NE1lz#a2I7HkB7Bib
zC_HKj-GeE>CgGy+9whdp4g`BLCvp!8U&;uov(!a2t+bEnRv6HXyi9t`-YglcEo`$K
zI8GTZXYLH1F5YE+b^&9-c%dfYc~N>X1MygiCdpZ8N*OKLV7W5+5rusvVP$KTgd_E;
zV`@J%*flk^Jhjj1)aX9cTQC5ItVZ(2W=FkE;*aH-)|+*kk6SET?pjmWaNEk+>D${o
z_#cmV%sNr-bj$gX%QW$m8{|&wA?SI;%go!uC))SCU%7vKz~jI-L0?1Ap^RZ7;i?hG
zB3+__P9{WW#uUa@#oavB8Q+`m==5;nXwvwZiR6j1<0+%5!{;8Q^`_s>XwIxTUvlAM
z)|rdpmprp=bM$iM@_6#8@((Vr7Q8HcP;{fXs3iGH;8nY8TBRaov}JqcixtC_ZB<vQ
z&(`?YTGlDob2eZb{<=Qi*ws{eBju)Fvqg(SE88vPt>w07?YBCLI#1vB=rX<|d6)j~
z?!9;SA9XkN<n~7QIX}?p7apJ-*nBuW*z>4rDD83J6N{$`!z{xG&lW}=KCd6md=WHe
zF)la3F!5t@`sLkMS6?Sg5vR3gcxTbGOK%>(y*_twKH{Cjg64anMViI^4{J-a%g<LH
ze(YYoyLRud-cLiHC*aZaV}tse#HQ(1*!QCCfgc+?yt~$WY5Tng2XOpFKtGg+ut(A%
z-=j*=VHi8C5{?;<#;=l0k&Y4u$@<72Q1np_Pz_U$(Jatz(CstOF!D30Fk7<(v!=0C
zv-fel=R$LH^XTz<^JVZi3XBSF3Nwow5p@<z60edNko+Kxkr9;Dk#m)gS143$Q5sTy
zqw-A^tHz`*q@kp#r)8miT*qBEKrc)`${^M-!6?x<@o1vS>0=3|@n*5+(H4=G;Z`Bm
z0XDw2UUnY#t`5ZG&W<NeoN#h<c62#GbaHieBf7hKczXJJ`FjWZMEFMgo$^l#ND0gg
zx)_`nQW$zAtSr1TqB^oJs{Uj{bVE#iY<*nqsjB$0ge#{D6LXTzpE;Wxe>ObDFO_)C
zCe0{aEki1k_dNXt+=U-mA1_W_8p^(%Q<IyQmz*DR*{Q&|P@#yYn4);6Wa-M-)$VKc
zrTJw^<pCA;mAX|@)oe8+HQ#IB*NxV9H#A-^X-sd5zTtDzzS*!vsa50_Ya2Pdf^Bpx
zbxz-Y-ZgON&fVsFb@$7<i+XZ;v-&a~r1qx_oPC%wnEEJVDC=?lldDf_hFhL>j|@Mb
z9sM+h7-yIepVWIvd=>Y)XzKR#)XeT1j<?!#-t!mU-CCGlggyu?nJkB_<bCW|eYv*z
ziR!cPy2cmFuWlPb-=a6;w@!ag*pB@X{?m8o#IEU{#=iIg)35lmfin;%)QHeRj3K>H
zI8-@&65hs?W6g0$Tn9b?K9MevmJ{6JljSOT6GbGYHWfM5G<6M41g#z&E8Qx6H$yI?
z50eHn6Z1ODBi1suSavH8F-{EUJXaTYHjh8AJ|73)7XPq7g<!OhsW6Z5mdKE3k(jTz
zvII_ILb6OMKw41-$PCI}kh79!lAl#5RdiS4Q~ID>t>OirQ5IDz)!g7S$y<#pnvPn`
zTCcP(>sag3>W=B<=vx}l7>pa{8`&AN7|$LpGx0noeC)GnyV)so9SefRgyl6WA8Q%w
zeVfO&`F8I1(hk7k+3~B6fhW|RD4pIpx4EPekGo2^q1>k2n?25Xx_BviQ+coYJoGK~
zi}SY&kPV~?{2VkK+z^r;>Jw%VE)ao<Sd8qCsyuln+BHTamLqmAZW>-y@)AN%A4?QY
z!X(X~xtpASHaNvFl_z!g+(cSqdP;^<ru=#G^Gg@HvkESTTr$e$&EC&>mD`$^mG5`i
zpn$&+Rk%>pUtCp^dd2Um*){o6wlZ|t=klqF!OHfk>gs};%-W>7nEHr@(CeX%5lwM7
zQg7xp*S7SwzHLLbOLn+*Uc0?`NAB*$d)wWCJsW)~{h|X4gV%@BpPU*_8L1qd8t0!(
zdySmVd!st{bK%K{=9Rj&=Ffv)KX1|hFxkC)82{hg(&3(fkq6-NB>?O?0kGBtAd?QJ
zm0$~|LIBLj0I*U5i1iA9XzK$|?dCuG2lOlFq=GX}9v}f{nuc(O=>uZH1yBw;!3bD_
zU{(i`gLA_m=mOLPjX+-zbO<GcGvX|w5itp8bfQQ*WD2qkUVrG}Of3%84DUniXj}9p
z^ivET&bv}EkFcb0hLww*!SUjPaCh){ydAy{ze{3HQcDVv9w+S}&=VpFugO%&D#;1t
zQRE9078DOC<tZDexT%V%>8W#QsA+O&>1m7Uxak_`<>>nu%o*kx!<ev4CCswS!z|vc
zXx2(L9kvhb=^T<AuQ(IA#JHxo(|J^QKJ!-bo#1ES9~a0Gv=U+xdMjKf5+SN1MlZG|
z-Y;=YGG59>T2DqomQ{`*59GHMHWa@qZ7S~^!Kl)z@vEz7SZjuAWovinywxMoS2FN7
zEH|1t%4A}H?2754xrD_j%Moi{n>gE7_6iP##}7_;J59Lg5Ifz(-D^B~y{dc!eQ)?H
z1`GsQ2d{)Cgfm98MOmHv9&;s5@6?xs(nO0hxa6LcxN|CLdl`M_GqP+i31t7w9nHU9
zkY40hVt!S*RG^%pl2DDR1@+)Ms)_U_Lks^c#r9*J-d)LeEAIFAEIl9{kQ}rbihXiz
zxOZfJbZ?wtQtXx5l+ld&8>=~scSi5kK8P(dtn9DO{nh=s_)Emb(M`^+uiKA)7VrA)
zEB#tO5ODlSVZM$P@WWh#2Fx+Iz|6u~m`%6|24UXdCqxG`1g0=2kOkd@#-Q&AR(P%P
zMdTpvAy(jBM;jT2tUyk{D~~EF3{{U>K(nFk;T(JdLx-`&6l3PF0@xsI7Y>87!d2q7
z@J9GD{0|aKlAELyq`{<P1SLWl8G$UCY?<7Fe1t-eqMK5MvWrTEsvFK%hiUBK+%$p?
zLsvvEML)#g&IlN<F{v}XV@_oeW0_<<%_hn=1;_jd$2w;fmm@bl_bAT=UK2hFz8Cxj
z0?vX$f}2A9!g(UzqH1CcVqe9dOEgR7NySNf$XLi~$;ru!DDWzBDsd|Fs0baAQPog0
zQ+L;h)y&gs)gIT`*5lMSHi$92ZZv<C*~A7u*Ss=kxA3yOWsS12wQaP=IJh3~J;CFY
z>in5#@A}YP&ZEYQ#XH-V)Gsvv6_^~14ao?j4lj=6k7|w9iW!UZJhhvUlPHq(FxfQ)
zq?V>>q`%8dxgeZ1aw#H*HTOZjUjc35y<*QR6jwV-iRB~}tyPXS=-S45n}+?ysv9OZ
zzqJ(K(rR1j$hs}xHG4PtzG(M&@2Lj@{VyISJQ5#z^W@U7{<DMUcV5Jft50HIPQ0$2
zPMURnYc#L4Ah{^9q_ASN>hV|l=i6Vte3RLV-yYuK+dKCw{z!laG%#N$3ABJM%p<0O
zYA^skKqQbP%m$r-WBwLFh0ujLomRwONMWQ8vL5*f<`CmhgJ?Rm2f718hVj63W7)9r
z*mpQXTq~XnpG|@xNg&xFjU_!Gq>|CVvs#J#1w}9=HDxE2J2egUAWZ`85!yYvKKcv>
zJ4PYKJ*G+KW|m8=VQlv7TJY|}%0<qN;Qq<;jdz`Io&T%Amf)T+R)kiRM@&}SK*Cw_
zq*ShSi_DnpraX&+wqk%%q4J>0wyKDli~41a=UN19Bb{{JVSQ=?d&3H&&qviwE*<+|
zre<Dfv2EpG-EAvqm*;>!9^?4cDF}{Txa*#Kx+jZQvyZXwvVVG@WYFu7)G)>HwaCho
zPBE;pGpDX4cqED@Z6)`nTsY^LE}F4-ek7|Lj+#LpTmF}Vfuf?4z^j_2v}GSEI;v7@
ztn0YySFg7=Mcq_r{?^*qM(m*I?Cd&z=li|$-7G!jeOwO;25=992S<m5AICiD8%90T
z7zuw~Iy(4bWgIcVFe&g-@|E0cxhctM!5QXRlG$%>X5MzsmCeV$vtN*Wk9q%cvGzm6
zlGZYQ`Nc~9M~79`)tR-DzwAEIeH!_EZe4SI`^$~5?i-97Prt=)N^Q<<UHq>3ePg@o
zht*Hi&(<B6UGm+oJ@<X`{k8*#!}y1J%19sX9snX}!Sv&sgM$rz0I=Br>|HuI*eO3a
z*sFk(4fq>KkN@xQ6^F(cm~$_2K14li9;XkV|9<@!M&f%8Nam8p00009a7bBm000XU
z000XU0RWnu7ytkil}SWFRCodHT?u#;Rkr@KbUNvfeG_5`YY-wNfPp{+o{ADgGcxep
z5O;8ydCWk3pWowCbe1RjK4lzy;4&jKqk}U-a1=+ud7z@;LLwlFC>S)v1jwFrI_XY2
zop;WyuIf%_F<t2n-Ic!IS6#QNZr{59Ip>~x?x|CCgE~7q5lBOq0>MKUdH^|7ARquk
zTn+*P5D<VNPJtl~KcE2u0x-lWFvQ`9_7{IpP!Ma2Ag8UrfAvN=8&%31Xc{{x^H{Co
z5Bf3s5%3VAbLh&IH$@c|uDl{TIw}*k&TyNplZ9bWQ*bafH>lHMG@8ELxbaVWHf?&T
znHpfF&E_pZ&^rD;1;7qozi0Q$(`V)7{8<+kI>wdbHk%E>!9AN2eO+^{$KB)hHtVU6
z4;0@%KYw`%{kM%aj|)L>`1``u*EM%B_Ep|f_<QCY-I>7iHT~t6&rZsneaT;XVt##n
z3*O&%0=#!k4Gq$@x_XoAC663)d$?Wm=U<vc73yj8s7C4ye?8%O?|sJH9hZdp8`Rda
zV)AbC6#gCsfWR{`**=IUEL^cLOs^a53WvcKmwX4axn1m?g`lRU8J`}Mu9cL0kEW(p
z*S~w@nA^2_^<7tUJ89F|+dbgAd&6H>XTrha?_sgD)BZa!4dhf)W5g$)o+5f!@!6p=
z7>#E6lGpa0z~7?)*juclePn!mT$U>W2F?VqT7?}(LqHHhL#3+DoNXk5_#Pb{(lwSP
zZ<=X|iSbjYeFoatR`H}3=!RdX3qeSTbc>FTPC<i4yKox}dW8l|7rAq|($3Sz1I6E)
z0PuMxfT6vuJ<7S!uFMY$3&G48V-OV;j??9JIDGgtTjt^~;o&;W${mB481A6nNfvWk
zG}X)k!0qh~kGnVT5kF&`L0DN3I5DFSlBQz~XE21}l1oOBfJjYXO~BgEYVC0M-fop2
zDE{6E#$wJW!`duuVRDQ5Vt#39HNO0^LN1EPNWEbGuwmiKId5c0K$-Du1N3`J+mCki
zh@bx#O+t{9<6v}qdxy{j&CM3X#uzYdYMN4g6Ac50s6((}=&dwR{Cx?4=mpk}4k7Lh
zja}8gbI(8i^%;C~;;e!(COSeeKO#br9p##*Gxqi|^SQmA8UnbUF8H0>&5WKoW3<xS
z+uAzF<~aqE$%3S$Xk2`8s#2WoA?J>vT<}n4p!j<e0B)WjD5-NnRG#WdwxQH=@&k}$
zOR8R0dom`Lc=?)q&RcuB7q?Ti)x-Dbg5R}n#wZVUe2wiO-Z*u-^!Q-=uplma_-CN_
z8Dw8c=~1luSn8d}KeqH68(WbvJ_VB|4Oi+hk~;kO6RAfiC;c;{;CF65PXUWd$`m2j
z>ve)Qtntp05&Y$`N~L&mauhNrjZlt#E%Rdnz*4RdA(~WsS0P~4Cker*^h9K3rID79
zAhx!)2_f*-6tD+E@|~5o_HbR*DQEm#fix64W;xPOIEsuwz3>ej`Mg}wlx+M?%^s;7
zt7<_1|D+24j|zb6{d*Duo)R*nQ%A&N`m}UK6}Gim#oV|jr-^I5{&3u6Y!z0&JjK=N
zf~iA{0UNr_&1RH*=FkdaRxmwXu@ih1pW6b!KwO1@&&hNB<m70YDveg=#B4Xn!v>f0
z=VYU~zns|bF>|Ig{pE8Oi&e4q8Sf>;d>$HnJ*g4^2E{@!BWJXj|MK2>t{)#4iCiKM
z_X3_Wd3!22SVWGECF_5t9Wx1ebd<xtk&rM9vu2GY%sRn*f5AMUxid3;-rS5H$2b;{
z-mVQ6!@O~ym}o<<xL6?O&7IIAJ-pEtPt>Ve1IRabo*K&M<b7Y)<TC!a?+aI7*L)G{
zfCa!8K?T~>e+mp(08G`jsI~A7O*rz=A?*I(Ym_y4*Z<vrc-}ABZsXaPWE_55X~M~}
zTIJaLcOUp224n-2zVikie7yn%YnEfm!5#SRe*5FmzkQB}p11+o{Fp|=>BHj<`2EIL
z@XCfeu<hfsczydWJ6bb}o?S)x0g-o5D5qWnbezFoiTRwunk%qdX1+L6i)|mA5jidY
z{0}jolhGWEzbnk=GvPS&`Ekzpd{Qp}jg2@(@Fp)^ip7aWa9kF~U5c$wF9JUgt8d{b
zE$t=?X?=dwwZ3ox^kn_>GtW8G6RGFlFM<@CjE-OtU#5a;0kB%yXw(N%<3n(~<Qsu8
zZ|+#=eP8Q&gdZK`A?MLQF(CmT9+sYjF8fKIN9rETN>sBeG(H{~)Y9EAyo%kT#Rg2j
zpdOnacnjrpoDswQL%S&=xD)LJZ^c?^7~tUKxVSW2U-+UJ`I8c2{Q|sd4FLUcTr-0M
zaqMa26wFKpz7U~s3AlNV<LRJWL`!?_{RP)vp+K58DGni_Hng{J{G%C}6UQspd^w|9
z18~65>^qhrHMbm9<`9gTLcVV_VCkY<jGCK;u}-q#%LEum+wQu00v693hqg8|E+F?e
zr#1I3!kny;lw%R`{p7X)SelGzum&R`J{*=hGlr!kVHo3VG@_P-B6~JTzO@CWjwnQm
zSR&sif;%k~Gk*Hv$KL#GDAqtA4UzOxf>cW$bp+1aV?*4j`n;5NQvl5P$NHC1)DVqF
ze?14Uta}S5dTDmrRR#Fn;tPAZ>c6M&cw`%zt17X5(`x+mXPZPMYENh$xHA{IIn#Q&
z<WuL|wgH?ccYQ_s*J*Sb=&abC^R8TNFm``&#I=ks9J{Fj!m;GPmSYjgeN40tkN)>^
zG}YF_5*3HIuofIEDMeLB1jc8M#;C+D(d52>)gx`#@~i9ZqkAV_+e~x*&R~QFvHtHw
zX=O8P?QIyJ9<S9QfZh~|rm@{l%9`SLom|1Yuu=8~?mQv6f;~cV->Ss9*B|&g;0hMp
z3Alm-uHb+xn7Ts16&!E{`__2XkJh+p1UhOAxPk+&;D9SQ;0g}7f`^<d*f9ujPKWd{
zNwA(ei(>~4p*<oBsnf?}Y_tY_z+HhWM{(ruZlLc|d{fqlC~^zWp3Kn@&XYx@s8428
z+W?HF@nkVpZ73FT0nX?mGjp-|=hGB_f%7#j<?C8Quwu<K-PpY#ZzNXSa*ba=b=wuJ
zprVklr=6G|NiR*v4kFXBPw?d4Dy67aGq*W0e|U1NJv0<}=fG*4{I=9DV5`;u*19?o
z5{B?pJoNC@$d2dNpxSZdpKs&0Z&je7$@!^)U(mv>Mp`Hum_uHM8Ep9TllPO>m-^Cs
zpVwg1bK6i`-w1z*2vDs7WXVaJJHyU=rk@Vk3#W^iKzdl}7D4^3u#E2B8*>%rGlt8u
z5=Bg)^vMF>N2OW-kTeo=C=#;#Uwg6hiz=At%UPznGuZL$9uX3jIcgXzEoL+}ne7De
zePX!NLIZ__1sfvp<d*$}QF`*n@c|qD`l$#Bn=o#aJJp@HvT{duzX6s;924PRO?YW@
z5nghs!8um&p{H&~w#Fia`1v1)e}&ROaMy!>aY5fTR<P*VpVM{qeXU#N9z3DGRUK^<
z`8ISD3xNMj_Il@&dJSMm!=vkNfHgFv_xE4u6KU4aaHMFg_cJ-U=WN?_Kka)L1Nw>(
zUH5HKQ7-^w@TCk-ATqS$+;^2Y-9Yg{p~En8>~LcE&~OCN2SO-y!qgT7qsff0kWR!$
z^D81!lBm$TfXL;}=Y9YJK+SF{!{d*<g%X#sAaChntiD3qVd2a){!IDu!avzV!TF4o
z9xGgkjB%soL!aKOw*koiw_`(h<}=y@MgPFg@A$O|ZKq#6{$@AkGvc*R?Si#UlQDr#
z3~i~w{&Resf;DUNF*}X?fEFV@D0-Vh!)meXqjD6C9^>=}ZDsk}pA}{0WdF3_)n|T5
zFNK7P(SF;zrP#jx9qieE2>F-K@p;gyHGt(@rI_!hEt)McpP}lbFn3v=a0JCAI=-Ld
z^HfmLKw}#PgVO)j-n&3BpR3@}{)WrPilHHGIK3w22T8R6=u<`rMwjnBh~jFy5zt}A
zN81hv!KkMXNNPDnh1mq7H@>uwma1@k3;2!wtQCOj+9tn%uigkWBw{AL|5)BofhX2&
zA+XZ302%fCsUzg9CimQPVv`f;C6O8|{n>ML#6sZcPqU_9<QKRdEq1%Pk+|_c^D$ZZ
zpTp?8<hlYZ8Sj`5DMd7vdH&JsA-;|ya5+f%=&CEBr`QO8Na{5Jql+JdN1k1QvXkYg
zs3w&k5rOovX-K7)MutMt)1dz*eGsClwG%b<Mnp!CZzRsYY@Y!MK;y}yczWw0oN9FR
ze!+NJMke*DMR;s+x~Aozt~D6zfBzb)s_VMoY^H}`58OE$vvRV#ocF~8^#ULQLiXc}
zUQkuE$PY|=DWHMVx2^eKu~LOA6rK$<8jY@}eezhf09Y%}*qLuY{&mxkmF%SY=ox@6
zF~uhVd#X`JdPSa>DPe!$!>g7c<pOq#ONp0~Ns93gcerW+2#u6Nz)abBxO;&=#J<Sv
z@ytFfgdg<)OXiP7*2GlLRC*q2ahR^|-aI}H3;8oRM~>oyleK6R!5=0O9Kit+4(r(6
ziv6QJ8-P(X4Sa3SakRGjFIv?a0G4_jZD3}d!^RD-cH>&cq5?d2jrKkeAp_;!Ur#;&
z9W7Y4e9epUX=T6m-g%gom8l&2YDT>Vpn#D2K2TLOYC9;D1)wkDRn>N#8T3J_^Lk0W
z2GEDo5^3Wxdgdfd9w7&WOIUcVywJ$#^9sz{H)rNATQUdN%*}+3f?}K#TL)6Cfb&`3
z%&Qjw3IaWJ_$1z;4dDsM&%YQ~=42pUgopbkSWmW!9lu+5e2Bl(Hp~!=)psw#l#5d7
z<59t4!9`Er%bRtn7l4p3WRMY9&31sf7Q0{HC$^-K>G(;07G_Pk5PmWfQbk{$>nD;C
z$aX+;iw(co_@<~Qn^p+B=a%_MiWA>XQ&sn1{z<(6(1#*dufHEF>#Fe8m!&8!F2%dw
zHlg}-8UFYJZG<8tdn)d^eHPNC3G-m$^7_440RBMV3*u1l6Q_-M<NrRRMWepLk*_Lo
z-`bbN{R`%2;?^Z|DbJ;SGxcT1^sB{@@yW=TI0_Rc#C0ow;>ckXuK!rmQ$k)#dR$sG
z@^U71!@qOSF|2)@pOpG;Qm+AE#NKTmpy<6aRJ-8I$ex7UR10>zRSMI&Dx4*+aC%oe
z$>ksZdHCl3@33X-u5M#~!F>8s>bP;(@Z1iZ5DQ57E<W`F&{Exn`H?9pD7b1W{*9JP
zOVBg7&1l$Lik;?q)RUhiU1Oo)Yy>(pe>^RmdH=2Rkv1Y;;r0f4a|kUQI?AO7tZbEf
zJ(*E203jiWBR5FKRnt*$=_L9l06hS)bRb+XpPQ(|6)W>G1u?i-W6WoCJgU<Q$tb+@
zD$Jffj@}LJ{)WH{zrK~CgV-P9o6Kh%ssy0+_)+}f)#J#erD6&w@S}9^J6Kv&%Joq)
zVCl;1aO2ciq)r?I|MikfXnOjsHU;7FXSU<j4Ywhj0zZnsHR6e<UbYvt;`zsJK|)+S
z62!-nogR|)RM~kx=eB4z9^a39Htt8sxmx;YG^zT!qj*qZKBH+t(F@!0k$*VC+$MjO
zDHQiTunA>lRkTWYJ9y;~2lKhQP~5|72z2_#^8q&npdI^OKWZnM4)jd~l<t#{4Y+Cf
zJmk?iep_h!z8~!6w6%M3%WgX$8wEf`;a??H_G2r4_wX%<Bi{h4{7*`b`2}z{1;B(K
z1YRBPTX!e&hj*a(pcMU@bm?3ynVpH1H&2p)w`(&X<ooZ-37-u`n^q8?b=dLG_XXa&
z|9vVMfYDf2c#}A8B~Ph8B=-cMr4A=W1b8$?C6oGZsl=!338XY{hCK?{ZdF8R@clrx
z<=e!HAY|AD>xFIKK%PKOm(9u+`!IG4P>PAtq9@Rh0<!HW6ucJiefk6H>JE!{0DuH!
zkK`y|6ZXDM&ju*fYcM2?dkd?0BQd?AvKl9=rI$l^%Bzo%82<d<-V^wnjLl;2>pwp_
z3!t@d`N^j}MPee&>2}gr!FRvB)4o^~UCPYDMfxiI>b@c+MsVI_ZG?n%#SdILF9)yD
z8iBv~&32h6$j=)^`5;_--)1F7aK==Pycf`JwRRcIa&EjD`NGhX@h9M+TM4YCmA;oJ
zrO3=nv3MeD1n(z%`&dZj&7(JU#eehVv~0XE^yJ%^arZ3+;^s6cinJi_LRv*8MlRsh
z{Xp^er2%-zvwii|iPQND<~cxwB;)S&_u$&{D%8_7aQMh%>8YP30yAe!z=De>;j*0J
zN>6b7(K|VAAJyy)=J$-BZpMp7n5{I{+sN@1<}jm{UYm<6az<l#QupT~-a!ux6n{`>
zC)2KLBDKeY!To$ha&qG2BZqfAotPNM^BbQ^H8u4$*;5z(vZ|_v=c1LgH4&aJ8cR)s
zhZ25=_;#ffO9d0sLd30K^&jiDoI6+3R|Htse-FYDw`bL=buUu;*yY6jR@v$9iMtOO
z{Jm)a77X@ba<Dr`c8-aSjUKyf*>%$<P8>f%7edh>l!!{woQDqvAyLn?wOiY*$B%zo
zv32X~pEWczvH$rLZ56cfy6vr`0a$epDA9d}4E`PkfT>4BU?%e$j!CrfB%e1P1~}M{
zuQ8DZRRHLI>|J6XE5CNbPoY`u^Tv~L_DESt0J@K9biv&;RPgs@1TwMtC4bqg&n_U&
z^RqpU@fmCZV8(Krcxd8Db|Y=v9v+%_sqO*ye5%7a4GH|cY5=AL^#T?U?(IAraOf}Z
znfd(s?_l?Sx}{(;kM%5!ES&ry9?r8?uz9NYQ(Ynr1^j&q08@d8z|&jaWMSaE-1`Sx
z2*lKk?$1KN8*2mJGw(g3`l+riN$dE3Q~;P7L<D9yC)d4ncJO{>Cd=wx?7hW&8J3pu
z_e%g|LIn2Oqk!C_wTCQ#s9zKa2tdEcq}@UR0njdQ`-LnZ0R1A9b_)drK)bx{7qWl=
z^ovZ|Eff#{?eex?$N~b;FEVMjP(T2*%iDe-`+v|7m{y$1dn*6{002ovPDHLkV1lnB
B5rhB$

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Build.html b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Build.html
new file mode 100644
index 0000000..24b3670
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Build.html
@@ -0,0 +1,27 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Build</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-Build.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-Build.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-0.1.0.0: TensorFlow bindings.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.Build</p></div><div id="table-of-contents"><p class="caption">Contents</p><ul><li><a href="#g:1">Graph node types</a></li><li><a href="#g:2">Ops</a></li><li><a href="#g:3">The Build monad</a></li><li><a href="#g:4">Creating and looking up Ops</a></li><li><a href="#g:5">Modifying all nodes in a Build action</a></li><li><a href="#g:6">Internal Summary related bits.</a></li></ul></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><span class="keyword">newtype</span> <a href="#t:ControlNode">ControlNode</a> = <a href="#v:ControlNode">ControlNode</a> {<ul class="subs"><li><a href="#v:unControlNode">unControlNode</a> :: <a href="TensorFlow-Output.html#t:Op">Op</a></li></ul>}</li><li class="src short"><span class="keyword">data</span> <a href="#t:Unique">Unique</a></li><li class="src short"><a href="#v:explicitName">explicitName</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> -&gt; <a href="TensorFlow-Output.html#t:PendingNodeName">PendingNodeName</a></li><li class="src short"><a href="#v:implicitName">implicitName</a> :: <a href="TensorFlow-Output.html#t:PendingNodeName">PendingNodeName</a></li><li class="src short"><a href="#v:opDef">opDef</a> :: <a href="TensorFlow-Output.html#t:OpType">OpType</a> -&gt; <a href="TensorFlow-Output.html#t:OpDef">OpDef</a></li><li class="src short"><a href="#v:opDefWithName">opDefWithName</a> :: <a href="TensorFlow-Output.html#t:PendingNodeName">PendingNodeName</a> -&gt; <a href="TensorFlow-Output.html#t:OpType">OpType</a> -&gt; <a href="TensorFlow-Output.html#t:OpDef">OpDef</a></li><li class="src short"><a href="#v:opName">opName</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> <a href="TensorFlow-Output.html#t:PendingNodeName">PendingNodeName</a></li><li class="src short"><a href="#v:opType">opType</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> <a href="TensorFlow-Output.html#t:OpType">OpType</a></li><li class="src short"><a href="#v:opAttr">opAttr</a> :: <a href="TensorFlow-Types.html#t:Attribute">Attribute</a> a =&gt; <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> -&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> a</li><li class="src short"><a href="#v:opInputs">opInputs</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> [<a href="TensorFlow-Output.html#t:Output">Output</a>]</li><li class="src short"><a href="#v:opControlInputs">opControlInputs</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> [<a href="TensorFlow-Output.html#t:NodeName">NodeName</a>]</li><li class="src short"><span class="keyword">data</span> <a href="#t:GraphState">GraphState</a></li><li class="src short"><a href="#v:render">render</a> :: <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a)</li><li class="src short"><a href="#v:renderNodeName">renderNodeName</a> :: <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> <a href="TensorFlow-Output.html#t:NodeName">NodeName</a></li><li class="src short"><a href="#v:renderedNodeDefs">renderedNodeDefs</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Build.html#t:GraphState">GraphState</a> (<a href="../containers-0.5.6.2/Data-Map-Lazy.html#t:Map">Map</a> <a href="TensorFlow-Output.html#t:NodeName">NodeName</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a>)</li><li class="src short"><span class="keyword">data</span> <a href="#t:BuildT">BuildT</a> m a</li><li class="src short"><span class="keyword">type</span> <a href="#t:Build">Build</a> = <a href="TensorFlow-Build.html#t:BuildT">BuildT</a> <a href="../base-4.8.2.0/Data-Functor-Identity.html#t:Identity">Identity</a></li><li class="src short"><a href="#v:addInitializer">addInitializer</a> :: <a href="TensorFlow-Build.html#t:ControlNode">ControlNode</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> ()</li><li class="src short"><a href="#v:hoistBuildT">hoistBuildT</a> :: (<span class="keyword">forall</span> a. m a -&gt; n a) -&gt; <a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m b -&gt; <a href="TensorFlow-Build.html#t:BuildT">BuildT</a> n b</li><li class="src short"><a href="#v:evalBuildT">evalBuildT</a> :: <a href="../base-4.8.2.0/Control-Monad.html#t:Monad">Monad</a> m =&gt; <a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m a -&gt; m a</li><li class="src short"><a href="#v:runBuildT">runBuildT</a> :: <a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m a -&gt; m (a, <a href="TensorFlow-Build.html#t:GraphState">GraphState</a>)</li><li class="src short"><a href="#v:asGraphDef">asGraphDef</a> :: <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a></li><li class="src short"><a href="#v:addGraphDef">addGraphDef</a> :: <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> ()</li><li class="src short"><a href="#v:flushInitializers">flushInitializers</a> :: <a href="../base-4.8.2.0/Control-Monad.html#t:Monad">Monad</a> m =&gt; <a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m [<a href="TensorFlow-Output.html#t:NodeName">NodeName</a>]</li><li class="src short"><a href="#v:flushNodeBuffer">flushNodeBuffer</a> :: <a href="../base-4.8.2.0/Control-Monad.html#t:Monad">Monad</a> m =&gt; <a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m [<a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a>]</li><li class="src short"><a href="#v:getOrAddOp">getOrAddOp</a> :: <a href="TensorFlow-Output.html#t:Op">Op</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> <a href="TensorFlow-Output.html#t:NodeName">NodeName</a></li><li class="src short"><a href="#v:addNewOp">addNewOp</a> :: <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a></li><li class="src short"><a href="#v:renderOutput">renderOutput</a> :: <a href="TensorFlow-Output.html#t:Output">Output</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></li><li class="src short"><a href="#v:colocateWith">colocateWith</a> :: <span class="keyword">forall</span> a v b. <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v b -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a</li><li class="src short"><a href="#v:withStateLens">withStateLens</a> :: <a href="../lens-family-1.2.0/Lens-Family2-State-Lazy.html#t:MonadState">MonadState</a> s m =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> s a -&gt; (a -&gt; a) -&gt; m b -&gt; m b</li><li class="src short"><a href="#v:withDevice">withDevice</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="TensorFlow-Output.html#t:Device">Device</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a</li><li class="src short"><a href="#v:withNameScope">withNameScope</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a</li><li class="src short"><a href="#v:withNodeDependencies">withNodeDependencies</a> :: <a href="../containers-0.5.6.2/Data-Set.html#t:Set">Set</a> <a href="TensorFlow-Output.html#t:NodeName">NodeName</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a</li><li class="src short"><a href="#v:addSummary">addSummary</a> :: <a href="TensorFlow-Build.html#t:SummaryTensor">SummaryTensor</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> ()</li><li class="src short"><span class="keyword">type</span> <a href="#t:SummaryTensor">SummaryTensor</a> = <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:collectAllSummaries">collectAllSummaries</a> :: <a href="../base-4.8.2.0/Control-Monad.html#t:Monad">Monad</a> m =&gt; <a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m [<a href="TensorFlow-Build.html#t:SummaryTensor">SummaryTensor</a>]</li></ul></div><div id="interface"><h1 id="g:1">Graph node types</h1><div class="top"><p class="src"><span class="keyword">newtype</span> <a name="t:ControlNode" class="def">ControlNode</a> <a href="src/TensorFlow-Output.html#ControlNode" class="link">Source</a></p><div class="doc"><p>A type of graph node which has no outputs. These nodes are
+ valuable for causing side effects when they are run.</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:ControlNode" class="def">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:unControlNode" class="def">unControlNode</a> :: <a href="TensorFlow-Output.html#t:Op">Op</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:ControlNode" class="caption collapser" onclick="toggleSection('i:ControlNode')">Instances</p><div id="section.i:ControlNode" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> <a href="TensorFlow-Build.html#t:ControlNode">ControlNode</a></span> <a href="src/TensorFlow-BuildOp.html#line-144" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> <a href="TensorFlow-Build.html#t:ControlNode">ControlNode</a></span> <a href="src/TensorFlow-BuildOp.html#line-92" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> <a href="TensorFlow-Build.html#t:ControlNode">ControlNode</a></span> <a href="src/TensorFlow-Nodes.html#line-91" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(~) * a () =&gt; <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> <a href="TensorFlow-Build.html#t:ControlNode">ControlNode</a> a</span> <a href="src/TensorFlow-Nodes.html#line-98" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:Unique" class="def">Unique</a> <a href="src/TensorFlow-Build.html#Unique" class="link">Source</a></p><div class="subs instances"><p id="control.i:Unique" class="caption collapser" onclick="toggleSection('i:Unique')">Instances</p><div id="section.i:Unique" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Prelude.html#t:Enum">Enum</a> <a href="TensorFlow-Build.html#t:Unique">Unique</a></span> <a href="src/TensorFlow-Build.html#line-98" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="TensorFlow-Build.html#t:Unique">Unique</a></span> <a href="src/TensorFlow-Build.html#line-98" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Ord.html#t:Ord">Ord</a> <a href="TensorFlow-Build.html#t:Unique">Unique</a></span> <a href="src/TensorFlow-Build.html#line-98" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><h1 id="g:2">Ops</h1><div class="top"><p class="src"><a name="v:explicitName" class="def">explicitName</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> -&gt; <a href="TensorFlow-Output.html#t:PendingNodeName">PendingNodeName</a> <a href="src/TensorFlow-Build.html#explicitName" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:implicitName" class="def">implicitName</a> :: <a href="TensorFlow-Output.html#t:PendingNodeName">PendingNodeName</a> <a href="src/TensorFlow-Build.html#implicitName" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:opDef" class="def">opDef</a> :: <a href="TensorFlow-Output.html#t:OpType">OpType</a> -&gt; <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> <a href="src/TensorFlow-Build.html#opDef" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:opDefWithName" class="def">opDefWithName</a> :: <a href="TensorFlow-Output.html#t:PendingNodeName">PendingNodeName</a> -&gt; <a href="TensorFlow-Output.html#t:OpType">OpType</a> -&gt; <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> <a href="src/TensorFlow-Build.html#opDefWithName" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:opName" class="def">opName</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> <a href="TensorFlow-Output.html#t:PendingNodeName">PendingNodeName</a> <a href="src/TensorFlow-Output.html#opName" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:opType" class="def">opType</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> <a href="TensorFlow-Output.html#t:OpType">OpType</a> <a href="src/TensorFlow-Output.html#opType" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:opAttr" class="def">opAttr</a> :: <a href="TensorFlow-Types.html#t:Attribute">Attribute</a> a =&gt; <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> -&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> a <a href="src/TensorFlow-Output.html#opAttr" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:opInputs" class="def">opInputs</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> [<a href="TensorFlow-Output.html#t:Output">Output</a>] <a href="src/TensorFlow-Output.html#opInputs" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:opControlInputs" class="def">opControlInputs</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> [<a href="TensorFlow-Output.html#t:NodeName">NodeName</a>] <a href="src/TensorFlow-Output.html#opControlInputs" class="link">Source</a></p></div><h1 id="g:3">The Build monad</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:GraphState" class="def">GraphState</a> <a href="src/TensorFlow-Build.html#GraphState" class="link">Source</a></p><div class="subs instances"><p id="control.i:GraphState" class="caption collapser" onclick="toggleSection('i:GraphState')">Instances</p><div id="section.i:GraphState" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Control-Monad.html#t:Monad">Monad</a> m =&gt; <a href="../lens-family-1.2.0/Lens-Family2-State-Lazy.html#t:MonadState">MonadState</a> <a href="TensorFlow-Build.html#t:GraphState">GraphState</a> (<a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m)</span> <a href="src/TensorFlow-Build.html#line-199" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><a name="v:render" class="def">render</a> :: <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a) <a href="src/TensorFlow-Build.html#render" class="link">Source</a></p><div class="doc"><p>Render a <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code>, fixing its name, scope, device and control inputs from
+ the <code><a href="TensorFlow-Build.html#t:Build">Build</a></code> context.  Also renders any dependencies of the <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> that
+ weren't already rendered.</p><p>This operation is idempotent; <code>render &gt;=&gt; render === render</code>.  However,
+ rendering a (previously un-rendered) <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> in two different contexts
+ may result in two different <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code>s.</p></div></div><div class="top"><p class="src"><a name="v:renderNodeName" class="def">renderNodeName</a> :: <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> <a href="TensorFlow-Output.html#t:NodeName">NodeName</a> <a href="src/TensorFlow-Build.html#renderNodeName" class="link">Source</a></p><div class="doc"><p>Render a <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> and get its node's name.</p></div></div><div class="top"><p class="src"><a name="v:renderedNodeDefs" class="def">renderedNodeDefs</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Build.html#t:GraphState">GraphState</a> (<a href="../containers-0.5.6.2/Data-Map-Lazy.html#t:Map">Map</a> <a href="TensorFlow-Output.html#t:NodeName">NodeName</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a>) <a href="src/TensorFlow-Build.html#renderedNodeDefs" class="link">Source</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:BuildT" class="def">BuildT</a> m a <a href="src/TensorFlow-Build.html#BuildT" class="link">Source</a></p><div class="doc"><p>An action for building nodes in a TensorFlow graph.
+ Used to manage build state internally as part of the <code>Session</code> monad.</p></div><div class="subs instances"><p id="control.i:BuildT" class="caption collapser" onclick="toggleSection('i:BuildT')">Instances</p><div id="section.i:BuildT" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../transformers-0.4.2.0/Control-Monad-Trans-Class.html#t:MonadTrans">MonadTrans</a> <a href="TensorFlow-Build.html#t:BuildT">BuildT</a></span> <a href="src/TensorFlow-Build.html#line-198" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Control-Monad.html#t:Monad">Monad</a> m =&gt; <a href="../lens-family-1.2.0/Lens-Family2-State-Lazy.html#t:MonadState">MonadState</a> <a href="TensorFlow-Build.html#t:GraphState">GraphState</a> (<a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m)</span> <a href="src/TensorFlow-Build.html#line-199" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Control-Monad.html#t:Monad">Monad</a> m =&gt; <a href="../base-4.8.2.0/Control-Monad.html#t:Monad">Monad</a> (<a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m)</span> <a href="src/TensorFlow-Build.html#line-198" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Functor.html#t:Functor">Functor</a> m =&gt; <a href="../base-4.8.2.0/Data-Functor.html#t:Functor">Functor</a> (<a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m)</span> <a href="src/TensorFlow-Build.html#line-198" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Control-Monad.html#t:Monad">Monad</a> m =&gt; <a href="../base-4.8.2.0/Control-Applicative.html#t:Applicative">Applicative</a> (<a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m)</span> <a href="src/TensorFlow-Build.html#line-198" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../transformers-0.4.2.0/Control-Monad-IO-Class.html#t:MonadIO">MonadIO</a> m =&gt; <a href="../transformers-0.4.2.0/Control-Monad-IO-Class.html#t:MonadIO">MonadIO</a> (<a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m)</span> <a href="src/TensorFlow-Build.html#line-198" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a =&gt; <a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> (<a href="TensorFlow-Build.html#t:Build">Build</a> a)</span> <a href="src/TensorFlow-BuildOp.html#line-180" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">type</span> <a name="t:Build" class="def">Build</a> = <a href="TensorFlow-Build.html#t:BuildT">BuildT</a> <a href="../base-4.8.2.0/Data-Functor-Identity.html#t:Identity">Identity</a> <a href="src/TensorFlow-Build.html#Build" class="link">Source</a></p><div class="doc"><p>An action for building nodes in a TensorFlow graph.</p></div></div><div class="top"><p class="src"><a name="v:addInitializer" class="def">addInitializer</a> :: <a href="TensorFlow-Build.html#t:ControlNode">ControlNode</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> () <a href="src/TensorFlow-Build.html#addInitializer" class="link">Source</a></p><div class="doc"><p>Registers the given node to be executed before the next
+ <code><a href="TensorFlow-Session.html#v:run">run</a></code>.</p></div></div><div class="top"><p class="src"><a name="v:hoistBuildT" class="def">hoistBuildT</a> :: (<span class="keyword">forall</span> a. m a -&gt; n a) -&gt; <a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m b -&gt; <a href="TensorFlow-Build.html#t:BuildT">BuildT</a> n b <a href="src/TensorFlow-Build.html#hoistBuildT" class="link">Source</a></p><div class="doc"><p>This is Control.Monad.Morph.hoist sans the dependency.</p></div></div><div class="top"><p class="src"><a name="v:evalBuildT" class="def">evalBuildT</a> :: <a href="../base-4.8.2.0/Control-Monad.html#t:Monad">Monad</a> m =&gt; <a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m a -&gt; m a <a href="src/TensorFlow-Build.html#evalBuildT" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:runBuildT" class="def">runBuildT</a> :: <a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m a -&gt; m (a, <a href="TensorFlow-Build.html#t:GraphState">GraphState</a>) <a href="src/TensorFlow-Build.html#runBuildT" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:asGraphDef" class="def">asGraphDef</a> :: <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a> <a href="src/TensorFlow-Build.html#asGraphDef" class="link">Source</a></p><div class="doc"><p>Produce a GraphDef proto representation of the nodes that are rendered in
+ the given <code><a href="TensorFlow-Build.html#t:Build">Build</a></code> action.</p></div></div><div class="top"><p class="src"><a name="v:addGraphDef" class="def">addGraphDef</a> :: <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> () <a href="src/TensorFlow-Build.html#addGraphDef" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:flushInitializers" class="def">flushInitializers</a> :: <a href="../base-4.8.2.0/Control-Monad.html#t:Monad">Monad</a> m =&gt; <a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m [<a href="TensorFlow-Output.html#t:NodeName">NodeName</a>] <a href="src/TensorFlow-Build.html#flushInitializers" class="link">Source</a></p><div class="doc"><p>Get all the initializers that have accumulated so far, and clear
+ that buffer.</p></div></div><div class="top"><p class="src"><a name="v:flushNodeBuffer" class="def">flushNodeBuffer</a> :: <a href="../base-4.8.2.0/Control-Monad.html#t:Monad">Monad</a> m =&gt; <a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m [<a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a>] <a href="src/TensorFlow-Build.html#flushNodeBuffer" class="link">Source</a></p><div class="doc"><p>Get all the NodeDefs that have accumulated so far, and clear that buffer.</p></div></div><h1 id="g:4">Creating and looking up Ops</h1><div class="top"><p class="src"><a name="v:getOrAddOp" class="def">getOrAddOp</a> :: <a href="TensorFlow-Output.html#t:Op">Op</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> <a href="TensorFlow-Output.html#t:NodeName">NodeName</a> <a href="src/TensorFlow-Build.html#getOrAddOp" class="link">Source</a></p><div class="doc"><p>Render the given op if it hasn't been rendered already, and return its
+ name.</p></div></div><div class="top"><p class="src"><a name="v:addNewOp" class="def">addNewOp</a> :: <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a> <a href="src/TensorFlow-Build.html#addNewOp" class="link">Source</a></p><div class="doc"><p>Add a new node for a given <code><a href="TensorFlow-Output.html#t:OpDef">OpDef</a></code>.  This is used for making &quot;stateful&quot; ops
+ which are not safe to dedup (e.g, &quot;variable&quot; and &quot;assign&quot;).</p></div></div><div class="top"><p class="src"><a name="v:renderOutput" class="def">renderOutput</a> :: <a href="TensorFlow-Output.html#t:Output">Output</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> <a href="src/TensorFlow-Build.html#renderOutput" class="link">Source</a></p><div class="doc"><p>Render an <code><a href="TensorFlow-Output.html#t:Output">Output</a></code> and return a string representation for the TensorFlow
+ foreign APIs.</p></div></div><h1 id="g:5">Modifying all nodes in a Build action</h1><div class="top"><p class="src"><a name="v:colocateWith" class="def">colocateWith</a> :: <span class="keyword">forall</span> a v b. <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v b -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a <a href="src/TensorFlow-Build.html#colocateWith" class="link">Source</a></p><div class="doc"><p>Places all nodes rendered in the given <code><a href="TensorFlow-Build.html#t:Build">Build</a></code> action on the same
+ device as the given Tensor (see also <code><a href="TensorFlow-Build.html#v:withDevice">withDevice</a></code>). Make sure that
+ the action has side effects of rendering the desired tensors. A pure
+ return would not have the desired effect.</p></div></div><div class="top"><p class="src"><a name="v:withStateLens" class="def">withStateLens</a> :: <a href="../lens-family-1.2.0/Lens-Family2-State-Lazy.html#t:MonadState">MonadState</a> s m =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> s a -&gt; (a -&gt; a) -&gt; m b -&gt; m b <a href="src/TensorFlow-Build.html#withStateLens" class="link">Source</a></p><div class="doc"><p>Modify some part of the state, run an action, and restore the state
+ after that action is done.</p></div></div><div class="top"><p class="src"><a name="v:withDevice" class="def">withDevice</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="TensorFlow-Output.html#t:Device">Device</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a <a href="src/TensorFlow-Build.html#withDevice" class="link">Source</a></p><div class="doc"><p>Set a device for all nodes rendered in the given <code><a href="TensorFlow-Build.html#t:Build">Build</a></code> action
+ (unless further overridden by another use of withDevice).</p></div></div><div class="top"><p class="src"><a name="v:withNameScope" class="def">withNameScope</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a <a href="src/TensorFlow-Build.html#withNameScope" class="link">Source</a></p><div class="doc"><p>Prepend a scope to all nodes rendered in the given <code><a href="TensorFlow-Build.html#t:Build">Build</a></code> action.</p></div></div><div class="top"><p class="src"><a name="v:withNodeDependencies" class="def">withNodeDependencies</a> :: <a href="../containers-0.5.6.2/Data-Set.html#t:Set">Set</a> <a href="TensorFlow-Output.html#t:NodeName">NodeName</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a <a href="src/TensorFlow-Build.html#withNodeDependencies" class="link">Source</a></p><div class="doc"><p>Add control inputs to all nodes rendered in the given <code><a href="TensorFlow-Build.html#t:Build">Build</a></code> action.</p></div></div><h1 id="g:6">Internal Summary related bits.</h1><div class="top"><p class="src"><a name="v:addSummary" class="def">addSummary</a> :: <a href="TensorFlow-Build.html#t:SummaryTensor">SummaryTensor</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> () <a href="src/TensorFlow-Build.html#addSummary" class="link">Source</a></p><div class="doc"><p>Records the given summary action in Build for retrieval with
+ <code><a href="TensorFlow-Build.html#v:collectAllSummaries">collectAllSummaries</a></code>. The summary op is required to produce a
+ Summary protocol buffer in string form. For safety, use the
+ pre-composed functions: Logging.scalarSummary and
+ Logging.histogramSummary.</p></div></div><div class="top"><p class="src"><span class="keyword">type</span> <a name="t:SummaryTensor" class="def">SummaryTensor</a> = <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> <a href="src/TensorFlow-Build.html#SummaryTensor" class="link">Source</a></p><div class="doc"><p>Synonym for the tensors that return serialized Summary proto.</p></div></div><div class="top"><p class="src"><a name="v:collectAllSummaries" class="def">collectAllSummaries</a> :: <a href="../base-4.8.2.0/Control-Monad.html#t:Monad">Monad</a> m =&gt; <a href="TensorFlow-Build.html#t:BuildT">BuildT</a> m [<a href="TensorFlow-Build.html#t:SummaryTensor">SummaryTensor</a>] <a href="src/TensorFlow-Build.html#collectAllSummaries" class="link">Source</a></p><div class="doc"><p>Retrieves the summary ops collected thus far. Typically this only
+ happens once, but if <code><a href="TensorFlow-Session.html#v:buildWithSummary">buildWithSummary</a></code> is used
+ repeatedly, the values accumulate.</p></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/TensorFlow-BuildOp.html b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-BuildOp.html
new file mode 100644
index 0000000..261e9ba
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-BuildOp.html
@@ -0,0 +1,6 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.BuildOp</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-BuildOp.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-BuildOp.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-0.1.0.0: TensorFlow bindings.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.BuildOp</p></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><span class="keyword">class</span> <a href="#t:OpResult">OpResult</a> a</li><li class="src short"><span class="keyword">class</span> <a href="#t:BuildOp">BuildOp</a> f</li><li class="src short"><a href="#v:buildOp">buildOp</a> :: <a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> f =&gt; <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> -&gt; f</li><li class="src short"><a href="#v:buildListOp">buildListOp</a> :: <a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> f =&gt; [<a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>] -&gt; <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> -&gt; f</li><li class="src short"><a href="#v:eqLengthGuard">eqLengthGuard</a> :: [(<a href="../base-4.8.2.0/Data-String.html#t:String">String</a>, [(<a href="../base-4.8.2.0/Data-String.html#t:String">String</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int">Int</a>)])] -&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">class</span> <a name="t:OpResult" class="def">OpResult</a> a <a href="src/TensorFlow-BuildOp.html#OpResult" class="link">Source</a></p><div class="doc"><p>Class of types that can be used as op outputs.</p></div><div class="subs minimal"><p class="caption">Minimal complete definition</p><p class="src">toResult</p></div><div class="subs instances"><p id="control.i:OpResult" class="caption collapser" onclick="toggleSection('i:OpResult')">Instances</p><div id="section.i:OpResult" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> <a href="TensorFlow-Output.html#t:ControlNode">ControlNode</a></span> <a href="src/TensorFlow-BuildOp.html#line-92" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a =&gt; <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> [a]</span> <a href="src/TensorFlow-BuildOp.html#line-95" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a1, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a2) =&gt; <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> (a1, a2)</span> <a href="src/TensorFlow-BuildOp.html#line-45" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Ref">Ref</a> a)</span> <a href="src/TensorFlow-BuildOp.html#line-89" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Value">Value</a> a)</span> <a href="src/TensorFlow-BuildOp.html#line-86" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a1, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a2, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a3) =&gt; <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> (a1, a2, a3)</span> <a href="src/TensorFlow-BuildOp.html#line-48" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a1, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a2, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a3, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a4) =&gt; <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> (a1, a2, a3, a4)</span> <a href="src/TensorFlow-BuildOp.html#line-51" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a1, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a2, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a3, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a4, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a5) =&gt; <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> (a1, a2, a3, a4, a5)</span> <a href="src/TensorFlow-BuildOp.html#line-55" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a1, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a2, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a3, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a4, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a5, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a6) =&gt; <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> (a1, a2, a3, a4, a5, a6)</span> <a href="src/TensorFlow-BuildOp.html#line-63" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">class</span> <a name="t:BuildOp" class="def">BuildOp</a> f <a href="src/TensorFlow-BuildOp.html#BuildOp" class="link">Source</a></p><div class="doc"><p>Class of types that can be used as op functions.</p></div><div class="subs minimal"><p class="caption">Minimal complete definition</p><p class="src">buildOp'</p></div><div class="subs instances"><p id="control.i:BuildOp" class="caption collapser" onclick="toggleSection('i:BuildOp')">Instances</p><div id="section.i:BuildOp" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> <a href="TensorFlow-Output.html#t:ControlNode">ControlNode</a></span> <a href="src/TensorFlow-BuildOp.html#line-144" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> [<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Value">Value</a> a]</span> <a href="src/TensorFlow-BuildOp.html#line-153" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> a =&gt; <a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> (<a href="TensorFlow-Build.html#t:Build">Build</a> a)</span> <a href="src/TensorFlow-BuildOp.html#line-180" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> f =&gt; <a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> ([<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a] -&gt; f)</span> <a href="src/TensorFlow-BuildOp.html#line-186" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> f =&gt; <a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; f)</span> <a href="src/TensorFlow-BuildOp.html#line-183" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t1, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t2) =&gt; <a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> (t1, t2)</span> <a href="src/TensorFlow-BuildOp.html#line-156" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Ref">Ref</a> a)</span> <a href="src/TensorFlow-BuildOp.html#line-150" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Value">Value</a> a)</span> <a href="src/TensorFlow-BuildOp.html#line-147" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t1, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t2, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t3) =&gt; <a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> (t1, t2, t3)</span> <a href="src/TensorFlow-BuildOp.html#line-159" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t1, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t2, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t3, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t4) =&gt; <a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> (t1, t2, t3, t4)</span> <a href="src/TensorFlow-BuildOp.html#line-162" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t1, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t2, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t3, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t4, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t5) =&gt; <a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> (t1, t2, t3, t4, t5)</span> <a href="src/TensorFlow-BuildOp.html#line-166" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t1, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t2, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t3, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t4, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t5, <a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> t6) =&gt; <a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> (t1, t2, t3, t4, t5, t6)</span> <a href="src/TensorFlow-BuildOp.html#line-170" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><a name="v:buildOp" class="def">buildOp</a> :: <a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> f =&gt; <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> -&gt; f <a href="src/TensorFlow-BuildOp.html#buildOp" class="link">Source</a></p><div class="doc"><p>Starts an operation that returns a structured set of tensors
+ (singletons or tuples).</p></div></div><div class="top"><p class="src"><a name="v:buildListOp" class="def">buildListOp</a> <a href="src/TensorFlow-BuildOp.html#buildListOp" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> f</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; [<a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]</td><td class="doc"><p>Cardinality of the corresponding list of tensors output.</p></td></tr><tr><td class="src">-&gt; <a href="TensorFlow-Output.html#t:OpDef">OpDef</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">-&gt; f</td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Starts an operation that returns a list of tensors.</p></div></div><div class="top"><p class="src"><a name="v:eqLengthGuard" class="def">eqLengthGuard</a> :: [(<a href="../base-4.8.2.0/Data-String.html#t:String">String</a>, [(<a href="../base-4.8.2.0/Data-String.html#t:String">String</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int">Int</a>)])] -&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> <a href="src/TensorFlow-BuildOp.html#eqLengthGuard" class="link">Source</a></p><div class="doc"><p>Returns true if all the integers in each tuple are identical.
+ Throws an error with a descriptive message if not.</p></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/TensorFlow-ControlFlow.html b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-ControlFlow.html
new file mode 100644
index 0000000..bfd5f5b
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-ControlFlow.html
@@ -0,0 +1,9 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.ControlFlow</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-ControlFlow.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-ControlFlow.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-0.1.0.0: TensorFlow bindings.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.ControlFlow</p></div><div id="table-of-contents"><p class="caption">Contents</p><ul><li><a href="#g:1">Dependencies</a></li><li><a href="#g:2">Operations</a></li></ul></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><a href="#v:withControlDependencies">withControlDependencies</a> :: <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t =&gt; t -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a</li><li class="src short"><a href="#v:group">group</a> :: <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t =&gt; t -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> <a href="TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:identity">identity</a> :: <a href="TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a</li><li class="src short"><a href="#v:noOp">noOp</a> :: <a href="TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:named">named</a> :: <a href="TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> -&gt; <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a</li></ul></div><div id="interface"><h1 id="g:1">Dependencies</h1><div class="top"><p class="src"><a name="v:withControlDependencies" class="def">withControlDependencies</a> :: <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t =&gt; t -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a <a href="src/TensorFlow-ControlFlow.html#withControlDependencies" class="link">Source</a></p><div class="doc"><p>Modify a <code><a href="TensorFlow-Build.html#t:Build">Build</a></code> action, such that all new ops rendered in it will depend
+ on the nodes in the first argument.</p></div></div><div class="top"><p class="src"><a name="v:group" class="def">group</a> :: <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t =&gt; t -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> <a href="TensorFlow-Output.html#t:ControlNode">ControlNode</a> <a href="src/TensorFlow-ControlFlow.html#group" class="link">Source</a></p><div class="doc"><p>Create an op that groups multiple operations.</p><p>When this op finishes, all ops in the input <code>n</code> have finished.  This op has
+ no output.</p></div></div><h1 id="g:2">Operations</h1><div class="top"><p class="src"><a name="v:identity" class="def">identity</a> :: <a href="TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a <a href="src/TensorFlow-ControlFlow.html#identity" class="link">Source</a></p><div class="doc"><p>Returns a <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> with the same shape and contents as the input.</p></div></div><div class="top"><p class="src"><a name="v:noOp" class="def">noOp</a> :: <a href="TensorFlow-Output.html#t:ControlNode">ControlNode</a> <a href="src/TensorFlow-ControlFlow.html#noOp" class="link">Source</a></p><div class="doc"><p>Does nothing.  Only useful as a placeholder for control edges.</p></div></div><div class="top"><p class="src"><a name="v:named" class="def">named</a> :: <a href="TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> -&gt; <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a <a href="src/TensorFlow-ControlFlow.html#named" class="link">Source</a></p><div class="doc"><p>Returns a <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> with a given name and the same shape and contents as
+ the input.</p><p>TODO(judahjacobson): This breaks when used with uninitialize <code>Tensor Ref</code>s,
+ since <code>RefIdentity</code> doesn't have SetAllowsUninitializedInput().  Look into
+ whether we can change that op.</p></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html
new file mode 100644
index 0000000..5c669e5
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Internal-FFI.html
@@ -0,0 +1,8 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Internal.FFI</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-Internal-FFI.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-Internal-FFI.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-0.1.0.0: TensorFlow bindings.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.Internal.FFI</p></div><div id="table-of-contents"><p class="caption">Contents</p><ul><li><a href="#g:1">Internal helper.</a></li></ul></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><span class="keyword">data</span> <a href="#t:TensorFlowException">TensorFlowException</a> = <a href="#v:TensorFlowException">TensorFlowException</a> Code <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></li><li class="src short"><span class="keyword">data</span> <a href="#t:Session">Session</a></li><li class="src short"><a href="#v:withSession">withSession</a> :: (SessionOptions -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> ()) -&gt; ((<a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> () -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> ()) -&gt; <a href="TensorFlow-Internal-FFI.html#t:Session">Session</a> -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> a) -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> a</li><li class="src short"><a href="#v:extendGraph">extendGraph</a> :: <a href="TensorFlow-Internal-FFI.html#t:Session">Session</a> -&gt; <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a> -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> ()</li><li class="src short"><a href="#v:run">run</a> :: <a href="TensorFlow-Internal-FFI.html#t:Session">Session</a> -&gt; [(<a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>, <a href="TensorFlow-Internal-FFI.html#t:TensorData">TensorData</a>)] -&gt; [<a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>] -&gt; [<a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>] -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> [<a href="TensorFlow-Internal-FFI.html#t:TensorData">TensorData</a>]</li><li class="src short"><span class="keyword">data</span> <a href="#t:TensorData">TensorData</a> = <a href="#v:TensorData">TensorData</a> {<ul class="subs"><li><a href="#v:tensorDataDimensions">tensorDataDimensions</a> :: [<a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]</li><li><a href="#v:tensorDataType">tensorDataType</a> :: !<a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></li><li><a href="#v:tensorDataBytes">tensorDataBytes</a> :: !(<a href="../vector-0.11.0.0/Data-Vector-Storable.html#t:Vector">Vector</a> <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>)</li></ul>}</li><li class="src short"><a href="#v:setSessionConfig">setSessionConfig</a> :: <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> -&gt; SessionOptions -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> ()</li><li class="src short"><a href="#v:setSessionTarget">setSessionTarget</a> :: <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; SessionOptions -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> ()</li><li class="src short"><a href="#v:getAllOpList">getAllOpList</a> :: <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:useProtoAsVoidPtrLen">useProtoAsVoidPtrLen</a> :: (Message msg, <a href="../base-4.8.2.0/Prelude.html#t:Num">Num</a> c) =&gt; msg -&gt; (<a href="../base-4.8.2.0/Foreign-Ptr.html#t:Ptr">Ptr</a> b -&gt; c -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> a) -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> a</li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:TensorFlowException" class="def">TensorFlowException</a> <a href="src/TensorFlow-Internal-FFI.html#TensorFlowException" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:TensorFlowException" class="def">TensorFlowException</a> Code <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="subs instances"><p id="control.i:TensorFlowException" class="caption collapser" onclick="toggleSection('i:TensorFlowException')">Instances</p><div id="section.i:TensorFlowException" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="TensorFlow-Internal-FFI.html#t:TensorFlowException">TensorFlowException</a></span> <a href="src/TensorFlow-Internal-FFI.html#line-60" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="TensorFlow-Internal-FFI.html#t:TensorFlowException">TensorFlowException</a></span> <a href="src/TensorFlow-Internal-FFI.html#line-60" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Control-Exception-Base.html#t:Exception">Exception</a> <a href="TensorFlow-Internal-FFI.html#t:TensorFlowException">TensorFlowException</a></span> <a href="src/TensorFlow-Internal-FFI.html#line-62" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:Session" class="def">Session</a> <a href="src/TensorFlow-Internal-Raw.html#Session" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:withSession" class="def">withSession</a> <a href="src/TensorFlow-Internal-FFI.html#withSession" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (SessionOptions -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> ())</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">-&gt; ((<a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> () -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> ()) -&gt; <a href="TensorFlow-Internal-FFI.html#t:Session">Session</a> -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> a)</td><td class="doc"><p>The action can spawn concurrent tasks which will
+ be canceled before withSession returns.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> a</td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Runs the given action after creating a session with options
+ populated by the given optionSetter.</p></div></div><div class="top"><p class="src"><a name="v:extendGraph" class="def">extendGraph</a> :: <a href="TensorFlow-Internal-FFI.html#t:Session">Session</a> -&gt; <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a> -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> () <a href="src/TensorFlow-Internal-FFI.html#extendGraph" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:run" class="def">run</a> <a href="src/TensorFlow-Internal-FFI.html#run" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="TensorFlow-Internal-FFI.html#t:Session">Session</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">-&gt; [(<a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>, <a href="TensorFlow-Internal-FFI.html#t:TensorData">TensorData</a>)]</td><td class="doc"><p>Feeds.</p></td></tr><tr><td class="src">-&gt; [<a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>]</td><td class="doc"><p>Fetches.</p></td></tr><tr><td class="src">-&gt; [<a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>]</td><td class="doc"><p>Targets.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> [<a href="TensorFlow-Internal-FFI.html#t:TensorData">TensorData</a>]</td><td class="doc empty">&nbsp;</td></tr></table></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:TensorData" class="def">TensorData</a> <a href="src/TensorFlow-Internal-FFI.html#TensorData" class="link">Source</a></p><div class="doc"><p>All of the data needed to represent a tensor.</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:TensorData" class="def">TensorData</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:tensorDataDimensions" class="def">tensorDataDimensions</a> :: [<a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:tensorDataType" class="def">tensorDataType</a> :: !<a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:tensorDataBytes" class="def">tensorDataBytes</a> :: !(<a href="../vector-0.11.0.0/Data-Vector-Storable.html#t:Vector">Vector</a> <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>)</dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:TensorData" class="caption collapser" onclick="toggleSection('i:TensorData')">Instances</p><div id="section.i:TensorData" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="TensorFlow-Internal-FFI.html#t:TensorData">TensorData</a></span> <a href="src/TensorFlow-Internal-FFI.html#line-70" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="TensorFlow-Internal-FFI.html#t:TensorData">TensorData</a></span> <a href="src/TensorFlow-Internal-FFI.html#line-70" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><a name="v:setSessionConfig" class="def">setSessionConfig</a> :: <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> -&gt; SessionOptions -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> () <a href="src/TensorFlow-Internal-FFI.html#setSessionConfig" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:setSessionTarget" class="def">setSessionTarget</a> :: <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; SessionOptions -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> () <a href="src/TensorFlow-Internal-FFI.html#setSessionTarget" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:getAllOpList" class="def">getAllOpList</a> :: <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> <a href="src/TensorFlow-Internal-FFI.html#getAllOpList" class="link">Source</a></p><div class="doc"><p>Returns the serialized OpList of all OpDefs defined in this
+ address space.</p></div></div><h1 id="g:1">Internal helper.</h1><div class="top"><p class="src"><a name="v:useProtoAsVoidPtrLen" class="def">useProtoAsVoidPtrLen</a> :: (Message msg, <a href="../base-4.8.2.0/Prelude.html#t:Num">Num</a> c) =&gt; msg -&gt; (<a href="../base-4.8.2.0/Foreign-Ptr.html#t:Ptr">Ptr</a> b -&gt; c -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> a) -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> a <a href="src/TensorFlow-Internal-FFI.html#useProtoAsVoidPtrLen" class="link">Source</a></p><div class="doc"><p>Serializes the given msg and provides it as (ptr,len) argument
+ to the given action.</p></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Internal-VarInt.html b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Internal-VarInt.html
new file mode 100644
index 0000000..14c3dcc
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Internal-VarInt.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Internal.VarInt</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-Internal-VarInt.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-Internal-VarInt.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-0.1.0.0: TensorFlow bindings.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>Safe</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.Internal.VarInt</p></div><div id="description"><p class="caption">Description</p><div class="doc"><p>Originally taken from internal proto-lens code.</p></div></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><a href="#v:getVarInt">getVarInt</a> :: <a href="../attoparsec-0.13.0.2/Data-Attoparsec-ByteString.html#t:Parser">Parser</a> <a href="../base-4.8.2.0/Data-Word.html#t:Word64">Word64</a></li><li class="src short"><a href="#v:putVarInt">putVarInt</a> :: <a href="../base-4.8.2.0/Data-Word.html#t:Word64">Word64</a> -&gt; <a href="../bytestring-0.10.6.0/Data-ByteString-Builder.html#t:Builder">Builder</a></li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><a name="v:getVarInt" class="def">getVarInt</a> :: <a href="../attoparsec-0.13.0.2/Data-Attoparsec-ByteString.html#t:Parser">Parser</a> <a href="../base-4.8.2.0/Data-Word.html#t:Word64">Word64</a> <a href="src/TensorFlow-Internal-VarInt.html#getVarInt" class="link">Source</a></p><div class="doc"><p>Decode an unsigned varint.</p></div></div><div class="top"><p class="src"><a name="v:putVarInt" class="def">putVarInt</a> :: <a href="../base-4.8.2.0/Data-Word.html#t:Word64">Word64</a> -&gt; <a href="../bytestring-0.10.6.0/Data-ByteString-Builder.html#t:Builder">Builder</a> <a href="src/TensorFlow-Internal-VarInt.html#putVarInt" class="link">Source</a></p><div class="doc"><p>Encode a Word64.</p></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Nodes.html b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Nodes.html
new file mode 100644
index 0000000..5971080
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Nodes.html
@@ -0,0 +1,6 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Nodes</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-Nodes.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-Nodes.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-0.1.0.0: TensorFlow bindings.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.Nodes</p></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><span class="keyword">class</span> <a href="#t:Nodes">Nodes</a> t <span class="keyword">where</span><ul class="subs"><li><a href="#v:getNodes">getNodes</a> :: t -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> (<a href="../containers-0.5.6.2/Data-Set.html#t:Set">Set</a> <a href="TensorFlow-Output.html#t:NodeName">NodeName</a>)</li></ul></li><li class="src short"><span class="keyword">class</span> <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t =&gt; <a href="#t:Fetchable">Fetchable</a> t a <span class="keyword">where</span><ul class="subs"><li><a href="#v:getFetch">getFetch</a> :: t -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> (<a href="TensorFlow-Nodes.html#t:Fetch">Fetch</a> a)</li></ul></li><li class="src short"><span class="keyword">data</span> <a href="#t:Fetch">Fetch</a> a = <a href="#v:Fetch">Fetch</a> {<ul class="subs"><li><a href="#v:fetches">fetches</a> :: <a href="../containers-0.5.6.2/Data-Set.html#t:Set">Set</a> <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></li><li><a href="#v:fetchRestore">fetchRestore</a> :: <a href="../containers-0.5.6.2/Data-Map-Lazy.html#t:Map">Map</a> <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> <a href="TensorFlow-Internal-FFI.html#t:TensorData">TensorData</a> -&gt; a</li></ul>}</li><li class="src short"><a href="#v:nodesUnion">nodesUnion</a> :: (<a href="../base-4.8.2.0/Data-Monoid.html#t:Monoid">Monoid</a> b, <a href="../base-4.8.2.0/Data-Traversable.html#t:Traversable">Traversable</a> t, <a href="../base-4.8.2.0/Control-Applicative.html#t:Applicative">Applicative</a> f) =&gt; t (f b) -&gt; f b</li><li class="src short"><a href="#v:fetchTensorList">fetchTensorList</a> :: <a href="TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> (<a href="TensorFlow-Nodes.html#t:Fetch">Fetch</a> (<a href="TensorFlow-Types.html#t:Shape">Shape</a>, [a]))</li><li class="src short"><a href="#v:fetchTensorVector">fetchTensorVector</a> :: <span class="keyword">forall</span> a v. <a href="TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> (<a href="TensorFlow-Nodes.html#t:Fetch">Fetch</a> (<a href="TensorFlow-Types.html#t:Shape">Shape</a>, <a href="../vector-0.11.0.0/Data-Vector.html#t:Vector">Vector</a> a))</li><li class="src short"><span class="keyword">newtype</span> <a href="#t:Scalar">Scalar</a> a = <a href="#v:Scalar">Scalar</a> {<ul class="subs"><li><a href="#v:unScalar">unScalar</a> :: a</li></ul>}</li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">class</span> <a name="t:Nodes" class="def">Nodes</a> t <span class="keyword">where</span> <a href="src/TensorFlow-Nodes.html#Nodes" class="link">Source</a></p><div class="doc"><p>Types that contain ops which can be run.</p></div><div class="subs methods"><p class="caption">Methods</p><p class="src"><a name="v:getNodes" class="def">getNodes</a> :: t -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> (<a href="../containers-0.5.6.2/Data-Set.html#t:Set">Set</a> <a href="TensorFlow-Output.html#t:NodeName">NodeName</a>) <a href="src/TensorFlow-Nodes.html#getNodes" class="link">Source</a></p></div><div class="subs instances"><p id="control.i:Nodes" class="caption collapser" onclick="toggleSection('i:Nodes')">Instances</p><div id="section.i:Nodes" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> <a href="TensorFlow-Output.html#t:ControlNode">ControlNode</a></span> <a href="src/TensorFlow-Nodes.html#line-91" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t =&gt; <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> [t]</span> <a href="src/TensorFlow-Nodes.html#line-85" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t1, <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t2) =&gt; <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> (t1, t2)</span> <a href="src/TensorFlow-Nodes.html#line-71" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a)</span> <a href="src/TensorFlow-Nodes.html#line-101" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t1, <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t2, <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t3) =&gt; <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> (t1, t2, t3)</span> <a href="src/TensorFlow-Nodes.html#line-74" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">class</span> <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t =&gt; <a name="t:Fetchable" class="def">Fetchable</a> t a <span class="keyword">where</span> <a href="src/TensorFlow-Nodes.html#Fetchable" class="link">Source</a></p><div class="doc"><p>Types that tensor representations (e.g. <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code>, <code><a href="TensorFlow-Output.html#t:ControlNode">ControlNode</a></code>) can be
+ fetched into.</p><p>Includes collections of tensors (e.g. tuples).</p></div><div class="subs methods"><p class="caption">Methods</p><p class="src"><a name="v:getFetch" class="def">getFetch</a> :: t -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> (<a href="TensorFlow-Nodes.html#t:Fetch">Fetch</a> a) <a href="src/TensorFlow-Nodes.html#getFetch" class="link">Source</a></p></div><div class="subs instances"><p id="control.i:Fetchable" class="caption collapser" onclick="toggleSection('i:Fetchable')">Instances</p><div id="section.i:Fetchable" class="show"><table><tr><td class="src clearfix"><span class="inst-left">(~) * a () =&gt; <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> <a href="TensorFlow-Output.html#t:ControlNode">ControlNode</a> a</span> <a href="src/TensorFlow-Nodes.html#line-98" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> t a =&gt; <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> [t] [a]</span> <a href="src/TensorFlow-Nodes.html#line-88" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-Types.html#t:TensorType">TensorType</a> a, (~) * a a') =&gt; <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a) (<a href="TensorFlow-Nodes.html#t:Scalar">Scalar</a> a')</span> <a href="src/TensorFlow-Nodes.html#line-135" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-Types.html#t:TensorType">TensorType</a> a, (~) * a a') =&gt; <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a) (<a href="../vector-0.11.0.0/Data-Vector.html#t:Vector">Vector</a> a')</span> <a href="src/TensorFlow-Nodes.html#line-128" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> t1 a1, <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> t2 a2) =&gt; <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> (t1, t2) (a1, a2)</span> <a href="src/TensorFlow-Nodes.html#line-77" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> t1 a1, <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> t2 a2, <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> t3 a3) =&gt; <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> (t1, t2, t3) (a1, a2, a3)</span> <a href="src/TensorFlow-Nodes.html#line-80" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:Fetch" class="def">Fetch</a> a <a href="src/TensorFlow-Nodes.html#Fetch" class="link">Source</a></p><div class="doc"><p>Fetch action. Keeps track of what needs to be fetched and how to decode
+ the fetched data.</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:Fetch" class="def">Fetch</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:fetches" class="def">fetches</a> :: <a href="../containers-0.5.6.2/Data-Set.html#t:Set">Set</a> <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc"><p>Nodes to fetch</p></dd><dt class="src"><a name="v:fetchRestore" class="def">fetchRestore</a> :: <a href="../containers-0.5.6.2/Data-Map-Lazy.html#t:Map">Map</a> <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> <a href="TensorFlow-Internal-FFI.html#t:TensorData">TensorData</a> -&gt; a</dt><dd class="doc"><p>Function to create an <code>a</code> from the fetched data.</p></dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:Fetch" class="caption collapser" onclick="toggleSection('i:Fetch')">Instances</p><div id="section.i:Fetch" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Functor.html#t:Functor">Functor</a> <a href="TensorFlow-Nodes.html#t:Fetch">Fetch</a></span> <a href="src/TensorFlow-Nodes.html#line-60" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Control-Applicative.html#t:Applicative">Applicative</a> <a href="TensorFlow-Nodes.html#t:Fetch">Fetch</a></span> <a href="src/TensorFlow-Nodes.html#line-63" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><a name="v:nodesUnion" class="def">nodesUnion</a> :: (<a href="../base-4.8.2.0/Data-Monoid.html#t:Monoid">Monoid</a> b, <a href="../base-4.8.2.0/Data-Traversable.html#t:Traversable">Traversable</a> t, <a href="../base-4.8.2.0/Control-Applicative.html#t:Applicative">Applicative</a> f) =&gt; t (f b) -&gt; f b <a href="src/TensorFlow-Nodes.html#nodesUnion" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:fetchTensorList" class="def">fetchTensorList</a> :: <a href="TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> (<a href="TensorFlow-Nodes.html#t:Fetch">Fetch</a> (<a href="TensorFlow-Types.html#t:Shape">Shape</a>, [a])) <a href="src/TensorFlow-Nodes.html#fetchTensorList" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:fetchTensorVector" class="def">fetchTensorVector</a> :: <span class="keyword">forall</span> a v. <a href="TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> (<a href="TensorFlow-Nodes.html#t:Fetch">Fetch</a> (<a href="TensorFlow-Types.html#t:Shape">Shape</a>, <a href="../vector-0.11.0.0/Data-Vector.html#t:Vector">Vector</a> a)) <a href="src/TensorFlow-Nodes.html#fetchTensorVector" class="link">Source</a></p></div><div class="top"><p class="src"><span class="keyword">newtype</span> <a name="t:Scalar" class="def">Scalar</a> a <a href="src/TensorFlow-Nodes.html#Scalar" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:Scalar" class="def">Scalar</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:unScalar" class="def">unScalar</a> :: a</dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:Scalar" class="caption collapser" onclick="toggleSection('i:Scalar')">Instances</p><div id="section.i:Scalar" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> a =&gt; <a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> (<a href="TensorFlow-Nodes.html#t:Scalar">Scalar</a> a)</span> <a href="src/TensorFlow-Nodes.html#line-132" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Prelude.html#t:Floating">Floating</a> a =&gt; <a href="../base-4.8.2.0/Prelude.html#t:Floating">Floating</a> (<a href="TensorFlow-Nodes.html#t:Scalar">Scalar</a> a)</span> <a href="src/TensorFlow-Nodes.html#line-132" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Prelude.html#t:Fractional">Fractional</a> a =&gt; <a href="../base-4.8.2.0/Prelude.html#t:Fractional">Fractional</a> (<a href="TensorFlow-Nodes.html#t:Scalar">Scalar</a> a)</span> <a href="src/TensorFlow-Nodes.html#line-132" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Prelude.html#t:Num">Num</a> a =&gt; <a href="../base-4.8.2.0/Prelude.html#t:Num">Num</a> (<a href="TensorFlow-Nodes.html#t:Scalar">Scalar</a> a)</span> <a href="src/TensorFlow-Nodes.html#line-132" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Ord.html#t:Ord">Ord</a> a =&gt; <a href="../base-4.8.2.0/Data-Ord.html#t:Ord">Ord</a> (<a href="TensorFlow-Nodes.html#t:Scalar">Scalar</a> a)</span> <a href="src/TensorFlow-Nodes.html#line-132" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Prelude.html#t:Real">Real</a> a =&gt; <a href="../base-4.8.2.0/Prelude.html#t:Real">Real</a> (<a href="TensorFlow-Nodes.html#t:Scalar">Scalar</a> a)</span> <a href="src/TensorFlow-Nodes.html#line-132" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Prelude.html#t:RealFloat">RealFloat</a> a =&gt; <a href="../base-4.8.2.0/Prelude.html#t:RealFloat">RealFloat</a> (<a href="TensorFlow-Nodes.html#t:Scalar">Scalar</a> a)</span> <a href="src/TensorFlow-Nodes.html#line-132" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Prelude.html#t:RealFrac">RealFrac</a> a =&gt; <a href="../base-4.8.2.0/Prelude.html#t:RealFrac">RealFrac</a> (<a href="TensorFlow-Nodes.html#t:Scalar">Scalar</a> a)</span> <a href="src/TensorFlow-Nodes.html#line-133" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> a =&gt; <a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> (<a href="TensorFlow-Nodes.html#t:Scalar">Scalar</a> a)</span> <a href="src/TensorFlow-Nodes.html#line-132" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-String.html#t:IsString">IsString</a> a =&gt; <a href="../base-4.8.2.0/Data-String.html#t:IsString">IsString</a> (<a href="TensorFlow-Nodes.html#t:Scalar">Scalar</a> a)</span> <a href="src/TensorFlow-Nodes.html#line-133" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-Types.html#t:TensorType">TensorType</a> a, (~) * a a') =&gt; <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a) (<a href="TensorFlow-Nodes.html#t:Scalar">Scalar</a> a')</span> <a href="src/TensorFlow-Nodes.html#line-135" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Output.html b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Output.html
new file mode 100644
index 0000000..4584e8a
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Output.html
@@ -0,0 +1,14 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Output</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-Output.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-Output.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-0.1.0.0: TensorFlow bindings.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.Output</p></div><div id="table-of-contents"><p class="caption">Contents</p><ul><li><a href="#g:1">Ops</a></li></ul></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><span class="keyword">newtype</span> <a href="#t:ControlNode">ControlNode</a> = <a href="#v:ControlNode">ControlNode</a> {<ul class="subs"><li><a href="#v:unControlNode">unControlNode</a> :: <a href="TensorFlow-Output.html#t:Op">Op</a></li></ul>}</li><li class="src short"><span class="keyword">newtype</span> <a href="#t:Device">Device</a> = <a href="#v:Device">Device</a> {<ul class="subs"><li><a href="#v:deviceName">deviceName</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></li></ul>}</li><li class="src short"><span class="keyword">newtype</span> <a href="#t:NodeName">NodeName</a> = <a href="#v:NodeName">NodeName</a> {<ul class="subs"><li><a href="#v:unNodeName">unNodeName</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></li></ul>}</li><li class="src short"><span class="keyword">data</span> <a href="#t:Op">Op</a><ul class="subs"><li>= <a href="#v:Rendered">Rendered</a> !<a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a></li><li>| <a href="#v:Unrendered">Unrendered</a> !<a href="TensorFlow-Output.html#t:OpDef">OpDef</a></li></ul></li><li class="src short"><a href="#v:opUnrendered">opUnrendered</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Traversal-39-">Traversal'</a> <a href="TensorFlow-Output.html#t:Op">Op</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a></li><li class="src short"><span class="keyword">data</span> <a href="#t:OpDef">OpDef</a> = <a href="#v:OpDef">OpDef</a> {<ul class="subs"><li><a href="#v:_opName">_opName</a> :: !<a href="TensorFlow-Output.html#t:PendingNodeName">PendingNodeName</a></li><li><a href="#v:_opType">_opType</a> :: !<a href="TensorFlow-Output.html#t:OpType">OpType</a></li><li><a href="#v:_opAttrs">_opAttrs</a> :: !(<a href="../containers-0.5.6.2/Data-Map-Lazy.html#t:Map">Map</a> <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a>)</li><li><a href="#v:_opInputs">_opInputs</a> :: [<a href="TensorFlow-Output.html#t:Output">Output</a>]</li><li><a href="#v:_opControlInputs">_opControlInputs</a> :: [<a href="TensorFlow-Output.html#t:NodeName">NodeName</a>]</li></ul>}</li><li class="src short"><a href="#v:opName">opName</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> <a href="TensorFlow-Output.html#t:PendingNodeName">PendingNodeName</a></li><li class="src short"><a href="#v:opType">opType</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> <a href="TensorFlow-Output.html#t:OpType">OpType</a></li><li class="src short"><a href="#v:opAttr">opAttr</a> :: <a href="TensorFlow-Types.html#t:Attribute">Attribute</a> a =&gt; <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> -&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> a</li><li class="src short"><a href="#v:opInputs">opInputs</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> [<a href="TensorFlow-Output.html#t:Output">Output</a>]</li><li class="src short"><a href="#v:opControlInputs">opControlInputs</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> [<a href="TensorFlow-Output.html#t:NodeName">NodeName</a>]</li><li class="src short"><span class="keyword">newtype</span> <a href="#t:OpType">OpType</a> = <a href="#v:OpType">OpType</a> {<ul class="subs"><li><a href="#v:unOpType">unOpType</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></li></ul>}</li><li class="src short"><span class="keyword">newtype</span> <a href="#t:OutputIx">OutputIx</a> = <a href="#v:OutputIx">OutputIx</a> {<ul class="subs"><li><a href="#v:unOutputIx">unOutputIx</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int">Int</a></li></ul>}</li><li class="src short"><span class="keyword">data</span> <a href="#t:Output">Output</a> = <a href="#v:Output">Output</a> !<a href="TensorFlow-Output.html#t:OutputIx">OutputIx</a> !<a href="TensorFlow-Output.html#t:Op">Op</a></li><li class="src short"><a href="#v:output">output</a> :: <a href="TensorFlow-Output.html#t:OutputIx">OutputIx</a> -&gt; <a href="TensorFlow-Output.html#t:Op">Op</a> -&gt; <a href="TensorFlow-Output.html#t:Output">Output</a></li><li class="src short"><a href="#v:outputIndex">outputIndex</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:Output">Output</a> <a href="TensorFlow-Output.html#t:OutputIx">OutputIx</a></li><li class="src short"><a href="#v:outputOp">outputOp</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:Output">Output</a> <a href="TensorFlow-Output.html#t:Op">Op</a></li><li class="src short"><span class="keyword">data</span> <a href="#t:PendingNodeName">PendingNodeName</a><ul class="subs"><li>= <a href="#v:ExplicitName">ExplicitName</a> !<a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></li><li>| <a href="#v:ImplicitName">ImplicitName</a></li></ul></li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">newtype</span> <a name="t:ControlNode" class="def">ControlNode</a> <a href="src/TensorFlow-Output.html#ControlNode" class="link">Source</a></p><div class="doc"><p>A type of graph node which has no outputs. These nodes are
+ valuable for causing side effects when they are run.</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:ControlNode" class="def">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:unControlNode" class="def">unControlNode</a> :: <a href="TensorFlow-Output.html#t:Op">Op</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:ControlNode" class="caption collapser" onclick="toggleSection('i:ControlNode')">Instances</p><div id="section.i:ControlNode" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> <a href="TensorFlow-Output.html#t:ControlNode">ControlNode</a></span> <a href="src/TensorFlow-BuildOp.html#line-144" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> <a href="TensorFlow-Output.html#t:ControlNode">ControlNode</a></span> <a href="src/TensorFlow-BuildOp.html#line-92" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> <a href="TensorFlow-Output.html#t:ControlNode">ControlNode</a></span> <a href="src/TensorFlow-Nodes.html#line-91" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(~) * a () =&gt; <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> <a href="TensorFlow-Output.html#t:ControlNode">ControlNode</a> a</span> <a href="src/TensorFlow-Nodes.html#line-98" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">newtype</span> <a name="t:Device" class="def">Device</a> <a href="src/TensorFlow-Output.html#Device" class="link">Source</a></p><div class="doc"><p>A device that a node can be assigned to.
+ There's a naming convention where the device names
+ are constructed from job and replica names.</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:Device" class="def">Device</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:deviceName" class="def">deviceName</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:Device" class="caption collapser" onclick="toggleSection('i:Device')">Instances</p><div id="section.i:Device" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="TensorFlow-Output.html#t:Device">Device</a></span> <a href="src/TensorFlow-Output.html#line-86" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Ord.html#t:Ord">Ord</a> <a href="TensorFlow-Output.html#t:Device">Device</a></span> <a href="src/TensorFlow-Output.html#line-86" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="TensorFlow-Output.html#t:Device">Device</a></span> <a href="src/TensorFlow-Output.html#line-88" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-String.html#t:IsString">IsString</a> <a href="TensorFlow-Output.html#t:Device">Device</a></span> <a href="src/TensorFlow-Output.html#line-86" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><h1 id="g:1">Ops</h1><div class="top"><p class="src"><span class="keyword">newtype</span> <a name="t:NodeName" class="def">NodeName</a> <a href="src/TensorFlow-Output.html#NodeName" class="link">Source</a></p><div class="doc"><p>The name of a node in the graph.  This corresponds to the proto field
+ NodeDef.name.  Includes the scope prefix (if any) and a unique identifier
+ (if the node was implicitly named).</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:NodeName" class="def">NodeName</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:unNodeName" class="def">unNodeName</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:NodeName" class="caption collapser" onclick="toggleSection('i:NodeName')">Instances</p><div id="section.i:NodeName" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="TensorFlow-Output.html#t:NodeName">NodeName</a></span> <a href="src/TensorFlow-Output.html#line-129" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Ord.html#t:Ord">Ord</a> <a href="TensorFlow-Output.html#t:NodeName">NodeName</a></span> <a href="src/TensorFlow-Output.html#line-129" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="TensorFlow-Output.html#t:NodeName">NodeName</a></span> <a href="src/TensorFlow-Output.html#line-129" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:Op" class="def">Op</a> <a href="src/TensorFlow-Output.html#Op" class="link">Source</a></p><div class="doc"><p>The representation of a node in a TensorFlow graph.</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:Rendered" class="def">Rendered</a> !<a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a></td><td class="doc"><p>Properties are fixed, including the
+ device, name, and scope.</p></td></tr><tr><td class="src"><a name="v:Unrendered" class="def">Unrendered</a> !<a href="TensorFlow-Output.html#t:OpDef">OpDef</a></td><td class="doc"><p>Properties are not fixed, and may change depending
+ on which context this op is rendered in.</p></td></tr></table></div><div class="subs instances"><p id="control.i:Op" class="caption collapser" onclick="toggleSection('i:Op')">Instances</p><div id="section.i:Op" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="TensorFlow-Output.html#t:Op">Op</a></span> <a href="src/TensorFlow-Output.html#line-97" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Ord.html#t:Ord">Ord</a> <a href="TensorFlow-Output.html#t:Op">Op</a></span> <a href="src/TensorFlow-Output.html#line-97" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="TensorFlow-Output.html#t:Op">Op</a></span> <a href="src/TensorFlow-Output.html#line-99" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><a name="v:opUnrendered" class="def">opUnrendered</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Traversal-39-">Traversal'</a> <a href="TensorFlow-Output.html#t:Op">Op</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> <a href="src/TensorFlow-Output.html#opUnrendered" class="link">Source</a></p><div class="doc"><p>Traverse on the <code><a href="TensorFlow-Output.html#v:Unrendered">Unrendered</a></code> of an <code><a href="TensorFlow-Output.html#t:Op">Op</a></code>.</p><p>Same implementation as _Left.</p></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:OpDef" class="def">OpDef</a> <a href="src/TensorFlow-Output.html#OpDef" class="link">Source</a></p><div class="doc"><p>Op definition. This corresponds somewhat to the <code><a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a></code> proto.</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:OpDef" class="def">OpDef</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_opName" class="def">_opName</a> :: !<a href="TensorFlow-Output.html#t:PendingNodeName">PendingNodeName</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_opType" class="def">_opType</a> :: !<a href="TensorFlow-Output.html#t:OpType">OpType</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_opAttrs" class="def">_opAttrs</a> :: !(<a href="../containers-0.5.6.2/Data-Map-Lazy.html#t:Map">Map</a> <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a>)</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_opInputs" class="def">_opInputs</a> :: [<a href="TensorFlow-Output.html#t:Output">Output</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_opControlInputs" class="def">_opControlInputs</a> :: [<a href="TensorFlow-Output.html#t:NodeName">NodeName</a>]</dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:OpDef" class="caption collapser" onclick="toggleSection('i:OpDef')">Instances</p><div id="section.i:OpDef" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a></span> <a href="src/TensorFlow-Output.html#line-117" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Ord.html#t:Ord">Ord</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a></span> <a href="src/TensorFlow-Output.html#line-117" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><a name="v:opName" class="def">opName</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> <a href="TensorFlow-Output.html#t:PendingNodeName">PendingNodeName</a> <a href="src/TensorFlow-Output.html#opName" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:opType" class="def">opType</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> <a href="TensorFlow-Output.html#t:OpType">OpType</a> <a href="src/TensorFlow-Output.html#opType" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:opAttr" class="def">opAttr</a> :: <a href="TensorFlow-Types.html#t:Attribute">Attribute</a> a =&gt; <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> -&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> a <a href="src/TensorFlow-Output.html#opAttr" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:opInputs" class="def">opInputs</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> [<a href="TensorFlow-Output.html#t:Output">Output</a>] <a href="src/TensorFlow-Output.html#opInputs" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:opControlInputs" class="def">opControlInputs</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:OpDef">OpDef</a> [<a href="TensorFlow-Output.html#t:NodeName">NodeName</a>] <a href="src/TensorFlow-Output.html#opControlInputs" class="link">Source</a></p></div><div class="top"><p class="src"><span class="keyword">newtype</span> <a name="t:OpType" class="def">OpType</a> <a href="src/TensorFlow-Output.html#OpType" class="link">Source</a></p><div class="doc"><p>The type of op of a node in the graph.  This corresponds to the proto field
+ NodeDef.op.</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:OpType" class="def">OpType</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:unOpType" class="def">unOpType</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:OpType" class="caption collapser" onclick="toggleSection('i:OpType')">Instances</p><div id="section.i:OpType" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="TensorFlow-Output.html#t:OpType">OpType</a></span> <a href="src/TensorFlow-Output.html#line-61" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Ord.html#t:Ord">Ord</a> <a href="TensorFlow-Output.html#t:OpType">OpType</a></span> <a href="src/TensorFlow-Output.html#line-61" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="TensorFlow-Output.html#t:OpType">OpType</a></span> <a href="src/TensorFlow-Output.html#line-61" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-String.html#t:IsString">IsString</a> <a href="TensorFlow-Output.html#t:OpType">OpType</a></span> <a href="src/TensorFlow-Output.html#line-63" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">newtype</span> <a name="t:OutputIx" class="def">OutputIx</a> <a href="src/TensorFlow-Output.html#OutputIx" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:OutputIx" class="def">OutputIx</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:unOutputIx" class="def">unOutputIx</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int">Int</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:OutputIx" class="caption collapser" onclick="toggleSection('i:OutputIx')">Instances</p><div id="section.i:OutputIx" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Prelude.html#t:Enum">Enum</a> <a href="TensorFlow-Output.html#t:OutputIx">OutputIx</a></span> <a href="src/TensorFlow-Output.html#line-80" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="TensorFlow-Output.html#t:OutputIx">OutputIx</a></span> <a href="src/TensorFlow-Output.html#line-80" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Prelude.html#t:Num">Num</a> <a href="TensorFlow-Output.html#t:OutputIx">OutputIx</a></span> <a href="src/TensorFlow-Output.html#line-80" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Ord.html#t:Ord">Ord</a> <a href="TensorFlow-Output.html#t:OutputIx">OutputIx</a></span> <a href="src/TensorFlow-Output.html#line-80" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="TensorFlow-Output.html#t:OutputIx">OutputIx</a></span> <a href="src/TensorFlow-Output.html#line-80" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:Output" class="def">Output</a> <a href="src/TensorFlow-Output.html#Output" class="link">Source</a></p><div class="doc"><p>An output of a TensorFlow node.</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:Output" class="def">Output</a> !<a href="TensorFlow-Output.html#t:OutputIx">OutputIx</a> !<a href="TensorFlow-Output.html#t:Op">Op</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="subs instances"><p id="control.i:Output" class="caption collapser" onclick="toggleSection('i:Output')">Instances</p><div id="section.i:Output" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="TensorFlow-Output.html#t:Output">Output</a></span> <a href="src/TensorFlow-Output.html#line-68" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Ord.html#t:Ord">Ord</a> <a href="TensorFlow-Output.html#t:Output">Output</a></span> <a href="src/TensorFlow-Output.html#line-68" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="TensorFlow-Output.html#t:Output">Output</a></span> <a href="src/TensorFlow-Output.html#line-68" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-String.html#t:IsString">IsString</a> <a href="TensorFlow-Output.html#t:Output">Output</a></span> <a href="src/TensorFlow-Output.html#line-150" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><a name="v:output" class="def">output</a> :: <a href="TensorFlow-Output.html#t:OutputIx">OutputIx</a> -&gt; <a href="TensorFlow-Output.html#t:Op">Op</a> -&gt; <a href="TensorFlow-Output.html#t:Output">Output</a> <a href="src/TensorFlow-Output.html#output" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:outputIndex" class="def">outputIndex</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:Output">Output</a> <a href="TensorFlow-Output.html#t:OutputIx">OutputIx</a> <a href="src/TensorFlow-Output.html#outputIndex" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:outputOp" class="def">outputOp</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-Output.html#t:Output">Output</a> <a href="TensorFlow-Output.html#t:Op">Op</a> <a href="src/TensorFlow-Output.html#outputOp" class="link">Source</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:PendingNodeName" class="def">PendingNodeName</a> <a href="src/TensorFlow-Output.html#PendingNodeName" class="link">Source</a></p><div class="doc"><p>The name specified for an unrendered Op.  If an Op has an
+ ImplicitName, it will be assigned based on the opType plus a
+ unique identifier.  Does not contain the &quot;scope&quot; prefix.</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:ExplicitName" class="def">ExplicitName</a> !<a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:ImplicitName" class="def">ImplicitName</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="subs instances"><p id="control.i:PendingNodeName" class="caption collapser" onclick="toggleSection('i:PendingNodeName')">Instances</p><div id="section.i:PendingNodeName" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="TensorFlow-Output.html#t:PendingNodeName">PendingNodeName</a></span> <a href="src/TensorFlow-Output.html#line-123" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Ord.html#t:Ord">Ord</a> <a href="TensorFlow-Output.html#t:PendingNodeName">PendingNodeName</a></span> <a href="src/TensorFlow-Output.html#line-123" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="TensorFlow-Output.html#t:PendingNodeName">PendingNodeName</a></span> <a href="src/TensorFlow-Output.html#line-123" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Session.html b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Session.html
new file mode 100644
index 0000000..5d52b86
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Session.html
@@ -0,0 +1,23 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Session</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-Session.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-Session.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-0.1.0.0: TensorFlow bindings.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.Session</p></div><div id="table-of-contents"><p class="caption">Contents</p><ul><li><a href="#g:1">Opaque value created via <code>sessionConfig</code> and <code>sessionTarget</code>.</a></li></ul></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><span class="keyword">data</span> <a href="#t:Session">Session</a> a</li><li class="src short"><span class="keyword">data</span> <a href="#t:SessionOption">SessionOption</a></li><li class="src short"><a href="#v:sessionConfig">sessionConfig</a> :: <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> -&gt; <a href="TensorFlow-Session.html#t:SessionOption">SessionOption</a></li><li class="src short"><a href="#v:sessionTarget">sessionTarget</a> :: <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="TensorFlow-Session.html#t:SessionOption">SessionOption</a></li><li class="src short"><a href="#v:runSession">runSession</a> :: <a href="TensorFlow-Session.html#t:Session">Session</a> a -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> a</li><li class="src short"><a href="#v:runSessionWithOptions">runSessionWithOptions</a> :: [<a href="TensorFlow-Session.html#t:SessionOption">SessionOption</a>] -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> a -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> a</li><li class="src short"><a href="#v:build">build</a> :: <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> a</li><li class="src short"><a href="#v:buildAnd">buildAnd</a> :: (a -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> b) -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> b</li><li class="src short"><a href="#v:buildWithSummary">buildWithSummary</a> :: <span class="keyword">forall</span> a. <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> (a, [<a href="TensorFlow-Build.html#t:SummaryTensor">SummaryTensor</a>])</li><li class="src short"><a href="#v:extend">extend</a> :: <a href="TensorFlow-Session.html#t:Session">Session</a> ()</li><li class="src short"><a href="#v:addGraphDef">addGraphDef</a> :: <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> ()</li><li class="src short"><a href="#v:run">run</a> :: <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> t a =&gt; t -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> a</li><li class="src short"><a href="#v:runWithFeeds">runWithFeeds</a> :: <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> t a =&gt; [<a href="TensorFlow-Tensor.html#t:Feed">Feed</a>] -&gt; t -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> a</li><li class="src short"><a href="#v:run_">run_</a> :: <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t =&gt; t -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> ()</li><li class="src short"><a href="#v:runWithFeeds_">runWithFeeds_</a> :: <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t =&gt; [<a href="TensorFlow-Tensor.html#t:Feed">Feed</a>] -&gt; t -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> ()</li><li class="src short"><a href="#v:asyncProdNodes">asyncProdNodes</a> :: <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t =&gt; t -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> ()</li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:Session" class="def">Session</a> a <a href="src/TensorFlow-Session.html#Session" class="link">Source</a></p><div class="subs instances"><p id="control.i:Session" class="caption collapser" onclick="toggleSection('i:Session')">Instances</p><div id="section.i:Session" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Control-Monad.html#t:Monad">Monad</a> <a href="TensorFlow-Session.html#t:Session">Session</a></span> <a href="src/TensorFlow-Session.html#line-72" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Functor.html#t:Functor">Functor</a> <a href="TensorFlow-Session.html#t:Session">Session</a></span> <a href="src/TensorFlow-Session.html#line-72" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Control-Applicative.html#t:Applicative">Applicative</a> <a href="TensorFlow-Session.html#t:Session">Session</a></span> <a href="src/TensorFlow-Session.html#line-72" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../transformers-0.4.2.0/Control-Monad-IO-Class.html#t:MonadIO">MonadIO</a> <a href="TensorFlow-Session.html#t:Session">Session</a></span> <a href="src/TensorFlow-Session.html#line-72" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><h1 id="g:1">Opaque value created via <code><a href="TensorFlow-Session.html#v:sessionConfig">sessionConfig</a></code> and <code><a href="TensorFlow-Session.html#v:sessionTarget">sessionTarget</a></code>.</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:SessionOption" class="def">SessionOption</a> <a href="src/TensorFlow-Session.html#SessionOption" class="link">Source</a></p><div class="doc"><p>Setting of an option for the session (see <code><a href="TensorFlow-Session.html#v:runSessionWithOptions">runSessionWithOptions</a></code>).</p></div></div><div class="top"><p class="src"><a name="v:sessionConfig" class="def">sessionConfig</a> :: <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> -&gt; <a href="TensorFlow-Session.html#t:SessionOption">SessionOption</a> <a href="src/TensorFlow-Session.html#sessionConfig" class="link">Source</a></p><div class="doc"><p>Uses the specified config for the created session.</p></div></div><div class="top"><p class="src"><a name="v:sessionTarget" class="def">sessionTarget</a> :: <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="TensorFlow-Session.html#t:SessionOption">SessionOption</a> <a href="src/TensorFlow-Session.html#sessionTarget" class="link">Source</a></p><div class="doc"><p>Target can be: &quot;local&quot;, ip:port, host:port.
+ The set of supported factories depends on the linked in libraries.
+ REQUIRES &quot;/<em>learning</em>brain/public:tensorflow_remote&quot; dependency for the binary.</p></div></div><div class="top"><p class="src"><a name="v:runSession" class="def">runSession</a> :: <a href="TensorFlow-Session.html#t:Session">Session</a> a -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> a <a href="src/TensorFlow-Session.html#runSession" class="link">Source</a></p><div class="doc"><p>Run <code><a href="TensorFlow-Session.html#t:Session">Session</a></code> actions in a new TensorFlow session.</p></div></div><div class="top"><p class="src"><a name="v:runSessionWithOptions" class="def">runSessionWithOptions</a> :: [<a href="TensorFlow-Session.html#t:SessionOption">SessionOption</a>] -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> a -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> a <a href="src/TensorFlow-Session.html#runSessionWithOptions" class="link">Source</a></p><div class="doc"><p>Run <code><a href="TensorFlow-Session.html#t:Session">Session</a></code> actions in a new TensorFlow session created with
+ the given option setter actions (<code><a href="TensorFlow-Session.html#v:sessionTarget">sessionTarget</a></code>, <code><a href="TensorFlow-Session.html#v:sessionConfig">sessionConfig</a></code>).</p></div></div><div class="top"><p class="src"><a name="v:build" class="def">build</a> :: <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> a <a href="src/TensorFlow-Session.html#build" class="link">Source</a></p><div class="doc"><p>Lift a <code><a href="TensorFlow-Build.html#t:Build">Build</a></code> action into a <code><a href="TensorFlow-Session.html#t:Session">Session</a></code>, including any explicit op
+ renderings.</p></div></div><div class="top"><p class="src"><a name="v:buildAnd" class="def">buildAnd</a> :: (a -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> b) -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> b <a href="src/TensorFlow-Session.html#buildAnd" class="link">Source</a></p><div class="doc"><p>Helper combinator for doing something with the result of a <code><a href="TensorFlow-Build.html#t:Build">Build</a></code> action.
+ Example usage:</p><pre>buildAnd run :: Fetchable t a =&gt; Build t -&gt; Session a</pre></div></div><div class="top"><p class="src"><a name="v:buildWithSummary" class="def">buildWithSummary</a> :: <span class="keyword">forall</span> a. <a href="TensorFlow-Build.html#t:Build">Build</a> a -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> (a, [<a href="TensorFlow-Build.html#t:SummaryTensor">SummaryTensor</a>]) <a href="src/TensorFlow-Session.html#buildWithSummary" class="link">Source</a></p><div class="doc"><p>Lift a <code><a href="TensorFlow-Build.html#t:Build">Build</a></code> action into a <code><a href="TensorFlow-Session.html#t:Session">Session</a></code>, including any explicit op
+ renderings. Returns the merged summary ops which can be used for
+ logging, see <code><a href="TensorFlow-Logging.html#v:build">build</a></code> for a convenient wrapper.</p></div></div><div class="top"><p class="src"><a name="v:extend" class="def">extend</a> :: <a href="TensorFlow-Session.html#t:Session">Session</a> () <a href="src/TensorFlow-Session.html#extend" class="link">Source</a></p><div class="doc"><p>Add all pending rendered nodes to the TensorFlow graph and runs
+ any pending initializers.</p><p>Note that run, runWithFeeds, etc. will all call this function implicitly.</p></div></div><div class="top"><p class="src"><a name="v:addGraphDef" class="def">addGraphDef</a> :: <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a> -&gt; <a href="TensorFlow-Build.html#t:Build">Build</a> () <a href="src/TensorFlow-Build.html#addGraphDef" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:run" class="def">run</a> :: <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> t a =&gt; t -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> a <a href="src/TensorFlow-Session.html#run" class="link">Source</a></p><div class="doc"><p>Run a subgraph <code>t</code>, rendering any dependent nodes that aren't already
+ rendered, and fetch the corresponding values for <code>a</code>.</p></div></div><div class="top"><p class="src"><a name="v:runWithFeeds" class="def">runWithFeeds</a> :: <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> t a =&gt; [<a href="TensorFlow-Tensor.html#t:Feed">Feed</a>] -&gt; t -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> a <a href="src/TensorFlow-Session.html#runWithFeeds" class="link">Source</a></p><div class="doc"><p>Run a subgraph <code>t</code>, rendering any dependent nodes that aren't already
+ rendered, feed the given input values, and fetch the corresponding result
+ values for <code>a</code>.</p></div></div><div class="top"><p class="src"><a name="v:run_" class="def">run_</a> :: <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t =&gt; t -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> () <a href="src/TensorFlow-Session.html#run_" class="link">Source</a></p><div class="doc"><p>Run a subgraph <code>t</code>, rendering and extending any dependent nodes that aren't
+ already rendered.  This behaves like <code><a href="TensorFlow-Session.html#v:run">run</a></code> except that it doesn't do any
+ fetches.</p></div></div><div class="top"><p class="src"><a name="v:runWithFeeds_" class="def">runWithFeeds_</a> :: <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t =&gt; [<a href="TensorFlow-Tensor.html#t:Feed">Feed</a>] -&gt; t -&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> () <a href="src/TensorFlow-Session.html#runWithFeeds_" class="link">Source</a></p><div class="doc"><p>Run a subgraph <code>t</code>, rendering any dependent nodes that aren't already
+ rendered, feed the given input values, and fetch the corresponding result
+ values for <code>a</code>.  This behaves like <code><a href="TensorFlow-Session.html#v:runWithFeeds">runWithFeeds</a></code> except that it doesn't do
+ any fetches.</p></div></div><div class="top"><p class="src"><a name="v:asyncProdNodes" class="def">asyncProdNodes</a> <a href="src/TensorFlow-Session.html#asyncProdNodes" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; t</td><td class="doc"><p>Node to evaluate concurrently.</p></td></tr><tr><td class="src">-&gt; <a href="TensorFlow-Session.html#t:Session">Session</a> ()</td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Starts a concurrent thread which evaluates the given Nodes
+ forever until runSession exits or an exception occurs. Graph
+ extension happens synchronously, but the resultant run proceeds as
+ a separate thread.</p></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Tensor.html b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Tensor.html
new file mode 100644
index 0000000..950e4c4
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Tensor.html
@@ -0,0 +1,17 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Tensor</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-Tensor.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-Tensor.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-0.1.0.0: TensorFlow bindings.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.Tensor</p></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><span class="keyword">data</span> <a href="#t:Tensor">Tensor</a> v a = <a href="#v:Tensor">Tensor</a> (<a href="TensorFlow-Tensor.html#t:TensorKind">TensorKind</a> v) <a href="TensorFlow-Output.html#t:Output">Output</a></li><li class="src short"><span class="keyword">data</span> <a href="#t:Value">Value</a></li><li class="src short"><span class="keyword">data</span> <a href="#t:Ref">Ref</a></li><li class="src short"><span class="keyword">data</span> <a href="#t:TensorKind">TensorKind</a> v <span class="keyword">where</span><ul class="subs"><li><a href="#v:ValueKind">ValueKind</a> ::  <a href="TensorFlow-Tensor.html#t:TensorKind">TensorKind</a> <a href="TensorFlow-Tensor.html#t:Value">Value</a></li><li><a href="#v:RefKind">RefKind</a> ::  <a href="TensorFlow-Tensor.html#t:TensorKind">TensorKind</a> <a href="TensorFlow-Tensor.html#t:Ref">Ref</a></li></ul></li><li class="src short"><a href="#v:tensorKind">tensorKind</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a) (<a href="TensorFlow-Tensor.html#t:TensorKind">TensorKind</a> v)</li><li class="src short"><a href="#v:tensorOutput">tensorOutput</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a) <a href="TensorFlow-Output.html#t:Output">Output</a></li><li class="src short"><a href="#v:tensorAttr">tensorAttr</a> :: <a href="TensorFlow-Types.html#t:Attribute">Attribute</a> attr =&gt; <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> -&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Traversal-39-">Traversal'</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a) attr</li><li class="src short"><a href="#v:value">value</a> :: <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Value">Value</a> a</li><li class="src short"><span class="keyword">data</span> <a href="#t:Feed">Feed</a> = <a href="#v:Feed">Feed</a> <a href="TensorFlow-Output.html#t:Output">Output</a> <a href="TensorFlow-Internal-FFI.html#t:TensorData">TensorData</a></li><li class="src short"><a href="#v:feed">feed</a> :: <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="TensorFlow-Types.html#t:TensorData">TensorData</a> a -&gt; <a href="TensorFlow-Tensor.html#t:Feed">Feed</a></li><li class="src short"><a href="#v:tensorFromName">tensorFromName</a> :: <a href="TensorFlow-Tensor.html#t:TensorKind">TensorKind</a> v -&gt; <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> -&gt; <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a</li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:Tensor" class="def">Tensor</a> v a <a href="src/TensorFlow-Tensor.html#Tensor" class="link">Source</a></p><div class="doc"><p>A named output of a TensorFlow operation.</p><p>The type parameter <code>a</code> is the type of the elements in the <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code>.  The
+ parameter <code>v</code> is either <code><a href="TensorFlow-Tensor.html#t:Value">Value</a></code> or <code><a href="TensorFlow-Tensor.html#t:Ref">Ref</a></code>, depending on whether the graph is
+ treating this op output as an immutable <code><a href="TensorFlow-Tensor.html#t:Value">Value</a></code> or a stateful <code><a href="TensorFlow-Tensor.html#t:Ref">Ref</a></code> (e.g., a
+ variable).  Note that a <code>Tensor Ref</code> can be casted into a <code>Tensor Value</code> via
+ <code><a href="TensorFlow-Tensor.html#v:value">value</a></code>.</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:Tensor" class="def">Tensor</a> (<a href="TensorFlow-Tensor.html#t:TensorKind">TensorKind</a> v) <a href="TensorFlow-Output.html#t:Output">Output</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="subs instances"><p id="control.i:Tensor" class="caption collapser" onclick="toggleSection('i:Tensor')">Instances</p><div id="section.i:Tensor" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> [<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Value">Value</a> a]</span> <a href="src/TensorFlow-BuildOp.html#line-153" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> f =&gt; <a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> ([<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a] -&gt; f)</span> <a href="src/TensorFlow-BuildOp.html#line-186" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> f =&gt; <a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; f)</span> <a href="src/TensorFlow-BuildOp.html#line-183" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Ref">Ref</a> a)</span> <a href="src/TensorFlow-BuildOp.html#line-150" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Value">Value</a> a)</span> <a href="src/TensorFlow-BuildOp.html#line-147" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Ref">Ref</a> a)</span> <a href="src/TensorFlow-BuildOp.html#line-89" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Value">Value</a> a)</span> <a href="src/TensorFlow-BuildOp.html#line-86" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Nodes.html#t:Nodes">Nodes</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a)</span> <a href="src/TensorFlow-Nodes.html#line-101" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-Types.html#t:TensorType">TensorType</a> a, (~) * a a') =&gt; <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a) (<a href="TensorFlow-Nodes.html#t:Scalar">Scalar</a> a')</span> <a href="src/TensorFlow-Nodes.html#line-135" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">(<a href="TensorFlow-Types.html#t:TensorType">TensorType</a> a, (~) * a a') =&gt; <a href="TensorFlow-Nodes.html#t:Fetchable">Fetchable</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a) (<a href="../vector-0.11.0.0/Data-Vector.html#t:Vector">Vector</a> a')</span> <a href="src/TensorFlow-Nodes.html#line-128" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:Value" class="def">Value</a> <a href="src/TensorFlow-Tensor.html#Value" class="link">Source</a></p><div class="subs instances"><p id="control.i:Value" class="caption collapser" onclick="toggleSection('i:Value')">Instances</p><div id="section.i:Value" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> [<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Value">Value</a> a]</span> <a href="src/TensorFlow-BuildOp.html#line-153" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Value">Value</a> a)</span> <a href="src/TensorFlow-BuildOp.html#line-147" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Value">Value</a> a)</span> <a href="src/TensorFlow-BuildOp.html#line-86" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:Ref" class="def">Ref</a> <a href="src/TensorFlow-Tensor.html#Ref" class="link">Source</a></p><div class="subs instances"><p id="control.i:Ref" class="caption collapser" onclick="toggleSection('i:Ref')">Instances</p><div id="section.i:Ref" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:BuildOp">BuildOp</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Ref">Ref</a> a)</span> <a href="src/TensorFlow-BuildOp.html#line-150" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-BuildOp.html#t:OpResult">OpResult</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Ref">Ref</a> a)</span> <a href="src/TensorFlow-BuildOp.html#line-89" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:TensorKind" class="def">TensorKind</a> v <span class="keyword">where</span> <a href="src/TensorFlow-Tensor.html#TensorKind" class="link">Source</a></p><div class="doc"><p>This class provides a runtime switch on whether a <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> should be
+ treated as a <code><a href="TensorFlow-Tensor.html#t:Value">Value</a></code> or as a <code><a href="TensorFlow-Tensor.html#t:Ref">Ref</a></code>.</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:ValueKind" class="def">ValueKind</a> ::  <a href="TensorFlow-Tensor.html#t:TensorKind">TensorKind</a> <a href="TensorFlow-Tensor.html#t:Value">Value</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:RefKind" class="def">RefKind</a> ::  <a href="TensorFlow-Tensor.html#t:TensorKind">TensorKind</a> <a href="TensorFlow-Tensor.html#t:Ref">Ref</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div><div class="top"><p class="src"><a name="v:tensorKind" class="def">tensorKind</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a) (<a href="TensorFlow-Tensor.html#t:TensorKind">TensorKind</a> v) <a href="src/TensorFlow-Tensor.html#tensorKind" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:tensorOutput" class="def">tensorOutput</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a) <a href="TensorFlow-Output.html#t:Output">Output</a> <a href="src/TensorFlow-Tensor.html#tensorOutput" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:tensorAttr" class="def">tensorAttr</a> :: <a href="TensorFlow-Types.html#t:Attribute">Attribute</a> attr =&gt; <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> -&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Traversal-39-">Traversal'</a> (<a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a) attr <a href="src/TensorFlow-Tensor.html#tensorAttr" class="link">Source</a></p><div class="doc"><p>Lens for the attributes of a tensor.</p><p>Only valid if the tensor has not yet been rendered. If the tensor has been
+ rendered, the traversal will be over nothing (nothing can be read or
+ written).</p></div></div><div class="top"><p class="src"><a name="v:value" class="def">value</a> :: <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="TensorFlow-Tensor.html#t:Value">Value</a> a <a href="src/TensorFlow-Tensor.html#value" class="link">Source</a></p><div class="doc"><p>Cast a 'Tensor *' into a 'Tensor Value'. Common usage is to cast a
+ Ref into Value. This behaves like a no-op.</p></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:Feed" class="def">Feed</a> <a href="src/TensorFlow-Tensor.html#Feed" class="link">Source</a></p><div class="doc"><p>A pair of a <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> and some data that should be fed into that <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code>
+ when running the graph.</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:Feed" class="def">Feed</a> <a href="TensorFlow-Output.html#t:Output">Output</a> <a href="TensorFlow-Internal-FFI.html#t:TensorData">TensorData</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div><div class="top"><p class="src"><a name="v:feed" class="def">feed</a> :: <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="TensorFlow-Types.html#t:TensorData">TensorData</a> a -&gt; <a href="TensorFlow-Tensor.html#t:Feed">Feed</a> <a href="src/TensorFlow-Tensor.html#feed" class="link">Source</a></p><div class="doc"><p>Create a <code><a href="TensorFlow-Tensor.html#t:Feed">Feed</a></code> for feeding the given data into a <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> when running
+ the graph.</p><p>Note that if a <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> is rendered, its identity may change; so feeding the
+ rendered <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> may be different than feeding the original <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code>.</p></div></div><div class="top"><p class="src"><a name="v:tensorFromName" class="def">tensorFromName</a> :: <a href="TensorFlow-Tensor.html#t:TensorKind">TensorKind</a> v -&gt; <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> -&gt; <a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a <a href="src/TensorFlow-Tensor.html#tensorFromName" class="link">Source</a></p><div class="doc"><p>Create a <code><a href="TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> for a given name.  This can be used to reference nodes
+ in a <code>GraphDef</code> that was loaded via <code>addGraphDef</code>.
+ TODO(judahjacobson): add more safety checks here.</p></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Types.html b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Types.html
new file mode 100644
index 0000000..28057b4
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/TensorFlow-Types.html
@@ -0,0 +1,13 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Types</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-Types.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-Types.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-0.1.0.0: TensorFlow bindings.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.Types</p></div><div id="table-of-contents"><p class="caption">Contents</p><ul><li><a href="#g:1">Type constraints</a><ul><li><a href="#g:2">Implementation of constraints</a></li></ul></li></ul></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><span class="keyword">class</span> <a href="#t:TensorType">TensorType</a> a <span class="keyword">where</span><ul class="subs"><li><a href="#v:tensorType">tensorType</a> :: a -&gt; <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></li><li><a href="#v:tensorRefType">tensorRefType</a> :: a -&gt; <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></li><li><a href="#v:tensorVal">tensorVal</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> [a]</li><li><a href="#v:decodeTensorData">decodeTensorData</a> :: <a href="TensorFlow-Types.html#t:TensorData">TensorData</a> a -&gt; <a href="../vector-0.11.0.0/Data-Vector.html#t:Vector">Vector</a> a</li><li><a href="#v:encodeTensorData">encodeTensorData</a> :: <a href="TensorFlow-Types.html#t:Shape">Shape</a> -&gt; <a href="../vector-0.11.0.0/Data-Vector.html#t:Vector">Vector</a> a -&gt; <a href="TensorFlow-Types.html#t:TensorData">TensorData</a> a</li></ul></li><li class="src short"><span class="keyword">newtype</span> <a href="#t:TensorData">TensorData</a> a = <a href="#v:TensorData">TensorData</a> {<ul class="subs"><li><a href="#v:unTensorData">unTensorData</a> :: <a href="TensorFlow-Internal-FFI.html#t:TensorData">TensorData</a></li></ul>}</li><li class="src short"><span class="keyword">newtype</span> <a href="#t:Shape">Shape</a> = <a href="#v:Shape">Shape</a> [<a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]</li><li class="src short"><a href="#v:protoShape">protoShape</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a> <a href="TensorFlow-Types.html#t:Shape">Shape</a></li><li class="src short"><span class="keyword">class</span> <a href="#t:Attribute">Attribute</a> a <span class="keyword">where</span><ul class="subs"><li><a href="#v:attrLens">attrLens</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> a</li></ul></li><li class="src short"><span class="keyword">type</span> <a href="#t:OneOf">OneOf</a> ts a = (<a href="TensorFlow-Types.html#t:TensorType">TensorType</a> a, <a href="TensorFlow-Types.html#t:TensorTypes">TensorTypes</a> ts, <a href="TensorFlow-Types.html#t:NoneOf">NoneOf</a> (<a href="TensorFlow-Types.html#t:AllTensorTypes">AllTensorTypes</a> <a href="TensorFlow-Types.html#t:-92--92-">\\</a> ts) a)</li><li class="src short"><span class="keyword">type family</span> a <a href="#t:-47--61-">/=</a> b :: <a href="../base-4.8.2.0/GHC-Exts.html#t:Constraint">Constraint</a></li><li class="src short"><span class="keyword">data</span> <a href="#t:TypeError">TypeError</a> a</li><li class="src short"><span class="keyword">data</span> <a href="#t:ExcludedCase">ExcludedCase</a></li><li class="src short"><span class="keyword">type family</span> <a href="#t:TensorTypes">TensorTypes</a> ts :: <a href="../base-4.8.2.0/GHC-Exts.html#t:Constraint">Constraint</a></li><li class="src short"><span class="keyword">type family</span> <a href="#t:NoneOf">NoneOf</a> ts a :: <a href="../base-4.8.2.0/GHC-Exts.html#t:Constraint">Constraint</a></li><li class="src short"><span class="keyword">type family</span> as <a href="#t:-92--92-">\\</a> bs</li><li class="src short"><span class="keyword">type family</span> <a href="#t:Delete">Delete</a> a as</li><li class="src short"><span class="keyword">type</span> <a href="#t:AllTensorTypes">AllTensorTypes</a> = `[<a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>, <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a>]`</li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">class</span> <a name="t:TensorType" class="def">TensorType</a> a <span class="keyword">where</span> <a href="src/TensorFlow-Types.html#TensorType" class="link">Source</a></p><div class="doc"><p>The class of scalar types supported by tensorflow.</p></div><div class="subs methods"><p class="caption">Methods</p><p class="src"><a name="v:tensorType" class="def">tensorType</a> :: a -&gt; <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a> <a href="src/TensorFlow-Types.html#tensorType" class="link">Source</a></p><p class="src"><a name="v:tensorRefType" class="def">tensorRefType</a> :: a -&gt; <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a> <a href="src/TensorFlow-Types.html#tensorRefType" class="link">Source</a></p><p class="src"><a name="v:tensorVal" class="def">tensorVal</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> [a] <a href="src/TensorFlow-Types.html#tensorVal" class="link">Source</a></p><p class="src"><a name="v:decodeTensorData" class="def">decodeTensorData</a> :: <a href="TensorFlow-Types.html#t:TensorData">TensorData</a> a -&gt; <a href="../vector-0.11.0.0/Data-Vector.html#t:Vector">Vector</a> a <a href="src/TensorFlow-Types.html#decodeTensorData" class="link">Source</a></p><div class="doc"><p>Decode the bytes of a TensorData into a Vector.</p></div><p class="src"><a name="v:encodeTensorData" class="def">encodeTensorData</a> :: <a href="TensorFlow-Types.html#t:Shape">Shape</a> -&gt; <a href="../vector-0.11.0.0/Data-Vector.html#t:Vector">Vector</a> a -&gt; <a href="TensorFlow-Types.html#t:TensorData">TensorData</a> a <a href="src/TensorFlow-Types.html#encodeTensorData" class="link">Source</a></p><div class="doc"><p>Encode a Vector into a TensorData.</p><p>The values should be in row major order, e.g.,</p><p>element 0:   index (0, ..., 0)
+   element 1:   index (0, ..., 1)
+   ...</p></div></div><div class="subs instances"><p id="control.i:TensorType" class="caption collapser" onclick="toggleSection('i:TensorType')">Instances</p><div id="section.i:TensorType" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:TensorType">TensorType</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/TensorFlow-Types.html#line-244" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:TensorType">TensorType</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a></span> <a href="src/TensorFlow-Types.html#line-137" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:TensorType">TensorType</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></span> <a href="src/TensorFlow-Types.html#line-130" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:TensorType">TensorType</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a></span> <a href="src/TensorFlow-Types.html#line-182" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:TensorType">TensorType</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a></span> <a href="src/TensorFlow-Types.html#line-175" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:TensorType">TensorType</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></span> <a href="src/TensorFlow-Types.html#line-144" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:TensorType">TensorType</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></span> <a href="src/TensorFlow-Types.html#line-151" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:TensorType">TensorType</a> <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a></span> <a href="src/TensorFlow-Types.html#line-161" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:TensorType">TensorType</a> <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a></span> <a href="src/TensorFlow-Types.html#line-168" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:TensorType">TensorType</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></span> <a href="src/TensorFlow-Types.html#line-189" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:TensorType">TensorType</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>)</span> <a href="src/TensorFlow-Types.html#line-258" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:TensorType">TensorType</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</span> <a href="src/TensorFlow-Types.html#line-251" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">newtype</span> <a name="t:TensorData" class="def">TensorData</a> a <a href="src/TensorFlow-Types.html#TensorData" class="link">Source</a></p><div class="doc"><p>Data about a tensor that is encoded for the TensorFlow APIs.</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:TensorData" class="def">TensorData</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:unTensorData" class="def">unTensorData</a> :: <a href="TensorFlow-Internal-FFI.html#t:TensorData">TensorData</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div></div><div class="top"><p class="src"><span class="keyword">newtype</span> <a name="t:Shape" class="def">Shape</a> <a href="src/TensorFlow-Types.html#Shape" class="link">Source</a></p><div class="doc"><p>Shape (dimensions) of a tensor.</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:Shape" class="def">Shape</a> [<a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]</td><td class="doc empty">&nbsp;</td></tr></table></div><div class="subs instances"><p id="control.i:Shape" class="caption collapser" onclick="toggleSection('i:Shape')">Instances</p><div id="section.i:Shape" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/GHC-Exts.html#t:IsList">IsList</a> <a href="TensorFlow-Types.html#t:Shape">Shape</a></span> <a href="src/TensorFlow-Types.html#line-268" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="TensorFlow-Types.html#t:Shape">Shape</a></span> <a href="src/TensorFlow-Types.html#line-266" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:Attribute">Attribute</a> <a href="TensorFlow-Types.html#t:Shape">Shape</a></span> <a href="src/TensorFlow-Types.html#line-301" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> <a href="../base-4.8.2.0/GHC-Exts.html#t:Item">Item</a> <a href="TensorFlow-Types.html#t:Shape">Shape</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></span> <a href="src/TensorFlow-Types.html#line-269" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><a name="v:protoShape" class="def">protoShape</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a> <a href="TensorFlow-Types.html#t:Shape">Shape</a> <a href="src/TensorFlow-Types.html#protoShape" class="link">Source</a></p></div><div class="top"><p class="src"><span class="keyword">class</span> <a name="t:Attribute" class="def">Attribute</a> a <span class="keyword">where</span> <a href="src/TensorFlow-Types.html#Attribute" class="link">Source</a></p><div class="subs methods"><p class="caption">Methods</p><p class="src"><a name="v:attrLens" class="def">attrLens</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> a <a href="src/TensorFlow-Types.html#attrLens" class="link">Source</a></p></div><div class="subs instances"><p id="control.i:Attribute" class="caption collapser" onclick="toggleSection('i:Attribute')">Instances</p><div id="section.i:Attribute" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:Attribute">Attribute</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/TensorFlow-Types.html#line-298" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:Attribute">Attribute</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></span> <a href="src/TensorFlow-Types.html#line-283" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:Attribute">Attribute</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></span> <a href="src/TensorFlow-Types.html#line-289" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:Attribute">Attribute</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></span> <a href="src/TensorFlow-Types.html#line-286" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:Attribute">Attribute</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a></span> <a href="src/TensorFlow-Types.html#line-305" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:Attribute">Attribute</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/TensorFlow-Types.html#line-295" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:Attribute">Attribute</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></span> <a href="src/TensorFlow-Types.html#line-292" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:Attribute">Attribute</a> <a href="TensorFlow-Types.html#t:Shape">Shape</a></span> <a href="src/TensorFlow-Types.html#line-301" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:Attribute">Attribute</a> [<a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]</span> <a href="src/TensorFlow-Types.html#line-311" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="TensorFlow-Types.html#t:Attribute">Attribute</a> [<a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a>]</span> <a href="src/TensorFlow-Types.html#line-308" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><h1 id="g:1">Type constraints</h1><div class="top"><p class="src"><span class="keyword">type</span> <a name="t:OneOf" class="def">OneOf</a> ts a = (<a href="TensorFlow-Types.html#t:TensorType">TensorType</a> a, <a href="TensorFlow-Types.html#t:TensorTypes">TensorTypes</a> ts, <a href="TensorFlow-Types.html#t:NoneOf">NoneOf</a> (<a href="TensorFlow-Types.html#t:AllTensorTypes">AllTensorTypes</a> <a href="TensorFlow-Types.html#t:-92--92-">\\</a> ts) a) <a href="src/TensorFlow-Types.html#OneOf" class="link">Source</a></p><div class="doc"><p>A <code><a href="../base-4.8.2.0/GHC-Exts.html#t:Constraint">Constraint</a></code> specifying the possible choices of a <code><a href="TensorFlow-Types.html#t:TensorType">TensorType</a></code>.</p><p>We implement a <code><a href="../base-4.8.2.0/GHC-Exts.html#t:Constraint">Constraint</a></code> like <code>OneOf '[Double, Float] a</code> by turning the
+ natural representation as a conjunction, i.e.,</p><pre>   a == Double || a == Float
+</pre><p>into a disjunction like</p><pre>    a /= Int32 &amp;&amp; a /= Int64 &amp;&amp; a /= ByteString &amp;&amp; ...
+</pre><p>using an enumeration of all the possible <code><a href="TensorFlow-Types.html#t:TensorType">TensorType</a></code>s.</p></div></div><div class="top"><p class="src"><span class="keyword">type family</span> a <a name="t:-47--61-" class="def">/=</a> b :: <a href="../base-4.8.2.0/GHC-Exts.html#t:Constraint">Constraint</a> <a href="src/TensorFlow-Types.html#%2F%3D" class="link">Source</a></p><div class="doc"><p>A constraint checking that two types are different.</p></div><div class="subs equations"><p class="caption">Equations</p><table><tr><td class="src">a <a href="TensorFlow-Types.html#t:-47--61-">/=</a> a = <a href="TensorFlow-Types.html#t:TypeError">TypeError</a> a ~ <a href="TensorFlow-Types.html#t:ExcludedCase">ExcludedCase</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">a <a href="TensorFlow-Types.html#t:-47--61-">/=</a> b = ()</td><td class="doc empty">&nbsp;</td></tr></table></div></div><h2 id="g:2">Implementation of constraints</h2><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:TypeError" class="def">TypeError</a> a <a href="src/TensorFlow-Types.html#TypeError" class="link">Source</a></p><div class="doc"><p>Helper types to produce a reasonable type error message when the Constraint
+ &quot;a /= a&quot; fails.
+ TODO(judahjacobson): Use ghc-8's CustomTypeErrors for this.</p></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:ExcludedCase" class="def">ExcludedCase</a> <a href="src/TensorFlow-Types.html#ExcludedCase" class="link">Source</a></p></div><div class="top"><p class="src"><span class="keyword">type family</span> <a name="t:TensorTypes" class="def">TensorTypes</a> ts :: <a href="../base-4.8.2.0/GHC-Exts.html#t:Constraint">Constraint</a> <a href="src/TensorFlow-Types.html#TensorTypes" class="link">Source</a></p><div class="doc"><p>A <code><a href="../base-4.8.2.0/GHC-Exts.html#t:Constraint">Constraint</a></code> checking that the input is a list of <code><a href="TensorFlow-Types.html#t:TensorType">TensorType</a></code>s.
+ Helps improve error messages when using <code><a href="TensorFlow-Types.html#t:OneOf">OneOf</a></code>.</p></div><div class="subs equations"><p class="caption">Equations</p><table><tr><td class="src"><a href="TensorFlow-Types.html#t:TensorTypes">TensorTypes</a> `[]` = ()</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a href="TensorFlow-Types.html#t:TensorTypes">TensorTypes</a> (t : ts) = (<a href="TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="TensorFlow-Types.html#t:TensorTypes">TensorTypes</a> ts)</td><td class="doc empty">&nbsp;</td></tr></table></div></div><div class="top"><p class="src"><span class="keyword">type family</span> <a name="t:NoneOf" class="def">NoneOf</a> ts a :: <a href="../base-4.8.2.0/GHC-Exts.html#t:Constraint">Constraint</a> <a href="src/TensorFlow-Types.html#NoneOf" class="link">Source</a></p><div class="doc"><p>A constraint that the type <code>a</code> doesn't appear in the type list <code>ts</code>.
+ Assumes that <code>a</code> and each of the elements of <code>ts</code> are <code><a href="TensorFlow-Types.html#t:TensorType">TensorType</a></code>s.</p></div><div class="subs equations"><p class="caption">Equations</p><table><tr><td class="src"><a href="TensorFlow-Types.html#t:NoneOf">NoneOf</a> `[]` a = ()</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a href="TensorFlow-Types.html#t:NoneOf">NoneOf</a> (t : ts) a = (a <a href="TensorFlow-Types.html#t:-47--61-">/=</a> t, <a href="TensorFlow-Types.html#t:NoneOf">NoneOf</a> ts a)</td><td class="doc empty">&nbsp;</td></tr></table></div></div><div class="top"><p class="src"><span class="keyword">type family</span> as <a name="t:-92--92-" class="def">\\</a> bs <a href="src/TensorFlow-Types.html#%5C%5C" class="link">Source</a></p><div class="doc"><p>Takes the difference of two lists of types.</p></div><div class="subs equations"><p class="caption">Equations</p><table><tr><td class="src">as <a href="TensorFlow-Types.html#t:-92--92-">\\</a> `[]` = as</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">as <a href="TensorFlow-Types.html#t:-92--92-">\\</a> (b : bs) = <a href="TensorFlow-Types.html#t:Delete">Delete</a> b as <a href="TensorFlow-Types.html#t:-92--92-">\\</a> bs</td><td class="doc empty">&nbsp;</td></tr></table></div></div><div class="top"><p class="src"><span class="keyword">type family</span> <a name="t:Delete" class="def">Delete</a> a as <a href="src/TensorFlow-Types.html#Delete" class="link">Source</a></p><div class="doc"><p>Removes a type from the given list of types.</p></div><div class="subs equations"><p class="caption">Equations</p><table><tr><td class="src"><a href="TensorFlow-Types.html#t:Delete">Delete</a> a `[]` = `[]`</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a href="TensorFlow-Types.html#t:Delete">Delete</a> a (a : as) = <a href="TensorFlow-Types.html#t:Delete">Delete</a> a as</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a href="TensorFlow-Types.html#t:Delete">Delete</a> a (b : as) = b : <a href="TensorFlow-Types.html#t:Delete">Delete</a> a as</td><td class="doc empty">&nbsp;</td></tr></table></div></div><div class="top"><p class="src"><span class="keyword">type</span> <a name="t:AllTensorTypes" class="def">AllTensorTypes</a> = `[<a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>, <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a>]` <a href="src/TensorFlow-Types.html#AllTensorTypes" class="link">Source</a></p><div class="doc"><p>An enumeration of all valid <code><a href="TensorFlow-Types.html#t:TensorType">TensorType</a></code>s.</p></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/doc-index.html b/docs/haddock/tensorflow-0.1.0.0/doc-index.html
new file mode 100644
index 0000000..dc781ec
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/doc-index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-0.1.0.0: TensorFlow bindings. (Index)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-0.1.0.0: TensorFlow bindings.</p></div><div id="content"><div id="index"><p class="caption">Index</p><table><tr><td class="src">/=</td><td class="module"><a href="TensorFlow-Types.html#t:-47--61-">TensorFlow.Types</a></td></tr><tr><td class="src">addGraphDef</td><td class="module"><a href="TensorFlow-Build.html#v:addGraphDef">TensorFlow.Build</a>, <a href="TensorFlow-Session.html#v:addGraphDef">TensorFlow.Session</a></td></tr><tr><td class="src">addInitializer</td><td class="module"><a href="TensorFlow-Build.html#v:addInitializer">TensorFlow.Build</a></td></tr><tr><td class="src">addNewOp</td><td class="module"><a href="TensorFlow-Build.html#v:addNewOp">TensorFlow.Build</a></td></tr><tr><td class="src">addSummary</td><td class="module"><a href="TensorFlow-Build.html#v:addSummary">TensorFlow.Build</a></td></tr><tr><td class="src">AllTensorTypes</td><td class="module"><a href="TensorFlow-Types.html#t:AllTensorTypes">TensorFlow.Types</a></td></tr><tr><td class="src">asGraphDef</td><td class="module"><a href="TensorFlow-Build.html#v:asGraphDef">TensorFlow.Build</a></td></tr><tr><td class="src">asyncProdNodes</td><td class="module"><a href="TensorFlow-Session.html#v:asyncProdNodes">TensorFlow.Session</a></td></tr><tr><td class="src">Attribute</td><td class="module"><a href="TensorFlow-Types.html#t:Attribute">TensorFlow.Types</a></td></tr><tr><td class="src">attrLens</td><td class="module"><a href="TensorFlow-Types.html#v:attrLens">TensorFlow.Types</a></td></tr><tr><td class="src">Build</td><td class="module"><a href="TensorFlow-Build.html#t:Build">TensorFlow.Build</a></td></tr><tr><td class="src">build</td><td class="module"><a href="TensorFlow-Session.html#v:build">TensorFlow.Session</a></td></tr><tr><td class="src">buildAnd</td><td class="module"><a href="TensorFlow-Session.html#v:buildAnd">TensorFlow.Session</a></td></tr><tr><td class="src">buildListOp</td><td class="module"><a href="TensorFlow-BuildOp.html#v:buildListOp">TensorFlow.BuildOp</a></td></tr><tr><td class="src">BuildOp</td><td class="module"><a href="TensorFlow-BuildOp.html#t:BuildOp">TensorFlow.BuildOp</a></td></tr><tr><td class="src">buildOp</td><td class="module"><a href="TensorFlow-BuildOp.html#v:buildOp">TensorFlow.BuildOp</a></td></tr><tr><td class="src">BuildT</td><td class="module"><a href="TensorFlow-Build.html#t:BuildT">TensorFlow.Build</a></td></tr><tr><td class="src">buildWithSummary</td><td class="module"><a href="TensorFlow-Session.html#v:buildWithSummary">TensorFlow.Session</a></td></tr><tr><td class="src">collectAllSummaries</td><td class="module"><a href="TensorFlow-Build.html#v:collectAllSummaries">TensorFlow.Build</a></td></tr><tr><td class="src">colocateWith</td><td class="module"><a href="TensorFlow-Build.html#v:colocateWith">TensorFlow.Build</a></td></tr><tr><td class="src">ControlNode</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="TensorFlow-Output.html#t:ControlNode">TensorFlow.Output</a>, <a href="TensorFlow-Build.html#t:ControlNode">TensorFlow.Build</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="TensorFlow-Output.html#v:ControlNode">TensorFlow.Output</a>, <a href="TensorFlow-Build.html#v:ControlNode">TensorFlow.Build</a></td></tr><tr><td class="src">decodeTensorData</td><td class="module"><a href="TensorFlow-Types.html#v:decodeTensorData">TensorFlow.Types</a></td></tr><tr><td class="src">Delete</td><td class="module"><a href="TensorFlow-Types.html#t:Delete">TensorFlow.Types</a></td></tr><tr><td class="src">Device</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="TensorFlow-Output.html#t:Device">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="TensorFlow-Output.html#v:Device">TensorFlow.Output</a></td></tr><tr><td class="src">deviceName</td><td class="module"><a href="TensorFlow-Output.html#v:deviceName">TensorFlow.Output</a></td></tr><tr><td class="src">encodeTensorData</td><td class="module"><a href="TensorFlow-Types.html#v:encodeTensorData">TensorFlow.Types</a></td></tr><tr><td class="src">eqLengthGuard</td><td class="module"><a href="TensorFlow-BuildOp.html#v:eqLengthGuard">TensorFlow.BuildOp</a></td></tr><tr><td class="src">evalBuildT</td><td class="module"><a href="TensorFlow-Build.html#v:evalBuildT">TensorFlow.Build</a></td></tr><tr><td class="src">ExcludedCase</td><td class="module"><a href="TensorFlow-Types.html#t:ExcludedCase">TensorFlow.Types</a></td></tr><tr><td class="src">ExplicitName</td><td class="module"><a href="TensorFlow-Output.html#v:ExplicitName">TensorFlow.Output</a></td></tr><tr><td class="src">explicitName</td><td class="module"><a href="TensorFlow-Build.html#v:explicitName">TensorFlow.Build</a></td></tr><tr><td class="src">extend</td><td class="module"><a href="TensorFlow-Session.html#v:extend">TensorFlow.Session</a></td></tr><tr><td class="src">extendGraph</td><td class="module"><a href="TensorFlow-Internal-FFI.html#v:extendGraph">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">Feed</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="TensorFlow-Tensor.html#t:Feed">TensorFlow.Tensor</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="TensorFlow-Tensor.html#v:Feed">TensorFlow.Tensor</a></td></tr><tr><td class="src">feed</td><td class="module"><a href="TensorFlow-Tensor.html#v:feed">TensorFlow.Tensor</a></td></tr><tr><td class="src">Fetch</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="TensorFlow-Nodes.html#t:Fetch">TensorFlow.Nodes</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="TensorFlow-Nodes.html#v:Fetch">TensorFlow.Nodes</a></td></tr><tr><td class="src">Fetchable</td><td class="module"><a href="TensorFlow-Nodes.html#t:Fetchable">TensorFlow.Nodes</a></td></tr><tr><td class="src">fetches</td><td class="module"><a href="TensorFlow-Nodes.html#v:fetches">TensorFlow.Nodes</a></td></tr><tr><td class="src">fetchRestore</td><td class="module"><a href="TensorFlow-Nodes.html#v:fetchRestore">TensorFlow.Nodes</a></td></tr><tr><td class="src">fetchTensorList</td><td class="module"><a href="TensorFlow-Nodes.html#v:fetchTensorList">TensorFlow.Nodes</a></td></tr><tr><td class="src">fetchTensorVector</td><td class="module"><a href="TensorFlow-Nodes.html#v:fetchTensorVector">TensorFlow.Nodes</a></td></tr><tr><td class="src">flushInitializers</td><td class="module"><a href="TensorFlow-Build.html#v:flushInitializers">TensorFlow.Build</a></td></tr><tr><td class="src">flushNodeBuffer</td><td class="module"><a href="TensorFlow-Build.html#v:flushNodeBuffer">TensorFlow.Build</a></td></tr><tr><td class="src">getAllOpList</td><td class="module"><a href="TensorFlow-Internal-FFI.html#v:getAllOpList">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">getFetch</td><td class="module"><a href="TensorFlow-Nodes.html#v:getFetch">TensorFlow.Nodes</a></td></tr><tr><td class="src">getNodes</td><td class="module"><a href="TensorFlow-Nodes.html#v:getNodes">TensorFlow.Nodes</a></td></tr><tr><td class="src">getOrAddOp</td><td class="module"><a href="TensorFlow-Build.html#v:getOrAddOp">TensorFlow.Build</a></td></tr><tr><td class="src">getVarInt</td><td class="module"><a href="TensorFlow-Internal-VarInt.html#v:getVarInt">TensorFlow.Internal.VarInt</a></td></tr><tr><td class="src">GraphState</td><td class="module"><a href="TensorFlow-Build.html#t:GraphState">TensorFlow.Build</a></td></tr><tr><td class="src">group</td><td class="module"><a href="TensorFlow-ControlFlow.html#v:group">TensorFlow.ControlFlow</a></td></tr><tr><td class="src">hoistBuildT</td><td class="module"><a href="TensorFlow-Build.html#v:hoistBuildT">TensorFlow.Build</a></td></tr><tr><td class="src">identity</td><td class="module"><a href="TensorFlow-ControlFlow.html#v:identity">TensorFlow.ControlFlow</a></td></tr><tr><td class="src">ImplicitName</td><td class="module"><a href="TensorFlow-Output.html#v:ImplicitName">TensorFlow.Output</a></td></tr><tr><td class="src">implicitName</td><td class="module"><a href="TensorFlow-Build.html#v:implicitName">TensorFlow.Build</a></td></tr><tr><td class="src">named</td><td class="module"><a href="TensorFlow-ControlFlow.html#v:named">TensorFlow.ControlFlow</a></td></tr><tr><td class="src">NodeName</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="TensorFlow-Output.html#t:NodeName">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="TensorFlow-Output.html#v:NodeName">TensorFlow.Output</a></td></tr><tr><td class="src">Nodes</td><td class="module"><a href="TensorFlow-Nodes.html#t:Nodes">TensorFlow.Nodes</a></td></tr><tr><td class="src">nodesUnion</td><td class="module"><a href="TensorFlow-Nodes.html#v:nodesUnion">TensorFlow.Nodes</a></td></tr><tr><td class="src">NoneOf</td><td class="module"><a href="TensorFlow-Types.html#t:NoneOf">TensorFlow.Types</a></td></tr><tr><td class="src">noOp</td><td class="module"><a href="TensorFlow-ControlFlow.html#v:noOp">TensorFlow.ControlFlow</a></td></tr><tr><td class="src">OneOf</td><td class="module"><a href="TensorFlow-Types.html#t:OneOf">TensorFlow.Types</a></td></tr><tr><td class="src">Op</td><td class="module"><a href="TensorFlow-Output.html#t:Op">TensorFlow.Output</a></td></tr><tr><td class="src">opAttr</td><td class="module"><a href="TensorFlow-Output.html#v:opAttr">TensorFlow.Output</a>, <a href="TensorFlow-Build.html#v:opAttr">TensorFlow.Build</a></td></tr><tr><td class="src">opControlInputs</td><td class="module"><a href="TensorFlow-Output.html#v:opControlInputs">TensorFlow.Output</a>, <a href="TensorFlow-Build.html#v:opControlInputs">TensorFlow.Build</a></td></tr><tr><td class="src">OpDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="TensorFlow-Output.html#t:OpDef">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="TensorFlow-Output.html#v:OpDef">TensorFlow.Output</a></td></tr><tr><td class="src">opDef</td><td class="module"><a href="TensorFlow-Build.html#v:opDef">TensorFlow.Build</a></td></tr><tr><td class="src">opDefWithName</td><td class="module"><a href="TensorFlow-Build.html#v:opDefWithName">TensorFlow.Build</a></td></tr><tr><td class="src">opInputs</td><td class="module"><a href="TensorFlow-Output.html#v:opInputs">TensorFlow.Output</a>, <a href="TensorFlow-Build.html#v:opInputs">TensorFlow.Build</a></td></tr><tr><td class="src">opName</td><td class="module"><a href="TensorFlow-Output.html#v:opName">TensorFlow.Output</a>, <a href="TensorFlow-Build.html#v:opName">TensorFlow.Build</a></td></tr><tr><td class="src">OpResult</td><td class="module"><a href="TensorFlow-BuildOp.html#t:OpResult">TensorFlow.BuildOp</a></td></tr><tr><td class="src">OpType</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="TensorFlow-Output.html#t:OpType">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="TensorFlow-Output.html#v:OpType">TensorFlow.Output</a></td></tr><tr><td class="src">opType</td><td class="module"><a href="TensorFlow-Output.html#v:opType">TensorFlow.Output</a>, <a href="TensorFlow-Build.html#v:opType">TensorFlow.Build</a></td></tr><tr><td class="src">opUnrendered</td><td class="module"><a href="TensorFlow-Output.html#v:opUnrendered">TensorFlow.Output</a></td></tr><tr><td class="src">Output</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="TensorFlow-Output.html#t:Output">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="TensorFlow-Output.html#v:Output">TensorFlow.Output</a></td></tr><tr><td class="src">output</td><td class="module"><a href="TensorFlow-Output.html#v:output">TensorFlow.Output</a></td></tr><tr><td class="src">outputIndex</td><td class="module"><a href="TensorFlow-Output.html#v:outputIndex">TensorFlow.Output</a></td></tr><tr><td class="src">OutputIx</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="TensorFlow-Output.html#t:OutputIx">TensorFlow.Output</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="TensorFlow-Output.html#v:OutputIx">TensorFlow.Output</a></td></tr><tr><td class="src">outputOp</td><td class="module"><a href="TensorFlow-Output.html#v:outputOp">TensorFlow.Output</a></td></tr><tr><td class="src">PendingNodeName</td><td class="module"><a href="TensorFlow-Output.html#t:PendingNodeName">TensorFlow.Output</a></td></tr><tr><td class="src">protoShape</td><td class="module"><a href="TensorFlow-Types.html#v:protoShape">TensorFlow.Types</a></td></tr><tr><td class="src">putVarInt</td><td class="module"><a href="TensorFlow-Internal-VarInt.html#v:putVarInt">TensorFlow.Internal.VarInt</a></td></tr><tr><td class="src">Ref</td><td class="module"><a href="TensorFlow-Tensor.html#t:Ref">TensorFlow.Tensor</a></td></tr><tr><td class="src">RefKind</td><td class="module"><a href="TensorFlow-Tensor.html#v:RefKind">TensorFlow.Tensor</a></td></tr><tr><td class="src">render</td><td class="module"><a href="TensorFlow-Build.html#v:render">TensorFlow.Build</a></td></tr><tr><td class="src">Rendered</td><td class="module"><a href="TensorFlow-Output.html#v:Rendered">TensorFlow.Output</a></td></tr><tr><td class="src">renderedNodeDefs</td><td class="module"><a href="TensorFlow-Build.html#v:renderedNodeDefs">TensorFlow.Build</a></td></tr><tr><td class="src">renderNodeName</td><td class="module"><a href="TensorFlow-Build.html#v:renderNodeName">TensorFlow.Build</a></td></tr><tr><td class="src">renderOutput</td><td class="module"><a href="TensorFlow-Build.html#v:renderOutput">TensorFlow.Build</a></td></tr><tr><td class="src">run</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="TensorFlow-Internal-FFI.html#v:run">TensorFlow.Internal.FFI</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="TensorFlow-Session.html#v:run">TensorFlow.Session</a></td></tr><tr><td class="src">runBuildT</td><td class="module"><a href="TensorFlow-Build.html#v:runBuildT">TensorFlow.Build</a></td></tr><tr><td class="src">runSession</td><td class="module"><a href="TensorFlow-Session.html#v:runSession">TensorFlow.Session</a></td></tr><tr><td class="src">runSessionWithOptions</td><td class="module"><a href="TensorFlow-Session.html#v:runSessionWithOptions">TensorFlow.Session</a></td></tr><tr><td class="src">runWithFeeds</td><td class="module"><a href="TensorFlow-Session.html#v:runWithFeeds">TensorFlow.Session</a></td></tr><tr><td class="src">runWithFeeds_</td><td class="module"><a href="TensorFlow-Session.html#v:runWithFeeds_">TensorFlow.Session</a></td></tr><tr><td class="src">run_</td><td class="module"><a href="TensorFlow-Session.html#v:run_">TensorFlow.Session</a></td></tr><tr><td class="src">Scalar</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="TensorFlow-Nodes.html#t:Scalar">TensorFlow.Nodes</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="TensorFlow-Nodes.html#v:Scalar">TensorFlow.Nodes</a></td></tr><tr><td class="src">Session</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="TensorFlow-Internal-FFI.html#t:Session">TensorFlow.Internal.FFI</a></td></tr><tr><td class="alt">2 (Type/Class)</td><td class="module"><a href="TensorFlow-Session.html#t:Session">TensorFlow.Session</a></td></tr><tr><td class="src">sessionConfig</td><td class="module"><a href="TensorFlow-Session.html#v:sessionConfig">TensorFlow.Session</a></td></tr><tr><td class="src">SessionOption</td><td class="module"><a href="TensorFlow-Session.html#t:SessionOption">TensorFlow.Session</a></td></tr><tr><td class="src">sessionTarget</td><td class="module"><a href="TensorFlow-Session.html#v:sessionTarget">TensorFlow.Session</a></td></tr><tr><td class="src">setSessionConfig</td><td class="module"><a href="TensorFlow-Internal-FFI.html#v:setSessionConfig">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">setSessionTarget</td><td class="module"><a href="TensorFlow-Internal-FFI.html#v:setSessionTarget">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">Shape</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="TensorFlow-Types.html#t:Shape">TensorFlow.Types</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="TensorFlow-Types.html#v:Shape">TensorFlow.Types</a></td></tr><tr><td class="src">SummaryTensor</td><td class="module"><a href="TensorFlow-Build.html#t:SummaryTensor">TensorFlow.Build</a></td></tr><tr><td class="src">Tensor</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="TensorFlow-Tensor.html#t:Tensor">TensorFlow.Tensor</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="TensorFlow-Tensor.html#v:Tensor">TensorFlow.Tensor</a></td></tr><tr><td class="src">tensorAttr</td><td class="module"><a href="TensorFlow-Tensor.html#v:tensorAttr">TensorFlow.Tensor</a></td></tr><tr><td class="src">TensorData</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="TensorFlow-Internal-FFI.html#t:TensorData">TensorFlow.Internal.FFI</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="TensorFlow-Internal-FFI.html#v:TensorData">TensorFlow.Internal.FFI</a></td></tr><tr><td class="alt">3 (Type/Class)</td><td class="module"><a href="TensorFlow-Types.html#t:TensorData">TensorFlow.Types</a></td></tr><tr><td class="alt">4 (Data Constructor)</td><td class="module"><a href="TensorFlow-Types.html#v:TensorData">TensorFlow.Types</a></td></tr><tr><td class="src">tensorDataBytes</td><td class="module"><a href="TensorFlow-Internal-FFI.html#v:tensorDataBytes">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">tensorDataDimensions</td><td class="module"><a href="TensorFlow-Internal-FFI.html#v:tensorDataDimensions">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">tensorDataType</td><td class="module"><a href="TensorFlow-Internal-FFI.html#v:tensorDataType">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">TensorFlowException</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="TensorFlow-Internal-FFI.html#t:TensorFlowException">TensorFlow.Internal.FFI</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="TensorFlow-Internal-FFI.html#v:TensorFlowException">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">tensorFromName</td><td class="module"><a href="TensorFlow-Tensor.html#v:tensorFromName">TensorFlow.Tensor</a></td></tr><tr><td class="src">TensorKind</td><td class="module"><a href="TensorFlow-Tensor.html#t:TensorKind">TensorFlow.Tensor</a></td></tr><tr><td class="src">tensorKind</td><td class="module"><a href="TensorFlow-Tensor.html#v:tensorKind">TensorFlow.Tensor</a></td></tr><tr><td class="src">tensorOutput</td><td class="module"><a href="TensorFlow-Tensor.html#v:tensorOutput">TensorFlow.Tensor</a></td></tr><tr><td class="src">tensorRefType</td><td class="module"><a href="TensorFlow-Types.html#v:tensorRefType">TensorFlow.Types</a></td></tr><tr><td class="src">TensorType</td><td class="module"><a href="TensorFlow-Types.html#t:TensorType">TensorFlow.Types</a></td></tr><tr><td class="src">tensorType</td><td class="module"><a href="TensorFlow-Types.html#v:tensorType">TensorFlow.Types</a></td></tr><tr><td class="src">TensorTypes</td><td class="module"><a href="TensorFlow-Types.html#t:TensorTypes">TensorFlow.Types</a></td></tr><tr><td class="src">tensorVal</td><td class="module"><a href="TensorFlow-Types.html#v:tensorVal">TensorFlow.Types</a></td></tr><tr><td class="src">TypeError</td><td class="module"><a href="TensorFlow-Types.html#t:TypeError">TensorFlow.Types</a></td></tr><tr><td class="src">unControlNode</td><td class="module"><a href="TensorFlow-Output.html#v:unControlNode">TensorFlow.Output</a>, <a href="TensorFlow-Build.html#v:unControlNode">TensorFlow.Build</a></td></tr><tr><td class="src">Unique</td><td class="module"><a href="TensorFlow-Build.html#t:Unique">TensorFlow.Build</a></td></tr><tr><td class="src">unNodeName</td><td class="module"><a href="TensorFlow-Output.html#v:unNodeName">TensorFlow.Output</a></td></tr><tr><td class="src">unOpType</td><td class="module"><a href="TensorFlow-Output.html#v:unOpType">TensorFlow.Output</a></td></tr><tr><td class="src">unOutputIx</td><td class="module"><a href="TensorFlow-Output.html#v:unOutputIx">TensorFlow.Output</a></td></tr><tr><td class="src">Unrendered</td><td class="module"><a href="TensorFlow-Output.html#v:Unrendered">TensorFlow.Output</a></td></tr><tr><td class="src">unScalar</td><td class="module"><a href="TensorFlow-Nodes.html#v:unScalar">TensorFlow.Nodes</a></td></tr><tr><td class="src">unTensorData</td><td class="module"><a href="TensorFlow-Types.html#v:unTensorData">TensorFlow.Types</a></td></tr><tr><td class="src">useProtoAsVoidPtrLen</td><td class="module"><a href="TensorFlow-Internal-FFI.html#v:useProtoAsVoidPtrLen">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">Value</td><td class="module"><a href="TensorFlow-Tensor.html#t:Value">TensorFlow.Tensor</a></td></tr><tr><td class="src">value</td><td class="module"><a href="TensorFlow-Tensor.html#v:value">TensorFlow.Tensor</a></td></tr><tr><td class="src">ValueKind</td><td class="module"><a href="TensorFlow-Tensor.html#v:ValueKind">TensorFlow.Tensor</a></td></tr><tr><td class="src">withControlDependencies</td><td class="module"><a href="TensorFlow-ControlFlow.html#v:withControlDependencies">TensorFlow.ControlFlow</a></td></tr><tr><td class="src">withDevice</td><td class="module"><a href="TensorFlow-Build.html#v:withDevice">TensorFlow.Build</a></td></tr><tr><td class="src">withNameScope</td><td class="module"><a href="TensorFlow-Build.html#v:withNameScope">TensorFlow.Build</a></td></tr><tr><td class="src">withNodeDependencies</td><td class="module"><a href="TensorFlow-Build.html#v:withNodeDependencies">TensorFlow.Build</a></td></tr><tr><td class="src">withSession</td><td class="module"><a href="TensorFlow-Internal-FFI.html#v:withSession">TensorFlow.Internal.FFI</a></td></tr><tr><td class="src">withStateLens</td><td class="module"><a href="TensorFlow-Build.html#v:withStateLens">TensorFlow.Build</a></td></tr><tr><td class="src">\\</td><td class="module"><a href="TensorFlow-Types.html#t:-92--92-">TensorFlow.Types</a></td></tr><tr><td class="src">_opAttrs</td><td class="module"><a href="TensorFlow-Output.html#v:_opAttrs">TensorFlow.Output</a></td></tr><tr><td class="src">_opControlInputs</td><td class="module"><a href="TensorFlow-Output.html#v:_opControlInputs">TensorFlow.Output</a></td></tr><tr><td class="src">_opInputs</td><td class="module"><a href="TensorFlow-Output.html#v:_opInputs">TensorFlow.Output</a></td></tr><tr><td class="src">_opName</td><td class="module"><a href="TensorFlow-Output.html#v:_opName">TensorFlow.Output</a></td></tr><tr><td class="src">_opType</td><td class="module"><a href="TensorFlow-Output.html#v:_opType">TensorFlow.Output</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/frames.html b/docs/haddock/tensorflow-0.1.0.0/frames.html
new file mode 100644
index 0000000..1b4e38d
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/frames.html
@@ -0,0 +1,30 @@
+<!DOCTYPE html 
+     PUBLIC "-//W3C//DTD XHTML 1.0 Frameset//EN"
+     "http://www.w3.org/TR/xhtml1/DTD/xhtml1-frameset.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<title></title>
+<script src="haddock-util.js" type="text/javascript"></script>
+<script type="text/javascript"><!--
+/*
+
+  The synopsis frame needs to be updated using javascript, so we hide
+  it by default and only show it if javascript is enabled.
+
+  TODO: provide some means to disable it.
+*/
+function load() {
+  var d = document.getElementById("inner-fs");
+  d.rows = "50%,50%";
+  postReframe();
+}
+--></script>
+</head>
+<frameset id="outer-fs" cols="25%,75%" onload="load()">
+  <frameset id="inner-fs" rows="100%,0%">
+    <frame src="index-frames.html" name="modules" />
+    <frame src="" name="synopsis" />
+  </frameset>
+  <frame src="index.html" name="main" />
+</frameset>
+</html>
diff --git a/docs/haddock/tensorflow-0.1.0.0/haddock-util.js b/docs/haddock/tensorflow-0.1.0.0/haddock-util.js
new file mode 100644
index 0000000..9a6fccf
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/haddock-util.js
@@ -0,0 +1,344 @@
+// Haddock JavaScript utilities
+
+var rspace = /\s\s+/g,
+	  rtrim = /^\s+|\s+$/g;
+
+function spaced(s) { return (" " + s + " ").replace(rspace, " "); }
+function trim(s)   { return s.replace(rtrim, ""); }
+
+function hasClass(elem, value) {
+  var className = spaced(elem.className || "");
+  return className.indexOf( " " + value + " " ) >= 0;
+}
+
+function addClass(elem, value) {
+  var className = spaced(elem.className || "");
+  if ( className.indexOf( " " + value + " " ) < 0 ) {
+    elem.className = trim(className + " " + value);
+  }
+}
+
+function removeClass(elem, value) {
+  var className = spaced(elem.className || "");
+  className = className.replace(" " + value + " ", " ");
+  elem.className = trim(className);
+}
+
+function toggleClass(elem, valueOn, valueOff, bool) {
+  if (bool == null) { bool = ! hasClass(elem, valueOn); }
+  if (bool) {
+    removeClass(elem, valueOff);
+    addClass(elem, valueOn);
+  }
+  else {
+    removeClass(elem, valueOn);
+    addClass(elem, valueOff);
+  }
+  return bool;
+}
+
+
+function makeClassToggle(valueOn, valueOff)
+{
+  return function(elem, bool) {
+    return toggleClass(elem, valueOn, valueOff, bool);
+  }
+}
+
+toggleShow = makeClassToggle("show", "hide");
+toggleCollapser = makeClassToggle("collapser", "expander");
+
+function toggleSection(id)
+{
+  var b = toggleShow(document.getElementById("section." + id));
+  toggleCollapser(document.getElementById("control." + id), b);
+  rememberCollapsed(id, b);
+  return b;
+}
+
+var collapsed = {};
+function rememberCollapsed(id, b)
+{
+  if(b)
+    delete collapsed[id]
+  else
+    collapsed[id] = null;
+
+  var sections = [];
+  for(var i in collapsed)
+  {
+    if(collapsed.hasOwnProperty(i))
+      sections.push(i);
+  }
+  // cookie specific to this page; don't use setCookie which sets path=/
+  document.cookie = "collapsed=" + escape(sections.join('+'));
+}
+
+function restoreCollapsed()
+{
+  var cookie = getCookie("collapsed");
+  if(!cookie)
+    return;
+
+  var ids = cookie.split('+');
+  for(var i in ids)
+  {
+    if(document.getElementById("section." + ids[i]))
+      toggleSection(ids[i]);
+  }
+}
+
+function setCookie(name, value) {
+  document.cookie = name + "=" + escape(value) + ";path=/;";
+}
+
+function clearCookie(name) {
+  document.cookie = name + "=;path=/;expires=Thu, 01-Jan-1970 00:00:01 GMT;";
+}
+
+function getCookie(name) {
+  var nameEQ = name + "=";
+  var ca = document.cookie.split(';');
+  for(var i=0;i < ca.length;i++) {
+    var c = ca[i];
+    while (c.charAt(0)==' ') c = c.substring(1,c.length);
+    if (c.indexOf(nameEQ) == 0) {
+      return unescape(c.substring(nameEQ.length,c.length));
+    }
+  }
+  return null;
+}
+
+
+
+var max_results = 75; // 50 is not enough to search for map in the base libraries
+var shown_range = null;
+var last_search = null;
+
+function quick_search()
+{
+    perform_search(false);
+}
+
+function full_search()
+{
+    perform_search(true);
+}
+
+
+function perform_search(full)
+{
+    var text = document.getElementById("searchbox").value.toLowerCase();
+    if (text == last_search && !full) return;
+    last_search = text;
+    
+    var table = document.getElementById("indexlist");
+    var status = document.getElementById("searchmsg");
+    var children = table.firstChild.childNodes;
+    
+    // first figure out the first node with the prefix
+    var first = bisect(-1);
+    var last = (first == -1 ? -1 : bisect(1));
+
+    if (first == -1)
+    {
+        table.className = "";
+        status.innerHTML = "No results found, displaying all";
+    }
+    else if (first == 0 && last == children.length - 1)
+    {
+        table.className = "";
+        status.innerHTML = "";
+    }
+    else if (last - first >= max_results && !full)
+    {
+        table.className = "";
+        status.innerHTML = "More than " + max_results + ", press Search to display";
+    }
+    else
+    {
+        // decide what you need to clear/show
+        if (shown_range)
+            setclass(shown_range[0], shown_range[1], "indexrow");
+        setclass(first, last, "indexshow");
+        shown_range = [first, last];
+        table.className = "indexsearch";
+        status.innerHTML = "";
+    }
+
+    
+    function setclass(first, last, status)
+    {
+        for (var i = first; i <= last; i++)
+        {
+            children[i].className = status;
+        }
+    }
+    
+    
+    // do a binary search, treating 0 as ...
+    // return either -1 (no 0's found) or location of most far match
+    function bisect(dir)
+    {
+        var first = 0, finish = children.length - 1;
+        var mid, success = false;
+
+        while (finish - first > 3)
+        {
+            mid = Math.floor((finish + first) / 2);
+
+            var i = checkitem(mid);
+            if (i == 0) i = dir;
+            if (i == -1)
+                finish = mid;
+            else
+                first = mid;
+        }
+        var a = (dir == 1 ? first : finish);
+        var b = (dir == 1 ? finish : first);
+        for (var i = b; i != a - dir; i -= dir)
+        {
+            if (checkitem(i) == 0) return i;
+        }
+        return -1;
+    }    
+    
+    
+    // from an index, decide what the result is
+    // 0 = match, -1 is lower, 1 is higher
+    function checkitem(i)
+    {
+        var s = getitem(i).toLowerCase().substr(0, text.length);
+        if (s == text) return 0;
+        else return (s > text ? -1 : 1);
+    }
+    
+    
+    // from an index, get its string
+    // this abstracts over alternates
+    function getitem(i)
+    {
+        for ( ; i >= 0; i--)
+        {
+            var s = children[i].firstChild.firstChild.data;
+            if (s.indexOf(' ') == -1)
+                return s;
+        }
+        return ""; // should never be reached
+    }
+}
+
+function setSynopsis(filename) {
+    if (parent.window.synopsis) {
+        if (parent.window.synopsis.location.replace) {
+            // In Firefox this avoids adding the change to the history.
+            parent.window.synopsis.location.replace(filename);
+        } else {
+            parent.window.synopsis.location = filename;
+        }
+    }
+}
+
+function addMenuItem(html) {
+  var menu = document.getElementById("page-menu");
+  if (menu) {
+    var btn = menu.firstChild.cloneNode(false);
+    btn.innerHTML = html;
+    menu.appendChild(btn);
+  }
+}
+
+function adjustForFrames() {
+  var bodyCls;
+  
+  if (parent.location.href == window.location.href) {
+    // not in frames, so add Frames button
+    addMenuItem("<a href='#' onclick='reframe();return true;'>Frames</a>");
+    bodyCls = "no-frame";
+  }
+  else {
+    bodyCls = "in-frame";
+  }
+  addClass(document.body, bodyCls);
+}
+
+function reframe() {
+  setCookie("haddock-reframe", document.URL);
+  window.location = "frames.html";
+}
+
+function postReframe() {
+  var s = getCookie("haddock-reframe");
+  if (s) {
+    parent.window.main.location = s;
+    clearCookie("haddock-reframe");
+  }
+}
+
+function styles() {
+  var i, a, es = document.getElementsByTagName("link"), rs = [];
+  for (i = 0; a = es[i]; i++) {
+    if(a.rel.indexOf("style") != -1 && a.title) {
+      rs.push(a);
+    }
+  }
+  return rs;
+}
+
+function addStyleMenu() {
+  var as = styles();
+  var i, a, btns = "";
+  for(i=0; a = as[i]; i++) {
+    btns += "<li><a href='#' onclick=\"setActiveStyleSheet('"
+      + a.title + "'); return false;\">"
+      + a.title + "</a></li>"
+  }
+  if (as.length > 1) {
+    var h = "<div id='style-menu-holder'>"
+      + "<a href='#' onclick='styleMenu(); return false;'>Style &#9662;</a>"
+      + "<ul id='style-menu' class='hide'>" + btns + "</ul>"
+      + "</div>";
+    addMenuItem(h);
+  }
+}
+
+function setActiveStyleSheet(title) {
+  var as = styles();
+  var i, a, found;
+  for(i=0; a = as[i]; i++) {
+    a.disabled = true;
+          // need to do this always, some browsers are edge triggered
+    if(a.title == title) {
+      found = a;
+    }
+  }
+  if (found) {
+    found.disabled = false;
+    setCookie("haddock-style", title);
+  }
+  else {
+    as[0].disabled = false;
+    clearCookie("haddock-style");
+  }
+  styleMenu(false);
+}
+
+function resetStyle() {
+  var s = getCookie("haddock-style");
+  if (s) setActiveStyleSheet(s);
+}
+
+
+function styleMenu(show) {
+  var m = document.getElementById('style-menu');
+  if (m) toggleShow(m, show);
+}
+
+
+function pageLoad() {
+  addStyleMenu();
+  adjustForFrames();
+  resetStyle();
+  restoreCollapsed();
+}
+
diff --git a/docs/haddock/tensorflow-0.1.0.0/hslogo-16.png b/docs/haddock/tensorflow-0.1.0.0/hslogo-16.png
new file mode 100644
index 0000000000000000000000000000000000000000..0ff8579fbd897417b0d6dad6e920f8882138a7c0
GIT binary patch
literal 1684
zcmV;F25b3=P)<h;3K|Lk000e1NJLTq000*N000mO1^@s6{V(Jg0008+X+uL$Nkc;*
zP;zf(X>4Tx0C)j~RL^S@K@|QrZmG~B2wH0nvUrdpNm;9CMbtL^5n^i$+aI<?kYqDS
z)^0ZI?k2URh~(nYD&9ryp$Gqf9(wR(FG_`4MJV2@C}N=(E%D86Otz)!z|Ma2-Z$@k
zZ+71R4RX>n^?(HA4aZWV5ov6ELTdbo0FI&wK{O>*+w4vx20?>!`FrQsdJlnHR>OPy
zcd~b_n$otK2Za4V;76L-DzNVtaSB-y0*E}{p()372;bw_^6ZZ}PI-92wGS&j#91PI
zKs7DSe@(bk%_Y-7gGe}(^>I=@oY#w#*Bu9GZf3^F5WP>3rn}7Ut74&?PWBFvy`A)a
zPP5)V!Xd&78LdA?xQ(9mjMYElVd13a#D+Z_7&Y|xU=_C-srWU*6kiZcC!$nw*)9$7
zn6CX+@=AhmkT}X@VSsa5NKe;HZuq)~1$`#h6R+ZTR#D-3j}vF!)ZOnz+5)dI4jl{{
z44Mr{P!L4~VVJN`K!!XTF*LGrKO?IK8<Tr7btG!LbYeuYL3=jbJ-1P$-8}v%B5{;M
zwFr{@LH;VQ$xr2Z`O93e*jD$Ht(%&<^58qg<(at}9@o>z<8w`3e3jI8lUGNUta*C8
zn(P`s>{pjD=7Kek#B;Fw@hxAK%$F&Q6vg9J^Xf~4by_hu-=A!MJ3Znq&n~srbFGPs
zH&&aMXZ>nO`|hf|ljc?VPhR!${AbO?W8x_>CU%PFA&Hm8F7cAsOREdwU~R_;ot1_u
z(ruCYB-LPGn!NQdT|ZlRy+(fw^-+`=%+gee_kY4FWHg<*4sZI8+sFJD<oAl_pC|$^
zY~aY5x@}W&?+~G7rEYVs0vEs0eekw!YomTR`~+A$s}`+NHJ>270UUORdLHO0nA4V)
z%{fwsET5CQ>B?eK%uw4yQc~9?*JVo<vzb+5>2}ze(;aRcp*ceL#HUJSllrgm5wQKR
zQu+C;QrUh^8rFfA`ftFz{YAidi-`aL010qNS#tmY4c7nw4c7reD4Tcy00T@(L_t(I
z5sj2vNEA^R$7gqDc6T=2^@fUA2(c`MltuL5<|<Z1YDGjBEr^N`3VJpQ$`C4&9(*Z6
z=|*B$y;-1^F6vXRAd7+(;)77Qps2euv)AeWmj>KW>RWz$&YbU@|M|{$E*8Tu-Ux!w
z1Y*Dr&Ubfr&v-nZaaB{3ilRumrjPmk{sZvQEWlW+{o~IH|8)=s6c#X9S5s5d%<Agu
zgM$153Ttk++ZPA~ZqgqN1|^pNV-|?9-zo48n9+Yiy$*su5czgp*VFj6!1h&DRZR>J
z4@)&QH5|xQY-)^L1n<EM!01(Q6kM8}ojndgRD~2KVIa@>0pTRu0Lx9`08YTjTwn^6
z0;b1+aQ@)n;Em$q;=7BBi)v0zj&o^g>0Whp^_^5IbxIUP8C@y9;R?*Ouu}rmfxbU=
zwtWVNke-m!=`7bYEhWpcI5#)9qp`8E0lr6IQ)ARL3Ui}Af@grj8aN1=r>Cb+prlzO
zNfJs*N_tUm2ZL<f%*@E?H`Ak_J`Q#Qri<;K{{H@Bg@uKWS5{WE+}vCx91geD*4DNm
zJ?D^~dGz7lWtbWH`o<0O#|{R6h9M*GMpjnV?Y_RgatOb~1le#ikPKTNIn%<;=>%5*
zPmL2??da$TR904gL(VDAQ<BE%k@5ZA>-Fv_Dk}Pdw*4T(%*f4MKLRg=4ekMjhe2mW
zMFsBwg%ftWT}0kxRaIk1k7qJ8*#cKB;Ft{i`zVIs-Nqge;!!Ld7#O&Qqu7e0sJmP)
z$MW*>L$vSB&dxp@iA3U9fo)-7!Czlr{|o7<LYh`Fd@X2-b0fIa*Wdy$({Zt@t7`&L
zhoaGF!l5{1P1Qm;<GaSj#=Oa7atf1JZ8n=F9*@uC?b(X48O9kPoH~h;!o2=WM?ZLl
z$Lj;4K~)fAdW@^N%nzT>Hv{1oyg3xsu%gn@(b1>$;SM-ZaQ`HV=V0s;lr%d8bd;xY
zGw<i+<$17ay36gdrqk(Lxc1|{y}dSeM5OIV(FjhBg5#OM<!Z*8(@<7crlVw+B-8R1
zl{3J6w9fYQ^vojauaRgZ=~SE&CK!R3pCe!v3>Nvm3=Iu=tyXIgtJnf@A(2S@M140N
ew{UA~tMxaJq;$xaSSi*30000<MNUMnLSTa0p)&^n

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-0.1.0.0/index-frames.html b/docs/haddock/tensorflow-0.1.0.0/index-frames.html
new file mode 100644
index 0000000..87bdf19
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/index-frames.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-0.1.0.0: TensorFlow bindings.</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-list"><p class="caption">Modules</p><ul><li class="module"><a href="TensorFlow-Build.html" target="main">TensorFlow.Build</a></li><li class="module"><a href="TensorFlow-BuildOp.html" target="main">TensorFlow.BuildOp</a></li><li class="module"><a href="TensorFlow-ControlFlow.html" target="main">TensorFlow.ControlFlow</a></li><li class="module"><a href="TensorFlow-Internal-FFI.html" target="main">TensorFlow.Internal.FFI</a></li><li class="module"><a href="TensorFlow-Internal-VarInt.html" target="main">TensorFlow.Internal.VarInt</a></li><li class="module"><a href="TensorFlow-Nodes.html" target="main">TensorFlow.Nodes</a></li><li class="module"><a href="TensorFlow-Output.html" target="main">TensorFlow.Output</a></li><li class="module"><a href="TensorFlow-Session.html" target="main">TensorFlow.Session</a></li><li class="module"><a href="TensorFlow-Tensor.html" target="main">TensorFlow.Tensor</a></li><li class="module"><a href="TensorFlow-Types.html" target="main">TensorFlow.Types</a></li></ul></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/index.html b/docs/haddock/tensorflow-0.1.0.0/index.html
new file mode 100644
index 0000000..d5426e3
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-0.1.0.0: TensorFlow bindings.</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-0.1.0.0: TensorFlow bindings.</p></div><div id="content"><div id="description"><h1>tensorflow-0.1.0.0: TensorFlow bindings.</h1><div class="doc"><p>Please see README.md</p></div></div><div id="module-list"><p class="caption">Modules</p><ul><li><span id="control.n.1" class="module collapser" onclick="toggleSection('n.1')">TensorFlow</span><ul id="section.n.1" class="show"><li><span class="module"><a href="TensorFlow-Build.html">TensorFlow.Build</a></span></li><li><span class="module"><a href="TensorFlow-BuildOp.html">TensorFlow.BuildOp</a></span></li><li><span class="module"><a href="TensorFlow-ControlFlow.html">TensorFlow.ControlFlow</a></span></li><li><span id="control.n.1.4" class="module collapser" onclick="toggleSection('n.1.4')">Internal</span><ul id="section.n.1.4" class="show"><li><span class="module"><a href="TensorFlow-Internal-FFI.html">TensorFlow.Internal.FFI</a></span></li><li><span class="module"><a href="TensorFlow-Internal-VarInt.html">TensorFlow.Internal.VarInt</a></span> Encoders and decoders for varint types.</li></ul></li><li><span class="module"><a href="TensorFlow-Nodes.html">TensorFlow.Nodes</a></span></li><li><span class="module"><a href="TensorFlow-Output.html">TensorFlow.Output</a></span></li><li><span class="module"><a href="TensorFlow-Session.html">TensorFlow.Session</a></span></li><li><span class="module"><a href="TensorFlow-Tensor.html">TensorFlow.Tensor</a></span></li><li><span class="module"><a href="TensorFlow-Types.html">TensorFlow.Types</a></span></li></ul></li></ul></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Build.html b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Build.html
new file mode 100644
index 0000000..a4b221c
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Build.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Build</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.Build</p></div><div id="interface"><h1>Graph node types</h1><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Build.html#t:ControlNode" target="main">ControlNode</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Build.html#t:Unique" target="main">Unique</a></p></div><h1>Ops</h1><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:explicitName" target="main">explicitName</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:implicitName" target="main">implicitName</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:opDef" target="main">opDef</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:opDefWithName" target="main">opDefWithName</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:opName" target="main">opName</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:opType" target="main">opType</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:opAttr" target="main">opAttr</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:opInputs" target="main">opInputs</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:opControlInputs" target="main">opControlInputs</a></p></div><h1>The Build monad</h1><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Build.html#t:GraphState" target="main">GraphState</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:render" target="main">render</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:renderNodeName" target="main">renderNodeName</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:renderedNodeDefs" target="main">renderedNodeDefs</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Build.html#t:BuildT" target="main">BuildT</a> m a</p></div><div class="top"><p class="src"><span class="keyword">type</span> <a href="TensorFlow-Build.html#t:Build" target="main">Build</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:addInitializer" target="main">addInitializer</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:hoistBuildT" target="main">hoistBuildT</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:evalBuildT" target="main">evalBuildT</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:runBuildT" target="main">runBuildT</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:asGraphDef" target="main">asGraphDef</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:addGraphDef" target="main">addGraphDef</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:flushInitializers" target="main">flushInitializers</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:flushNodeBuffer" target="main">flushNodeBuffer</a></p></div><h1>Creating and looking up Ops</h1><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:getOrAddOp" target="main">getOrAddOp</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:addNewOp" target="main">addNewOp</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:renderOutput" target="main">renderOutput</a></p></div><h1>Modifying all nodes in a Build action</h1><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:colocateWith" target="main">colocateWith</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:withStateLens" target="main">withStateLens</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:withDevice" target="main">withDevice</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:withNameScope" target="main">withNameScope</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:withNodeDependencies" target="main">withNodeDependencies</a></p></div><h1>Internal Summary related bits.</h1><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:addSummary" target="main">addSummary</a></p></div><div class="top"><p class="src"><span class="keyword">type</span> <a href="TensorFlow-Build.html#t:SummaryTensor" target="main">SummaryTensor</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Build.html#v:collectAllSummaries" target="main">collectAllSummaries</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-BuildOp.html b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-BuildOp.html
new file mode 100644
index 0000000..cd0e477
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-BuildOp.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.BuildOp</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.BuildOp</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">class</span> <a href="TensorFlow-BuildOp.html#t:OpResult" target="main">OpResult</a> a</p></div><div class="top"><p class="src"><span class="keyword">class</span> <a href="TensorFlow-BuildOp.html#t:BuildOp" target="main">BuildOp</a> f</p></div><div class="top"><p class="src"><a href="TensorFlow-BuildOp.html#v:buildOp" target="main">buildOp</a></p></div><div class="top"><p class="src"><a href="TensorFlow-BuildOp.html#v:buildListOp" target="main">buildListOp</a></p></div><div class="top"><p class="src"><a href="TensorFlow-BuildOp.html#v:eqLengthGuard" target="main">eqLengthGuard</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-ControlFlow.html b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-ControlFlow.html
new file mode 100644
index 0000000..3d5fc95
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-ControlFlow.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.ControlFlow</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.ControlFlow</p></div><div id="interface"><h1>Dependencies</h1><div class="top"><p class="src"><a href="TensorFlow-ControlFlow.html#v:withControlDependencies" target="main">withControlDependencies</a></p></div><div class="top"><p class="src"><a href="TensorFlow-ControlFlow.html#v:group" target="main">group</a></p></div><h1>Operations</h1><div class="top"><p class="src"><a href="TensorFlow-ControlFlow.html#v:identity" target="main">identity</a></p></div><div class="top"><p class="src"><a href="TensorFlow-ControlFlow.html#v:noOp" target="main">noOp</a></p></div><div class="top"><p class="src"><a href="TensorFlow-ControlFlow.html#v:named" target="main">named</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Internal-FFI.html b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Internal-FFI.html
new file mode 100644
index 0000000..bd4a296
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Internal-FFI.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Internal.FFI</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.Internal.FFI</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Internal-FFI.html#t:TensorFlowException" target="main">TensorFlowException</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Internal-FFI.html#t:Session" target="main">Session</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Internal-FFI.html#v:withSession" target="main">withSession</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Internal-FFI.html#v:extendGraph" target="main">extendGraph</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Internal-FFI.html#v:run" target="main">run</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Internal-FFI.html#t:TensorData" target="main">TensorData</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Internal-FFI.html#v:setSessionConfig" target="main">setSessionConfig</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Internal-FFI.html#v:setSessionTarget" target="main">setSessionTarget</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Internal-FFI.html#v:getAllOpList" target="main">getAllOpList</a></p></div><h1>Internal helper.</h1><div class="top"><p class="src"><a href="TensorFlow-Internal-FFI.html#v:useProtoAsVoidPtrLen" target="main">useProtoAsVoidPtrLen</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Internal-VarInt.html b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Internal-VarInt.html
new file mode 100644
index 0000000..cd36a75
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Internal-VarInt.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Internal.VarInt</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.Internal.VarInt</p></div><div id="interface"><div class="top"><p class="src"><a href="TensorFlow-Internal-VarInt.html#v:getVarInt" target="main">getVarInt</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Internal-VarInt.html#v:putVarInt" target="main">putVarInt</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Nodes.html b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Nodes.html
new file mode 100644
index 0000000..6765575
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Nodes.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Nodes</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.Nodes</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">class</span> <a href="TensorFlow-Nodes.html#t:Nodes" target="main">Nodes</a> t</p></div><div class="top"><p class="src"><span class="keyword">class</span> <a href="TensorFlow-Nodes.html#t:Fetchable" target="main">Fetchable</a> t a</p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Nodes.html#t:Fetch" target="main">Fetch</a> a</p></div><div class="top"><p class="src"><a href="TensorFlow-Nodes.html#v:nodesUnion" target="main">nodesUnion</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Nodes.html#v:fetchTensorList" target="main">fetchTensorList</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Nodes.html#v:fetchTensorVector" target="main">fetchTensorVector</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Nodes.html#t:Scalar" target="main">Scalar</a> a</p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Output.html b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Output.html
new file mode 100644
index 0000000..22c2d31
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Output.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Output</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.Output</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Output.html#t:ControlNode" target="main">ControlNode</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Output.html#t:Device" target="main">Device</a></p></div><h1>Ops</h1><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Output.html#t:NodeName" target="main">NodeName</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Output.html#t:Op" target="main">Op</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Output.html#v:opUnrendered" target="main">opUnrendered</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Output.html#t:OpDef" target="main">OpDef</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Output.html#v:opName" target="main">opName</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Output.html#v:opType" target="main">opType</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Output.html#v:opAttr" target="main">opAttr</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Output.html#v:opInputs" target="main">opInputs</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Output.html#v:opControlInputs" target="main">opControlInputs</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Output.html#t:OpType" target="main">OpType</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Output.html#t:OutputIx" target="main">OutputIx</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Output.html#t:Output" target="main">Output</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Output.html#v:output" target="main">output</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Output.html#v:outputIndex" target="main">outputIndex</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Output.html#v:outputOp" target="main">outputOp</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Output.html#t:PendingNodeName" target="main">PendingNodeName</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Session.html b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Session.html
new file mode 100644
index 0000000..a37ad20
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Session.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Session</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.Session</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Session.html#t:Session" target="main">Session</a> a</p></div><h1>Opaque value created via <code><a href="TensorFlow-Session.html#v:sessionConfig">sessionConfig</a></code> and <code><a href="TensorFlow-Session.html#v:sessionTarget">sessionTarget</a></code>.</h1><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Session.html#t:SessionOption" target="main">SessionOption</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Session.html#v:sessionConfig" target="main">sessionConfig</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Session.html#v:sessionTarget" target="main">sessionTarget</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Session.html#v:runSession" target="main">runSession</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Session.html#v:runSessionWithOptions" target="main">runSessionWithOptions</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Session.html#v:build" target="main">build</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Session.html#v:buildAnd" target="main">buildAnd</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Session.html#v:buildWithSummary" target="main">buildWithSummary</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Session.html#v:extend" target="main">extend</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Session.html#v:addGraphDef" target="main">addGraphDef</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Session.html#v:run" target="main">run</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Session.html#v:runWithFeeds" target="main">runWithFeeds</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Session.html#v:run_" target="main">run_</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Session.html#v:runWithFeeds_" target="main">runWithFeeds_</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Session.html#v:asyncProdNodes" target="main">asyncProdNodes</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Tensor.html b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Tensor.html
new file mode 100644
index 0000000..cdab973
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Tensor.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Tensor</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.Tensor</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Tensor.html#t:Tensor" target="main">Tensor</a> v a</p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Tensor.html#t:Value" target="main">Value</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Tensor.html#t:Ref" target="main">Ref</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Tensor.html#t:TensorKind" target="main">TensorKind</a> v</p></div><div class="top"><p class="src"><a href="TensorFlow-Tensor.html#v:tensorKind" target="main">tensorKind</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Tensor.html#v:tensorOutput" target="main">tensorOutput</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Tensor.html#v:tensorAttr" target="main">tensorAttr</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Tensor.html#v:value" target="main">value</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Tensor.html#t:Feed" target="main">Feed</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Tensor.html#v:feed" target="main">feed</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Tensor.html#v:tensorFromName" target="main">tensorFromName</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Types.html b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Types.html
new file mode 100644
index 0000000..5ba106f
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/mini_TensorFlow-Types.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Types</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.Types</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">class</span> <a href="TensorFlow-Types.html#t:TensorType" target="main">TensorType</a> a</p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Types.html#t:TensorData" target="main">TensorData</a> a</p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Types.html#t:Shape" target="main">Shape</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Types.html#v:protoShape" target="main">protoShape</a></p></div><div class="top"><p class="src"><span class="keyword">class</span> <a href="TensorFlow-Types.html#t:Attribute" target="main">Attribute</a> a</p></div><h1>Type constraints</h1><div class="top"><p class="src"><span class="keyword">type</span> <a href="TensorFlow-Types.html#t:OneOf" target="main">OneOf</a> ts a</p></div><div class="top"><p class="src"><span class="keyword">type family</span> a <a href="#t:-47--61-">/=</a> b :: <a href="../base-4.8.2.0/GHC-Exts.html#t:Constraint">Constraint</a></p></div><h2>Implementation of constraints</h2><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Types.html#t:TypeError" target="main">TypeError</a> a</p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Types.html#t:ExcludedCase" target="main">ExcludedCase</a></p></div><div class="top"><p class="src"><span class="keyword">type family</span> <a href="#t:TensorTypes">TensorTypes</a> ts :: <a href="../base-4.8.2.0/GHC-Exts.html#t:Constraint">Constraint</a></p></div><div class="top"><p class="src"><span class="keyword">type family</span> <a href="#t:NoneOf">NoneOf</a> ts a :: <a href="../base-4.8.2.0/GHC-Exts.html#t:Constraint">Constraint</a></p></div><div class="top"><p class="src"><span class="keyword">type family</span> as <a href="#t:-92--92-">\\</a> bs</p></div><div class="top"><p class="src"><span class="keyword">type family</span> <a href="#t:Delete">Delete</a> a as</p></div><div class="top"><p class="src"><span class="keyword">type</span> <a href="TensorFlow-Types.html#t:AllTensorTypes" target="main">AllTensorTypes</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-0.1.0.0/minus.gif b/docs/haddock/tensorflow-0.1.0.0/minus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..1deac2fe1a42e35b994f1b855488f392c50f6a89
GIT binary patch
literal 56
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS53HJ0a;6C;`j@u!^I(6E|
Jh4LZ{)&P#)4$%Mr

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-0.1.0.0/ocean.css b/docs/haddock/tensorflow-0.1.0.0/ocean.css
new file mode 100644
index 0000000..1110b40
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/ocean.css
@@ -0,0 +1,600 @@
+/* @group Fundamentals */
+
+* { margin: 0; padding: 0 }
+
+/* Is this portable? */
+html {
+  background-color: white;
+  width: 100%;
+  height: 100%;
+}
+
+body {
+  background: white;
+  color: black;
+  text-align: left;
+  min-height: 100%;
+  position: relative;
+}
+
+p {
+  margin: 0.8em 0;
+}
+
+ul, ol {
+  margin: 0.8em 0 0.8em 2em;
+}
+
+dl {
+  margin: 0.8em 0;
+}
+
+dt {
+  font-weight: bold;
+}
+dd {
+  margin-left: 2em;
+}
+
+a { text-decoration: none; }
+a[href]:link { color: rgb(196,69,29); }
+a[href]:visited { color: rgb(171,105,84); }
+a[href]:hover { text-decoration:underline; }
+
+/* @end */
+
+/* @group Fonts & Sizes */
+
+/* Basic technique & IE workarounds from YUI 3
+   For reasons, see:
+      http://yui.yahooapis.com/3.1.1/build/cssfonts/fonts.css
+ */
+
+body {
+	font:13px/1.4 sans-serif;
+	*font-size:small; /* for IE */
+	*font:x-small; /* for IE in quirks mode */
+}
+
+h1 { font-size: 146.5%; /* 19pt */ }
+h2 { font-size: 131%;   /* 17pt */ }
+h3 { font-size: 116%;   /* 15pt */ }
+h4 { font-size: 100%;   /* 13pt */ }
+h5 { font-size: 100%;   /* 13pt */ }
+
+select, input, button, textarea {
+	font:99% sans-serif;
+}
+
+table {
+	font-size:inherit;
+	font:100%;
+}
+
+pre, code, kbd, samp, tt, .src {
+	font-family:monospace;
+	*font-size:108%;
+	line-height: 124%;
+}
+
+.links, .link {
+  font-size: 85%; /* 11pt */
+}
+
+#module-header .caption {
+  font-size: 182%; /* 24pt */
+}
+
+.info  {
+  font-size: 85%; /* 11pt */
+}
+
+#table-of-contents, #synopsis  {
+  /* font-size: 85%; /* 11pt */
+}
+
+
+/* @end */
+
+/* @group Common */
+
+.caption, h1, h2, h3, h4, h5, h6 {
+  font-weight: bold;
+  color: rgb(78,98,114);
+  margin: 0.8em 0 0.4em;
+}
+
+* + h1, * + h2, * + h3, * + h4, * + h5, * + h6 {
+  margin-top: 2em;
+}
+
+h1 + h2, h2 + h3, h3 + h4, h4 + h5, h5 + h6 {
+  margin-top: inherit;
+}
+
+ul.links {
+  list-style: none;
+  text-align: left;
+  float: right;
+  display: inline-table;
+  margin: 0 0 0 1em;
+}
+
+ul.links li {
+  display: inline;
+  border-left: 1px solid #d5d5d5;
+  white-space: nowrap;
+  padding: 0;
+}
+
+ul.links li a {
+  padding: 0.2em 0.5em;
+}
+
+.hide { display: none; }
+.show { display: inherit; }
+.clear { clear: both; }
+
+.collapser {
+  background-image: url(minus.gif);
+  background-repeat: no-repeat;
+}
+.expander {
+  background-image: url(plus.gif);
+  background-repeat: no-repeat;
+}
+p.caption.collapser,
+p.caption.expander {
+  background-position: 0 0.4em;
+}
+.collapser, .expander {
+  padding-left: 14px;
+  margin-left: -14px;
+  cursor: pointer;
+}
+
+pre {
+  padding: 0.25em;
+  margin: 0.8em 0;
+  background: rgb(229,237,244);
+  overflow: auto;
+  border-bottom: 0.25em solid white;
+  /* white border adds some space below the box to compensate
+     for visual extra space that paragraphs have between baseline
+     and the bounding box */
+}
+
+.src {
+  background: #f0f0f0;
+  padding: 0.2em 0.5em;
+}
+
+.keyword { font-weight: normal; }
+.def { font-weight: bold; }
+
+
+/* @end */
+
+/* @group Page Structure */
+
+#content {
+  margin: 0 auto;
+  padding: 0 2em 6em;
+}
+
+#package-header {
+  background: rgb(41,56,69);
+  border-top: 5px solid rgb(78,98,114);
+  color: #ddd;
+  padding: 0.2em;
+  position: relative;
+  text-align: left;
+}
+
+#package-header .caption {
+  background: url(hslogo-16.png) no-repeat 0em;
+  color: white;
+  margin: 0 2em;
+  font-weight: normal;
+  font-style: normal;
+  padding-left: 2em;
+}
+
+#package-header a:link, #package-header a:visited { color: white; }
+#package-header a:hover { background: rgb(78,98,114); }
+
+#module-header .caption {
+  color: rgb(78,98,114);
+  font-weight: bold;
+  border-bottom: 1px solid #ddd;
+}
+
+table.info {
+  float: right;
+  padding: 0.5em 1em;
+  border: 1px solid #ddd;
+  color: rgb(78,98,114);
+  background-color: #fff;
+  max-width: 40%;
+  border-spacing: 0;
+  position: relative;
+  top: -0.5em;
+  margin: 0 0 0 2em;
+}
+
+.info th {
+	padding: 0 1em 0 0;
+}
+
+div#style-menu-holder {
+  position: relative;
+  z-index: 2;
+  display: inline;
+}
+
+#style-menu {
+  position: absolute;
+  z-index: 1;
+  overflow: visible;
+  background: #374c5e;
+  margin: 0;
+  text-align: center;
+  right: 0;
+  padding: 0;
+  top: 1.25em;
+}
+
+#style-menu li {
+	display: list-item;
+	border-style: none;
+	margin: 0;
+	padding: 0;
+	color: #000;
+	list-style-type: none;
+}
+
+#style-menu li + li {
+	border-top: 1px solid #919191;
+}
+
+#style-menu a {
+  width: 6em;
+  padding: 3px;
+  display: block;
+}
+
+#footer {
+  background: #ddd;
+  border-top: 1px solid #aaa;
+  padding: 0.5em 0;
+  color: #666;
+  text-align: center;
+  position: absolute;
+  bottom: 0;
+  width: 100%;
+  height: 3em;
+}
+
+/* @end */
+
+/* @group Front Matter */
+
+#table-of-contents {
+  float: right;
+  clear: right;
+  background: #faf9dc;
+  border: 1px solid #d8d7ad;
+  padding: 0.5em 1em;
+  max-width: 20em;
+  margin: 0.5em 0 1em 1em;
+}
+
+#table-of-contents .caption {
+  text-align: center;
+  margin: 0;
+}
+
+#table-of-contents ul {
+  list-style: none;
+  margin: 0;
+}
+
+#table-of-contents ul ul {
+  margin-left: 2em;
+}
+
+#description .caption {
+  display: none;
+}
+
+#synopsis {
+  display: none;
+}
+
+.no-frame #synopsis {
+  display: block;
+  position: fixed;
+  right: 0;
+  height: 80%;
+  top: 10%;
+  padding: 0;
+  max-width: 75%;
+}
+
+#synopsis .caption {
+  float: left;
+  width: 29px;
+  color: rgba(255,255,255,0);
+  height: 110px;
+  margin: 0;
+  font-size: 1px;
+  padding: 0;
+}
+
+#synopsis p.caption.collapser {
+  background: url(synopsis.png) no-repeat -64px -8px;
+}
+
+#synopsis p.caption.expander {
+  background: url(synopsis.png) no-repeat 0px -8px;
+}
+
+#synopsis ul {
+  height: 100%;
+  overflow: auto;
+  padding: 0.5em;
+  margin: 0;
+}
+
+#synopsis ul ul {
+  overflow: hidden;
+}
+
+#synopsis ul,
+#synopsis ul li.src {
+  background-color: #faf9dc;
+  white-space: nowrap;
+  list-style: none;
+  margin-left: 0;
+}
+
+/* @end */
+
+/* @group Main Content */
+
+#interface div.top { margin: 2em 0; }
+#interface h1 + div.top,
+#interface h2 + div.top,
+#interface h3 + div.top,
+#interface h4 + div.top,
+#interface h5 + div.top {
+ 	margin-top: 1em;
+}
+#interface p.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface td.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface span.fixity {
+  color: #919191;
+  border-left: 1px solid #919191;
+  padding: 0.2em 0.5em 0.2em 0.5em;
+  margin: 0 -1em 0 1em;
+}
+
+#interface span.rightedge {
+  border-left: 1px solid #919191;
+  padding: 0.2em 0 0.2em 0;
+  margin: 0 0 0 1em;
+}
+
+#interface table { border-spacing: 2px; }
+#interface td {
+  vertical-align: top;
+  padding-left: 0.5em;
+}
+#interface td.src {
+  white-space: nowrap;
+}
+#interface td.doc p {
+  margin: 0;
+}
+#interface td.doc p + p {
+  margin-top: 0.8em;
+}
+
+.clearfix:after {
+  clear: both;
+  content: " ";
+  display: block;
+  height: 0;
+  visibility: hidden;
+}
+
+.subs dl {
+  margin: 0;
+}
+
+.subs dt {
+  float: left;
+  clear: left;
+  display: block;
+  margin: 1px 0;
+}
+
+.subs dd {
+  float: right;
+  width: 90%;
+  display: block;
+  padding-left: 0.5em;
+  margin-bottom: 0.5em;
+}
+
+.subs dd.empty {
+  display: none;
+}
+
+.subs dd p {
+  margin: 0;
+}
+
+/* Render short-style data instances */
+.inst ul {
+  height: 100%;
+  padding: 0.5em;
+  margin: 0;
+}
+
+.inst, .inst li {
+  list-style: none;
+  margin-left: 1em;
+}
+
+/* Workaround for bug in Firefox (issue #384) */
+.inst-left {
+  float: left;
+}
+
+.top p.src {
+  border-top: 1px solid #ccc;
+}
+
+.subs, .doc {
+  /* use this selector for one level of indent */
+  padding-left: 2em;
+}
+
+.warning {
+  color: red;
+}
+
+.arguments {
+  margin-top: -0.4em;
+}
+.arguments .caption {
+  display: none;
+}
+
+.fields { padding-left: 1em; }
+
+.fields .caption { display: none; }
+
+.fields p { margin: 0 0; }
+
+/* this seems bulky to me
+.methods, .constructors {
+  background: #f8f8f8;
+  border: 1px solid #eee;
+}
+*/
+
+/* @end */
+
+/* @group Auxillary Pages */
+
+
+.extension-list {
+    list-style-type: none;
+    margin-left: 0;
+}
+
+#mini {
+  margin: 0 auto;
+  padding: 0 1em 1em;
+}
+
+#mini > * {
+  font-size: 93%; /* 12pt */
+}
+
+#mini #module-list .caption,
+#mini #module-header .caption {
+  font-size: 125%; /* 15pt */
+}
+
+#mini #interface h1,
+#mini #interface h2,
+#mini #interface h3,
+#mini #interface h4 {
+  font-size: 109%; /* 13pt */
+  margin: 1em 0 0;
+}
+
+#mini #interface .top,
+#mini #interface .src {
+  margin: 0;
+}
+
+#mini #module-list ul {
+  list-style: none;
+  margin: 0;
+}
+
+#alphabet ul {
+	list-style: none;
+	padding: 0;
+	margin: 0.5em 0 0;
+	text-align: center;
+}
+
+#alphabet li {
+	display: inline;
+	margin: 0 0.25em;
+}
+
+#alphabet a {
+	font-weight: bold;
+}
+
+#index .caption,
+#module-list .caption { font-size: 131%; /* 17pt */ }
+
+#index table {
+  margin-left: 2em;
+}
+
+#index .src {
+  font-weight: bold;
+}
+#index .alt {
+  font-size: 77%; /* 10pt */
+  font-style: italic;
+  padding-left: 2em;
+}
+
+#index td + td {
+  padding-left: 1em;
+}
+
+#module-list ul {
+  list-style: none;
+  margin: 0 0 0 2em;
+}
+
+#module-list li {
+  clear: right;
+}
+
+#module-list span.collapser,
+#module-list span.expander {
+  background-position: 0 0.3em;
+}
+
+#module-list .package {
+  float: right;
+}
+
+/* @end */
diff --git a/docs/haddock/tensorflow-0.1.0.0/plus.gif b/docs/haddock/tensorflow-0.1.0.0/plus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..2d15c14173d23f664b955cd24f51c82f5f09d91d
GIT binary patch
literal 59
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS5iS+a@;NG@rPSdH*>gbBX
M^XE!9f*2UA0nx1yDgXcg

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Build.html b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Build.html
new file mode 100644
index 0000000..57b54be
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Build.html
@@ -0,0 +1,387 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/Build.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE GeneralizedNewtypeDeriving #-}</span>
+<a name="line-16"></a><span class='hs-comment'>{-# LANGUAGE LambdaCase #-}</span>
+<a name="line-17"></a><span class='hs-comment'>{-# LANGUAGE Rank2Types #-}</span>
+<a name="line-18"></a><span class='hs-comment'>{-# LANGUAGE OverloadedStrings #-}</span>
+<a name="line-19"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Build</span>
+<a name="line-20"></a>    <span class='hs-layout'>(</span> <span class='hs-comment'>-- * Graph node types</span>
+<a name="line-21"></a>      <span class='hs-conid'>ControlNode</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-22"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Unique</span>
+<a name="line-23"></a>    <span class='hs-comment'>-- * Ops</span>
+<a name="line-24"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>explicitName</span>
+<a name="line-25"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>implicitName</span>
+<a name="line-26"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>opDef</span>
+<a name="line-27"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>opDefWithName</span>
+<a name="line-28"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>opName</span>
+<a name="line-29"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>opType</span>
+<a name="line-30"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>opAttr</span>
+<a name="line-31"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>opInputs</span>
+<a name="line-32"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>opControlInputs</span>
+<a name="line-33"></a>    <span class='hs-comment'>-- * The Build monad</span>
+<a name="line-34"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>GraphState</span>
+<a name="line-35"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>render</span>
+<a name="line-36"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>renderNodeName</span>
+<a name="line-37"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>renderedNodeDefs</span>
+<a name="line-38"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>BuildT</span>
+<a name="line-39"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Build</span>
+<a name="line-40"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>addInitializer</span>
+<a name="line-41"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>hoistBuildT</span>
+<a name="line-42"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>evalBuildT</span>
+<a name="line-43"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>runBuildT</span>
+<a name="line-44"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>asGraphDef</span>
+<a name="line-45"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>addGraphDef</span>
+<a name="line-46"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>flushInitializers</span>
+<a name="line-47"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>flushNodeBuffer</span>
+<a name="line-48"></a>    <span class='hs-comment'>-- * Creating and looking up Ops</span>
+<a name="line-49"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>getOrAddOp</span>
+<a name="line-50"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>addNewOp</span>
+<a name="line-51"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>renderOutput</span>
+<a name="line-52"></a>    <span class='hs-comment'>-- * Modifying all nodes in a Build action</span>
+<a name="line-53"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>colocateWith</span>
+<a name="line-54"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>withStateLens</span>
+<a name="line-55"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>withDevice</span>
+<a name="line-56"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>withNameScope</span>
+<a name="line-57"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>withNodeDependencies</span>
+<a name="line-58"></a>    <span class='hs-comment'>-- * Internal Summary related bits.</span>
+<a name="line-59"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>addSummary</span>
+<a name="line-60"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>SummaryTensor</span>
+<a name="line-61"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>collectAllSummaries</span>
+<a name="line-62"></a>    <span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-63"></a>
+<a name="line-64"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Monad</span><span class='hs-varop'>.</span><span class='hs-conid'>IO</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span> <span class='hs-layout'>(</span><span class='hs-conid'>MonadIO</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-65"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Monad</span><span class='hs-varop'>.</span><span class='hs-conid'>Trans</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span> <span class='hs-layout'>(</span><span class='hs-conid'>MonadTrans</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-66"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Monad</span><span class='hs-varop'>.</span><span class='hs-conid'>Trans</span><span class='hs-varop'>.</span><span class='hs-conid'>State</span><span class='hs-varop'>.</span><span class='hs-conid'>Strict</span><span class='hs-layout'>(</span><span class='hs-conid'>StateT</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-varid'>mapStateT</span><span class='hs-layout'>,</span> <span class='hs-varid'>evalStateT</span><span class='hs-layout'>)</span>
+<a name="line-67"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-layout'>(</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>
+<a name="line-68"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-layout'>(</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-69"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Functor</span><span class='hs-varop'>.</span><span class='hs-conid'>Identity</span> <span class='hs-layout'>(</span><span class='hs-conid'>Identity</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-70"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Strict</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Map</span>
+<a name="line-71"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Monoid</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varop'>&lt;&gt;</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-72"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Set</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Set</span>
+<a name="line-73"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Set</span> <span class='hs-layout'>(</span><span class='hs-conid'>Set</span><span class='hs-layout'>)</span>
+<a name="line-74"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>String</span> <span class='hs-layout'>(</span><span class='hs-conid'>IsString</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-75"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-layout'>(</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-76"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Text</span>
+<a name="line-77"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span> <span class='hs-layout'>(</span><span class='hs-conid'>Lens'</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>.~</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>^.</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>&amp;</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-78"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>State</span><span class='hs-varop'>.</span><span class='hs-conid'>Strict</span> <span class='hs-layout'>(</span><span class='hs-conid'>MonadState</span><span class='hs-layout'>,</span> <span class='hs-varid'>use</span><span class='hs-layout'>,</span> <span class='hs-varid'>uses</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>.=</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>&lt;&gt;=</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>%=</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-79"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span> <span class='hs-layout'>(</span><span class='hs-varid'>lens</span><span class='hs-layout'>)</span>
+<a name="line-80"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Graph</span>
+<a name="line-81"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>GraphDef</span>
+<a name="line-82"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>node</span>
+<a name="line-83"></a>    <span class='hs-layout'>)</span>
+<a name="line-84"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span>
+<a name="line-85"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>NodeDef</span>
+<a name="line-86"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>attr</span>
+<a name="line-87"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>input</span>
+<a name="line-88"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>device</span>
+<a name="line-89"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>name</span>
+<a name="line-90"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>op</span>
+<a name="line-91"></a>    <span class='hs-layout'>)</span>
+<a name="line-92"></a>
+<a name="line-93"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Orphans</span> <span class='hs-conid'>()</span>
+<a name="line-94"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Output</span>
+<a name="line-95"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span>
+<a name="line-96"></a>
+<a name="line-97"></a><a name="Unique"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>Unique</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Unique</span> <span class='hs-conid'>Int</span>
+<a name="line-98"></a>    <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Eq</span><span class='hs-layout'>,</span> <span class='hs-conid'>Ord</span><span class='hs-layout'>,</span> <span class='hs-conid'>Enum</span><span class='hs-layout'>)</span>
+<a name="line-99"></a>
+<a name="line-100"></a><span class='hs-comment'>--------------</span>
+<a name="line-101"></a>
+<a name="line-102"></a><a name="implicitName"></a><span class='hs-definition'>implicitName</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>PendingNodeName</span>
+<a name="line-103"></a><span class='hs-definition'>implicitName</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>ImplicitName</span>
+<a name="line-104"></a>
+<a name="line-105"></a><a name="explicitName"></a><span class='hs-definition'>explicitName</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>PendingNodeName</span>
+<a name="line-106"></a><span class='hs-definition'>explicitName</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>ExplicitName</span>
+<a name="line-107"></a>
+<a name="line-108"></a><a name="Scope"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>Scope</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Scope</span> <span class='hs-layout'>{</span><span class='hs-varid'>unScope</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span><span class='hs-layout'>}</span>
+<a name="line-109"></a>    <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Eq</span><span class='hs-layout'>,</span> <span class='hs-conid'>Ord</span><span class='hs-layout'>,</span> <span class='hs-conid'>IsString</span><span class='hs-layout'>)</span>
+<a name="line-110"></a>
+<a name="line-111"></a><a name="instance%20Show%20Scope"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Show</span> <span class='hs-conid'>Scope</span> <span class='hs-keyword'>where</span>
+<a name="line-112"></a>    <span class='hs-varid'>show</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>show</span> <span class='hs-varop'>.</span> <span class='hs-varid'>unScope</span>
+<a name="line-113"></a>
+<a name="line-114"></a><a name="opDef"></a><span class='hs-definition'>opDef</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpType</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>OpDef</span>
+<a name="line-115"></a><span class='hs-definition'>opDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>opDefWithName</span> <span class='hs-conid'>ImplicitName</span>
+<a name="line-116"></a>
+<a name="line-117"></a><a name="opDefWithName"></a><span class='hs-definition'>opDefWithName</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>PendingNodeName</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>OpType</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>OpDef</span>
+<a name="line-118"></a><span class='hs-definition'>opDefWithName</span> <span class='hs-varid'>n</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OpDef</span>
+<a name="line-119"></a>    <span class='hs-layout'>{</span> <span class='hs-sel'>_opName</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>n</span>
+<a name="line-120"></a>    <span class='hs-layout'>,</span> <span class='hs-sel'>_opType</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>t</span>
+<a name="line-121"></a>    <span class='hs-layout'>,</span> <span class='hs-sel'>_opAttrs</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>empty</span>
+<a name="line-122"></a>    <span class='hs-layout'>,</span> <span class='hs-sel'>_opInputs</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span>
+<a name="line-123"></a>    <span class='hs-layout'>,</span> <span class='hs-sel'>_opControlInputs</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span>
+<a name="line-124"></a>    <span class='hs-layout'>}</span>
+<a name="line-125"></a>
+<a name="line-126"></a><a name="SummaryTensor"></a><span class='hs-comment'>-- | Synonym for the tensors that return serialized Summary proto.</span>
+<a name="line-127"></a><a name="SummaryTensor"></a><span class='hs-keyword'>type</span> <span class='hs-conid'>SummaryTensor</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>ByteString</span>
+<a name="line-128"></a>
+<a name="line-129"></a><a name="GraphState"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>GraphState</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>GraphState</span>
+<a name="line-130"></a>    <span class='hs-layout'>{</span> <span class='hs-sel'>_renderedNodes</span> <span class='hs-keyglyph'>::</span> <span class='hs-varop'>!</span><span class='hs-layout'>(</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>PendingNode</span> <span class='hs-conid'>NodeDef</span><span class='hs-layout'>)</span>
+<a name="line-131"></a>        <span class='hs-comment'>-- ^ Nodes which have been rendered.  Keeps track of the unique ID we</span>
+<a name="line-132"></a>        <span class='hs-comment'>-- assign each implicitly-named node.  Also prevents us from adding the</span>
+<a name="line-133"></a>        <span class='hs-comment'>-- same node (implicit or explicit) more than once to the nodeBuffer.</span>
+<a name="line-134"></a>    <span class='hs-layout'>,</span> <span class='hs-sel'>_renderedNodeDefs</span> <span class='hs-keyglyph'>::</span> <span class='hs-varop'>!</span><span class='hs-layout'>(</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>NodeName</span> <span class='hs-conid'>NodeDef</span><span class='hs-layout'>)</span>
+<a name="line-135"></a>        <span class='hs-comment'>-- ^ The NodeDefs of nodes which have been rendered. Used by the</span>
+<a name="line-136"></a>        <span class='hs-comment'>-- Gradient module to inspect the node graph.</span>
+<a name="line-137"></a>    <span class='hs-layout'>,</span> <span class='hs-sel'>_nodeBuffer</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>NodeDef</span><span class='hs-keyglyph'>]</span>
+<a name="line-138"></a>        <span class='hs-comment'>-- ^ A list of nodes that should be passed to TensorFlow during</span>
+<a name="line-139"></a>        <span class='hs-comment'>-- the next call to Session.extend (TF_ExtendGraph).</span>
+<a name="line-140"></a>    <span class='hs-layout'>,</span> <span class='hs-sel'>_nextUnique</span> <span class='hs-keyglyph'>::</span> <span class='hs-varop'>!</span><span class='hs-conid'>Unique</span>
+<a name="line-141"></a>        <span class='hs-comment'>-- ^ Unique ID for the next node</span>
+<a name="line-142"></a>    <span class='hs-comment'>-- TODO(judahjacobson): watch for clashes between auto and user names.</span>
+<a name="line-143"></a>    <span class='hs-layout'>,</span> <span class='hs-sel'>_defaultDevice</span> <span class='hs-keyglyph'>::</span> <span class='hs-varop'>!</span><span class='hs-layout'>(</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Device</span><span class='hs-layout'>)</span>
+<a name="line-144"></a>    <span class='hs-layout'>,</span> <span class='hs-sel'>_currentScope</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Scope</span><span class='hs-keyglyph'>]</span>
+<a name="line-145"></a>    <span class='hs-layout'>,</span> <span class='hs-sel'>_defaultControlInputs</span> <span class='hs-keyglyph'>::</span> <span class='hs-varop'>!</span><span class='hs-layout'>(</span><span class='hs-conid'>Set</span> <span class='hs-conid'>NodeName</span><span class='hs-layout'>)</span>
+<a name="line-146"></a>    <span class='hs-layout'>,</span> <span class='hs-sel'>_initializationNodes</span>  <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>NodeName</span><span class='hs-keyglyph'>]</span>
+<a name="line-147"></a>      <span class='hs-comment'>-- ^ The nodes to run next time a TF.run is issued, typically</span>
+<a name="line-148"></a>      <span class='hs-comment'>-- variable initializers.</span>
+<a name="line-149"></a>    <span class='hs-layout'>,</span> <span class='hs-sel'>_summaries</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>SummaryTensor</span><span class='hs-keyglyph'>]</span>
+<a name="line-150"></a>      <span class='hs-comment'>-- ^ The tensors for summary</span>
+<a name="line-151"></a>    <span class='hs-layout'>}</span>
+<a name="line-152"></a>
+<a name="line-153"></a><a name="PendingNode"></a><span class='hs-comment'>-- | A node definition without its final name.  Used as a key in the</span>
+<a name="line-154"></a><a name="PendingNode"></a><span class='hs-comment'>-- "renderedNodes" map.</span>
+<a name="line-155"></a><a name="PendingNode"></a><span class='hs-comment'>-- The NodeDef contained inside has an empty "name" field.</span>
+<a name="line-156"></a><a name="PendingNode"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>PendingNode</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>PendingNode</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Scope</span><span class='hs-keyglyph'>]</span> <span class='hs-varop'>!</span><span class='hs-conid'>PendingNodeName</span> <span class='hs-varop'>!</span><span class='hs-conid'>NodeDef</span>
+<a name="line-157"></a>    <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Eq</span><span class='hs-layout'>,</span> <span class='hs-conid'>Ord</span><span class='hs-layout'>)</span>
+<a name="line-158"></a>
+<a name="line-159"></a><a name="pendingNodeDef"></a><span class='hs-comment'>-- Returns an _incomplete_ NodeDef. The name is fixed by addNewOpFromPending.</span>
+<a name="line-160"></a><span class='hs-definition'>pendingNodeDef</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>PendingNode</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>NodeDef</span>
+<a name="line-161"></a><span class='hs-definition'>pendingNodeDef</span> <span class='hs-layout'>(</span><span class='hs-conid'>PendingNode</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-varid'>n</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>n</span>
+<a name="line-162"></a>
+<a name="line-163"></a><a name="initGraphState"></a><span class='hs-definition'>initGraphState</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>GraphState</span>
+<a name="line-164"></a><span class='hs-definition'>initGraphState</span> <span class='hs-keyglyph'>=</span>
+<a name="line-165"></a>    <span class='hs-conid'>GraphState</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>empty</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>empty</span> <span class='hs-conid'>[]</span> <span class='hs-layout'>(</span><span class='hs-conid'>Unique</span> <span class='hs-num'>0</span><span class='hs-layout'>)</span> <span class='hs-conid'>Nothing</span> <span class='hs-conid'>[]</span> <span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>empty</span> <span class='hs-conid'>[]</span> <span class='hs-conid'>[]</span>
+<a name="line-166"></a>
+<a name="line-167"></a><a name="renderedNodes"></a><span class='hs-definition'>renderedNodes</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>GraphState</span> <span class='hs-layout'>(</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>PendingNode</span> <span class='hs-conid'>NodeDef</span><span class='hs-layout'>)</span>
+<a name="line-168"></a><span class='hs-definition'>renderedNodes</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_renderedNodes</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>g</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>g</span> <span class='hs-layout'>{</span> <span class='hs-sel'>_renderedNodes</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span> <span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-169"></a>
+<a name="line-170"></a><a name="renderedNodeDefs"></a><span class='hs-definition'>renderedNodeDefs</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>GraphState</span> <span class='hs-layout'>(</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>NodeName</span> <span class='hs-conid'>NodeDef</span><span class='hs-layout'>)</span>
+<a name="line-171"></a><span class='hs-definition'>renderedNodeDefs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_renderedNodeDefs</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>g</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>g</span> <span class='hs-layout'>{</span> <span class='hs-sel'>_renderedNodeDefs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span> <span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-172"></a>
+<a name="line-173"></a><a name="nodeBuffer"></a><span class='hs-definition'>nodeBuffer</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>GraphState</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>NodeDef</span><span class='hs-keyglyph'>]</span>
+<a name="line-174"></a><span class='hs-definition'>nodeBuffer</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_nodeBuffer</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>g</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>g</span> <span class='hs-layout'>{</span> <span class='hs-sel'>_nodeBuffer</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span> <span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-175"></a>
+<a name="line-176"></a><a name="nextUnique"></a><span class='hs-definition'>nextUnique</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>GraphState</span> <span class='hs-conid'>Unique</span>
+<a name="line-177"></a><span class='hs-definition'>nextUnique</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_nextUnique</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>g</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>g</span> <span class='hs-layout'>{</span> <span class='hs-sel'>_nextUnique</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span> <span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-178"></a>
+<a name="line-179"></a><a name="defaultDevice"></a><span class='hs-definition'>defaultDevice</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>GraphState</span> <span class='hs-layout'>(</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Device</span><span class='hs-layout'>)</span>
+<a name="line-180"></a><span class='hs-definition'>defaultDevice</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_defaultDevice</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>g</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>g</span> <span class='hs-layout'>{</span> <span class='hs-sel'>_defaultDevice</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span> <span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-181"></a>
+<a name="line-182"></a><a name="currentScope"></a><span class='hs-definition'>currentScope</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>GraphState</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Scope</span><span class='hs-keyglyph'>]</span>
+<a name="line-183"></a><span class='hs-definition'>currentScope</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_currentScope</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>g</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>g</span> <span class='hs-layout'>{</span> <span class='hs-sel'>_currentScope</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span> <span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-184"></a>
+<a name="line-185"></a><a name="defaultControlInputs"></a><span class='hs-definition'>defaultControlInputs</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>GraphState</span> <span class='hs-layout'>(</span><span class='hs-conid'>Set</span> <span class='hs-conid'>NodeName</span><span class='hs-layout'>)</span>
+<a name="line-186"></a><span class='hs-definition'>defaultControlInputs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_defaultControlInputs</span>
+<a name="line-187"></a>                          <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>g</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>g</span> <span class='hs-layout'>{</span> <span class='hs-sel'>_defaultControlInputs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span> <span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-188"></a>
+<a name="line-189"></a><a name="initializationNodes"></a><span class='hs-definition'>initializationNodes</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>GraphState</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>NodeName</span><span class='hs-keyglyph'>]</span>
+<a name="line-190"></a><span class='hs-definition'>initializationNodes</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_initializationNodes</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>g</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>g</span> <span class='hs-layout'>{</span> <span class='hs-sel'>_initializationNodes</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span> <span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-191"></a>
+<a name="line-192"></a><a name="summaries"></a><span class='hs-definition'>summaries</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>GraphState</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>SummaryTensor</span><span class='hs-keyglyph'>]</span>
+<a name="line-193"></a><span class='hs-definition'>summaries</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_summaries</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>g</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>g</span> <span class='hs-layout'>{</span> <span class='hs-sel'>_summaries</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span> <span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-194"></a>
+<a name="line-195"></a><a name="BuildT"></a><span class='hs-comment'>-- | An action for building nodes in a TensorFlow graph.</span>
+<a name="line-196"></a><a name="BuildT"></a><span class='hs-comment'>-- Used to manage build state internally as part of the @Session@ monad.</span>
+<a name="line-197"></a><a name="BuildT"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>BuildT</span> <span class='hs-varid'>m</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>BuildT</span> <span class='hs-layout'>(</span><span class='hs-conid'>StateT</span> <span class='hs-conid'>GraphState</span> <span class='hs-varid'>m</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-198"></a>    <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Functor</span><span class='hs-layout'>,</span> <span class='hs-conid'>Applicative</span><span class='hs-layout'>,</span> <span class='hs-conid'>Monad</span><span class='hs-layout'>,</span> <span class='hs-conid'>MonadIO</span><span class='hs-layout'>,</span> <span class='hs-conid'>MonadTrans</span><span class='hs-layout'>,</span>
+<a name="line-199"></a>              <span class='hs-conid'>MonadState</span> <span class='hs-conid'>GraphState</span><span class='hs-layout'>)</span>
+<a name="line-200"></a>
+<a name="line-201"></a><a name="Build"></a><span class='hs-comment'>-- | An action for building nodes in a TensorFlow graph.</span>
+<a name="line-202"></a><a name="Build"></a><span class='hs-keyword'>type</span> <span class='hs-conid'>Build</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>BuildT</span> <span class='hs-conid'>Identity</span>
+<a name="line-203"></a>
+<a name="line-204"></a><a name="hoistBuildT"></a><span class='hs-comment'>-- | This is Control.Monad.Morph.hoist sans the dependency.</span>
+<a name="line-205"></a><span class='hs-definition'>hoistBuildT</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varop'>.</span> <span class='hs-varid'>m</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>n</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>BuildT</span> <span class='hs-varid'>m</span> <span class='hs-varid'>b</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>BuildT</span> <span class='hs-varid'>n</span> <span class='hs-varid'>b</span>
+<a name="line-206"></a><span class='hs-definition'>hoistBuildT</span> <span class='hs-varid'>f</span> <span class='hs-layout'>(</span><span class='hs-conid'>BuildT</span> <span class='hs-varid'>m</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>BuildT</span> <span class='hs-varop'>$</span> <span class='hs-varid'>mapStateT</span> <span class='hs-varid'>f</span> <span class='hs-varid'>m</span>
+<a name="line-207"></a>
+<a name="line-208"></a><a name="runBuildT"></a><span class='hs-definition'>runBuildT</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>BuildT</span> <span class='hs-varid'>m</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>m</span> <span class='hs-layout'>(</span><span class='hs-varid'>a</span><span class='hs-layout'>,</span> <span class='hs-conid'>GraphState</span><span class='hs-layout'>)</span>
+<a name="line-209"></a><span class='hs-definition'>runBuildT</span> <span class='hs-layout'>(</span><span class='hs-conid'>BuildT</span> <span class='hs-varid'>f</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>runStateT</span> <span class='hs-varid'>f</span> <span class='hs-varid'>initGraphState</span>
+<a name="line-210"></a>
+<a name="line-211"></a><a name="evalBuildT"></a><span class='hs-definition'>evalBuildT</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Monad</span> <span class='hs-varid'>m</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>BuildT</span> <span class='hs-varid'>m</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>m</span> <span class='hs-varid'>a</span>
+<a name="line-212"></a><span class='hs-definition'>evalBuildT</span> <span class='hs-layout'>(</span><span class='hs-conid'>BuildT</span> <span class='hs-varid'>f</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>evalStateT</span> <span class='hs-varid'>f</span> <span class='hs-varid'>initGraphState</span>
+<a name="line-213"></a>
+<a name="line-214"></a><a name="flushNodeBuffer"></a><span class='hs-comment'>-- | Get all the NodeDefs that have accumulated so far, and clear that buffer.</span>
+<a name="line-215"></a><span class='hs-definition'>flushNodeBuffer</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Monad</span> <span class='hs-varid'>m</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>BuildT</span> <span class='hs-varid'>m</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>NodeDef</span><span class='hs-keyglyph'>]</span>
+<a name="line-216"></a><span class='hs-definition'>flushNodeBuffer</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-217"></a>    <span class='hs-varid'>ns</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>use</span> <span class='hs-varid'>nodeBuffer</span>
+<a name="line-218"></a>    <span class='hs-varid'>nodeBuffer</span> <span class='hs-varop'>.=</span> <span class='hs-conid'>[]</span>
+<a name="line-219"></a>    <span class='hs-varid'>return</span> <span class='hs-varid'>ns</span>
+<a name="line-220"></a>
+<a name="line-221"></a><a name="flushInitializers"></a><span class='hs-comment'>-- | Get all the initializers that have accumulated so far, and clear</span>
+<a name="line-222"></a><span class='hs-comment'>-- that buffer.</span>
+<a name="line-223"></a><span class='hs-definition'>flushInitializers</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Monad</span> <span class='hs-varid'>m</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>BuildT</span> <span class='hs-varid'>m</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>NodeName</span><span class='hs-keyglyph'>]</span>
+<a name="line-224"></a><span class='hs-definition'>flushInitializers</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-225"></a>    <span class='hs-varid'>ns</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>use</span> <span class='hs-varid'>initializationNodes</span>
+<a name="line-226"></a>    <span class='hs-varid'>initializationNodes</span> <span class='hs-varop'>.=</span> <span class='hs-conid'>[]</span>
+<a name="line-227"></a>    <span class='hs-varid'>return</span> <span class='hs-varid'>ns</span>
+<a name="line-228"></a>
+<a name="line-229"></a><a name="addInitializer"></a><span class='hs-comment'>-- | Registers the given node to be executed before the next</span>
+<a name="line-230"></a><span class='hs-comment'>-- 'TensorFlow.Session.run'.</span>
+<a name="line-231"></a><span class='hs-definition'>addInitializer</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>ControlNode</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-conid'>()</span>
+<a name="line-232"></a><span class='hs-definition'>addInitializer</span> <span class='hs-layout'>(</span><span class='hs-conid'>ControlNode</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-233"></a>    <span class='hs-varid'>i</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>getOrAddOp</span> <span class='hs-varid'>o</span>
+<a name="line-234"></a>    <span class='hs-varid'>initializationNodes</span> <span class='hs-varop'>%=</span> <span class='hs-layout'>(</span><span class='hs-varid'>i</span><span class='hs-conop'>:</span><span class='hs-layout'>)</span>
+<a name="line-235"></a>
+<a name="line-236"></a><a name="asGraphDef"></a><span class='hs-comment'>-- | Produce a GraphDef proto representation of the nodes that are rendered in</span>
+<a name="line-237"></a><span class='hs-comment'>-- the given 'Build' action.</span>
+<a name="line-238"></a><span class='hs-definition'>asGraphDef</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Build</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>GraphDef</span>
+<a name="line-239"></a><span class='hs-definition'>asGraphDef</span> <span class='hs-varid'>b</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>def</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>node</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>gs</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>nodeBuffer</span>
+<a name="line-240"></a>  <span class='hs-keyword'>where</span>
+<a name="line-241"></a>    <span class='hs-varid'>gs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>snd</span> <span class='hs-varop'>$</span> <span class='hs-varid'>runIdentity</span> <span class='hs-varop'>$</span> <span class='hs-varid'>runBuildT</span> <span class='hs-varid'>b</span>
+<a name="line-242"></a>
+<a name="line-243"></a><a name="addGraphDef"></a><span class='hs-comment'>-- TODO: check against existing nodes for conflicts?</span>
+<a name="line-244"></a><span class='hs-definition'>addGraphDef</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>GraphDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-conid'>()</span>
+<a name="line-245"></a><span class='hs-definition'>addGraphDef</span> <span class='hs-varid'>g</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>nodeBuffer</span> <span class='hs-varop'>&lt;&gt;=</span> <span class='hs-varid'>g</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>node</span>
+<a name="line-246"></a>
+<a name="line-247"></a><a name="getOrAddOp"></a><span class='hs-comment'>-- | Render the given op if it hasn't been rendered already, and return its</span>
+<a name="line-248"></a><span class='hs-comment'>-- name.</span>
+<a name="line-249"></a><span class='hs-definition'>getOrAddOp</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Op</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-conid'>NodeName</span>
+<a name="line-250"></a><span class='hs-definition'>getOrAddOp</span> <span class='hs-varid'>o</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NodeName</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-varop'>^.</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>resolveOp</span> <span class='hs-varid'>o</span>
+<a name="line-251"></a>
+<a name="line-252"></a><a name="resolveOp"></a><span class='hs-definition'>resolveOp</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Op</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-conid'>NodeDef</span>
+<a name="line-253"></a><span class='hs-definition'>resolveOp</span> <span class='hs-layout'>(</span><span class='hs-conid'>Rendered</span> <span class='hs-varid'>n</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>return</span> <span class='hs-varid'>n</span>
+<a name="line-254"></a><span class='hs-definition'>resolveOp</span> <span class='hs-layout'>(</span><span class='hs-conid'>Unrendered</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-255"></a>    <span class='hs-varid'>pending</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>getPendingNode</span> <span class='hs-varid'>o</span>
+<a name="line-256"></a>    <span class='hs-varid'>uses</span> <span class='hs-varid'>renderedNodes</span> <span class='hs-layout'>(</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>lookup</span> <span class='hs-varid'>pending</span><span class='hs-layout'>)</span> <span class='hs-varop'>&gt;&gt;=</span> <span class='hs-keyglyph'>\</span><span class='hs-keyword'>case</span>
+<a name="line-257"></a>        <span class='hs-conid'>Just</span> <span class='hs-varid'>n</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>return</span> <span class='hs-varid'>n</span>
+<a name="line-258"></a>        <span class='hs-conid'>Nothing</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>addNewOpFromPending</span> <span class='hs-varid'>pending</span>
+<a name="line-259"></a>
+<a name="line-260"></a><a name="addNewOp"></a><span class='hs-comment'>-- | Add a new node for a given 'OpDef'.  This is used for making "stateful" ops</span>
+<a name="line-261"></a><span class='hs-comment'>-- which are not safe to dedup (e.g, "variable" and "assign").</span>
+<a name="line-262"></a><span class='hs-definition'>addNewOp</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-conid'>NodeDef</span>
+<a name="line-263"></a><span class='hs-definition'>addNewOp</span> <span class='hs-varid'>o</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>getPendingNode</span> <span class='hs-varid'>o</span> <span class='hs-varop'>&gt;&gt;=</span> <span class='hs-varid'>addNewOpFromPending</span>
+<a name="line-264"></a>
+<a name="line-265"></a><a name="addNewOpFromPending"></a><span class='hs-definition'>addNewOpFromPending</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>PendingNode</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-conid'>NodeDef</span>
+<a name="line-266"></a><span class='hs-definition'>addNewOpFromPending</span> <span class='hs-varid'>pending</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-267"></a>    <span class='hs-varid'>nodeName</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>renderPendingNode</span> <span class='hs-varid'>pending</span>
+<a name="line-268"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>nodeDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>pendingNodeDef</span> <span class='hs-varid'>pending</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>name</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>unNodeName</span> <span class='hs-varid'>nodeName</span>
+<a name="line-269"></a>    <span class='hs-varid'>nodeBuffer</span> <span class='hs-varop'>%=</span> <span class='hs-layout'>(</span><span class='hs-varid'>nodeDef</span> <span class='hs-conop'>:</span><span class='hs-layout'>)</span>
+<a name="line-270"></a>    <span class='hs-varid'>renderedNodes</span> <span class='hs-varop'>%=</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>insert</span> <span class='hs-varid'>pending</span> <span class='hs-varid'>nodeDef</span>
+<a name="line-271"></a>    <span class='hs-varid'>renderedNodeDefs</span> <span class='hs-varop'>%=</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>insert</span> <span class='hs-varid'>nodeName</span> <span class='hs-varid'>nodeDef</span>
+<a name="line-272"></a>    <span class='hs-varid'>return</span> <span class='hs-varid'>nodeDef</span>
+<a name="line-273"></a>
+<a name="line-274"></a><a name="getPendingNode"></a><span class='hs-comment'>-- | Get the pending node corresponding to an OpDef, which may or may not have</span>
+<a name="line-275"></a><span class='hs-comment'>-- been rendered before.  Implicitly renders all of this node's inputs.</span>
+<a name="line-276"></a><span class='hs-definition'>getPendingNode</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-conid'>PendingNode</span>
+<a name="line-277"></a><span class='hs-definition'>getPendingNode</span> <span class='hs-varid'>o</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-278"></a>    <span class='hs-comment'>-- An empty string in the proto field means that no specific</span>
+<a name="line-279"></a>    <span class='hs-comment'>-- device is specified.</span>
+<a name="line-280"></a>    <span class='hs-varid'>dev</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>maybe</span> <span class='hs-str'>""</span> <span class='hs-varid'>deviceName</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>use</span> <span class='hs-varid'>defaultDevice</span>
+<a name="line-281"></a>    <span class='hs-varid'>inputs</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>mapM</span> <span class='hs-varid'>getInput</span> <span class='hs-layout'>(</span><span class='hs-varid'>o</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>opInputs</span><span class='hs-layout'>)</span>
+<a name="line-282"></a>    <span class='hs-varid'>scope</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>use</span> <span class='hs-varid'>currentScope</span>
+<a name="line-283"></a>    <span class='hs-varid'>controls</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>use</span> <span class='hs-varid'>defaultControlInputs</span>
+<a name="line-284"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>controlInputs</span>
+<a name="line-285"></a>            <span class='hs-keyglyph'>=</span> <span class='hs-varid'>map</span> <span class='hs-varid'>getDep</span> <span class='hs-layout'>(</span><span class='hs-varid'>o</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>opControlInputs</span> <span class='hs-varop'>++</span> <span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>toList</span> <span class='hs-varid'>controls</span><span class='hs-layout'>)</span>
+<a name="line-286"></a>    <span class='hs-varid'>return</span> <span class='hs-varop'>$</span> <span class='hs-conid'>PendingNode</span> <span class='hs-varid'>scope</span> <span class='hs-layout'>(</span><span class='hs-varid'>o</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>opName</span><span class='hs-layout'>)</span>
+<a name="line-287"></a>            <span class='hs-varop'>$</span> <span class='hs-varid'>def</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>op</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>unOpType</span> <span class='hs-layout'>(</span><span class='hs-varid'>o</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>opType</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-288"></a>                  <span class='hs-varop'>&amp;</span> <span class='hs-varid'>attr</span> <span class='hs-varop'>.~</span> <span class='hs-sel'>_opAttrs</span> <span class='hs-varid'>o</span>
+<a name="line-289"></a>                  <span class='hs-varop'>&amp;</span> <span class='hs-varid'>input</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>inputs</span> <span class='hs-varop'>++</span> <span class='hs-varid'>controlInputs</span><span class='hs-layout'>)</span>
+<a name="line-290"></a>                  <span class='hs-varop'>&amp;</span> <span class='hs-varid'>device</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>dev</span>
+<a name="line-291"></a>  <span class='hs-keyword'>where</span>
+<a name="line-292"></a>    <span class='hs-varid'>getInput</span> <span class='hs-layout'>(</span><span class='hs-conid'>Output</span> <span class='hs-layout'>(</span><span class='hs-conid'>OutputIx</span> <span class='hs-varid'>k</span><span class='hs-layout'>)</span> <span class='hs-varid'>subOp</span><span class='hs-layout'>)</span>
+<a name="line-293"></a>        <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-varop'>&lt;&gt;</span> <span class='hs-str'>":"</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>pack</span> <span class='hs-layout'>(</span><span class='hs-varid'>show</span> <span class='hs-varid'>k</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-varop'>.</span> <span class='hs-varid'>unNodeName</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>getOrAddOp</span> <span class='hs-varid'>subOp</span>
+<a name="line-294"></a>    <span class='hs-varid'>getDep</span> <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-str'>"^"</span> <span class='hs-varop'>&lt;&gt;</span><span class='hs-layout'>)</span> <span class='hs-varop'>.</span> <span class='hs-varid'>unNodeName</span>
+<a name="line-295"></a>
+<a name="line-296"></a><a name="renderPendingNode"></a><span class='hs-comment'>-- | Pick a name for a pending node.  If it has an explicit name, just use that;</span>
+<a name="line-297"></a><span class='hs-comment'>-- if the name is implicit, assign a new unique name based on the op type.</span>
+<a name="line-298"></a><span class='hs-definition'>renderPendingNode</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>PendingNode</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-conid'>NodeName</span>
+<a name="line-299"></a><span class='hs-definition'>renderPendingNode</span> <span class='hs-layout'>(</span><span class='hs-conid'>PendingNode</span> <span class='hs-varid'>scope</span> <span class='hs-varid'>pendingName</span> <span class='hs-varid'>nodeDef</span><span class='hs-layout'>)</span>
+<a name="line-300"></a>    <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NodeName</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-varid'>scopePrefix</span> <span class='hs-varop'>&lt;&gt;</span><span class='hs-layout'>)</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>getName</span>
+<a name="line-301"></a>  <span class='hs-keyword'>where</span>
+<a name="line-302"></a>    <span class='hs-varid'>scopePrefix</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>concat</span> <span class='hs-varop'>$</span> <span class='hs-varid'>fmap</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varop'>&lt;&gt;</span> <span class='hs-str'>"/"</span><span class='hs-layout'>)</span> <span class='hs-varop'>.</span> <span class='hs-varid'>unScope</span><span class='hs-layout'>)</span> <span class='hs-varid'>scope</span>
+<a name="line-303"></a>    <span class='hs-varid'>getName</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>case</span> <span class='hs-varid'>pendingName</span> <span class='hs-keyword'>of</span>
+<a name="line-304"></a>        <span class='hs-conid'>ExplicitName</span> <span class='hs-varid'>n</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>return</span> <span class='hs-varid'>n</span>
+<a name="line-305"></a>        <span class='hs-conid'>ImplicitName</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyword'>do</span>
+<a name="line-306"></a>            <span class='hs-varid'>u</span><span class='hs-keyglyph'>@</span><span class='hs-layout'>(</span><span class='hs-conid'>Unique</span> <span class='hs-varid'>k</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>use</span> <span class='hs-varid'>nextUnique</span>
+<a name="line-307"></a>            <span class='hs-varid'>nextUnique</span> <span class='hs-varop'>.=</span> <span class='hs-varid'>succ</span> <span class='hs-varid'>u</span>
+<a name="line-308"></a>            <span class='hs-varid'>return</span> <span class='hs-varop'>$</span> <span class='hs-varid'>nodeDef</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>op</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-str'>"_"</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>pack</span> <span class='hs-layout'>(</span><span class='hs-varid'>show</span> <span class='hs-varid'>k</span><span class='hs-layout'>)</span>
+<a name="line-309"></a>
+<a name="line-310"></a>
+<a name="line-311"></a><a name="renderOutput"></a><span class='hs-comment'>-- | Render an 'Output' and return a string representation for the TensorFlow</span>
+<a name="line-312"></a><span class='hs-comment'>-- foreign APIs.</span>
+<a name="line-313"></a><span class='hs-definition'>renderOutput</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Output</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-conid'>Text</span>
+<a name="line-314"></a><span class='hs-definition'>renderOutput</span> <span class='hs-layout'>(</span><span class='hs-conid'>Output</span> <span class='hs-layout'>(</span><span class='hs-conid'>OutputIx</span> <span class='hs-varid'>i</span><span class='hs-layout'>)</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-315"></a>    <span class='hs-varid'>n</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>getOrAddOp</span> <span class='hs-varid'>o</span>
+<a name="line-316"></a>    <span class='hs-varid'>return</span> <span class='hs-varop'>$</span> <span class='hs-varid'>unNodeName</span> <span class='hs-varid'>n</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>pack</span> <span class='hs-layout'>(</span><span class='hs-str'>":"</span> <span class='hs-varop'>++</span> <span class='hs-varid'>show</span> <span class='hs-varid'>i</span><span class='hs-layout'>)</span>
+<a name="line-317"></a>
+<a name="line-318"></a><a name="withStateLens"></a><span class='hs-comment'>-- | Modify some part of the state, run an action, and restore the state</span>
+<a name="line-319"></a><span class='hs-comment'>-- after that action is done.</span>
+<a name="line-320"></a><span class='hs-definition'>withStateLens</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>MonadState</span> <span class='hs-varid'>s</span> <span class='hs-varid'>m</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Lens'</span> <span class='hs-varid'>s</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>m</span> <span class='hs-varid'>b</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>m</span> <span class='hs-varid'>b</span>
+<a name="line-321"></a><span class='hs-definition'>withStateLens</span> <span class='hs-varid'>accessor</span> <span class='hs-varid'>f</span> <span class='hs-varid'>act</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-322"></a>    <span class='hs-varid'>old</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>use</span> <span class='hs-varid'>accessor</span>
+<a name="line-323"></a>    <span class='hs-varid'>accessor</span> <span class='hs-varop'>%=</span> <span class='hs-varid'>f</span>
+<a name="line-324"></a>    <span class='hs-varid'>result</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>act</span>
+<a name="line-325"></a>    <span class='hs-varid'>accessor</span> <span class='hs-varop'>.=</span> <span class='hs-varid'>old</span>
+<a name="line-326"></a>    <span class='hs-varid'>return</span> <span class='hs-varid'>result</span>
+<a name="line-327"></a>
+<a name="line-328"></a><a name="withDevice"></a><span class='hs-comment'>-- | Set a device for all nodes rendered in the given 'Build' action</span>
+<a name="line-329"></a><span class='hs-comment'>-- (unless further overridden by another use of withDevice).</span>
+<a name="line-330"></a><span class='hs-definition'>withDevice</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Maybe</span> <span class='hs-conid'>Device</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-varid'>a</span>
+<a name="line-331"></a><span class='hs-definition'>withDevice</span> <span class='hs-varid'>d</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>withStateLens</span> <span class='hs-varid'>defaultDevice</span> <span class='hs-layout'>(</span><span class='hs-varid'>const</span> <span class='hs-varid'>d</span><span class='hs-layout'>)</span>
+<a name="line-332"></a>
+<a name="line-333"></a><a name="colocateWith"></a><span class='hs-comment'>-- | Places all nodes rendered in the given 'Build' action on the same</span>
+<a name="line-334"></a><span class='hs-comment'>-- device as the given Tensor (see also 'withDevice'). Make sure that</span>
+<a name="line-335"></a><span class='hs-comment'>-- the action has side effects of rendering the desired tensors. A pure</span>
+<a name="line-336"></a><span class='hs-comment'>-- return would not have the desired effect.</span>
+<a name="line-337"></a><span class='hs-definition'>colocateWith</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varid'>v</span> <span class='hs-varid'>b</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>b</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-varid'>a</span>
+<a name="line-338"></a><span class='hs-definition'>colocateWith</span> <span class='hs-varid'>t</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-339"></a>    <span class='hs-varid'>d</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>Device</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-varop'>^.</span> <span class='hs-varid'>device</span><span class='hs-layout'>)</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>resolveOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>t</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>tensorOutput</span> <span class='hs-varop'>.</span> <span class='hs-varid'>outputOp</span><span class='hs-layout'>)</span>
+<a name="line-340"></a>    <span class='hs-varid'>withDevice</span> <span class='hs-layout'>(</span><span class='hs-conid'>Just</span> <span class='hs-varid'>d</span><span class='hs-layout'>)</span> <span class='hs-varid'>x</span>
+<a name="line-341"></a>
+<a name="line-342"></a><a name="withNameScope"></a><span class='hs-comment'>-- | Prepend a scope to all nodes rendered in the given 'Build' action.</span>
+<a name="line-343"></a><span class='hs-definition'>withNameScope</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-varid'>a</span>
+<a name="line-344"></a><span class='hs-definition'>withNameScope</span> <span class='hs-varid'>s</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>withStateLens</span> <span class='hs-varid'>currentScope</span> <span class='hs-layout'>(</span><span class='hs-conid'>Scope</span> <span class='hs-varid'>s</span> <span class='hs-conop'>:</span><span class='hs-layout'>)</span>
+<a name="line-345"></a>
+<a name="line-346"></a><a name="withNodeDependencies"></a><span class='hs-comment'>-- | Add control inputs to all nodes rendered in the given 'Build' action.</span>
+<a name="line-347"></a><span class='hs-definition'>withNodeDependencies</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Set</span> <span class='hs-conid'>NodeName</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-varid'>a</span>
+<a name="line-348"></a><span class='hs-definition'>withNodeDependencies</span> <span class='hs-varid'>nodes</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>withStateLens</span> <span class='hs-varid'>defaultControlInputs</span> <span class='hs-layout'>(</span><span class='hs-varop'>&lt;&gt;</span> <span class='hs-varid'>nodes</span><span class='hs-layout'>)</span>
+<a name="line-349"></a>
+<a name="line-350"></a><a name="render"></a><span class='hs-comment'>-- | Render a 'Tensor', fixing its name, scope, device and control inputs from</span>
+<a name="line-351"></a><span class='hs-comment'>-- the 'Build' context.  Also renders any dependencies of the 'Tensor' that</span>
+<a name="line-352"></a><span class='hs-comment'>-- weren't already rendered.</span>
+<a name="line-353"></a><span class='hs-comment'>--</span>
+<a name="line-354"></a><span class='hs-comment'>-- This operation is idempotent; @render &gt;=&gt; render === render@.  However,</span>
+<a name="line-355"></a><span class='hs-comment'>-- rendering a (previously un-rendered) 'Tensor' in two different contexts</span>
+<a name="line-356"></a><span class='hs-comment'>-- may result in two different 'Tensor's.</span>
+<a name="line-357"></a><span class='hs-definition'>render</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-358"></a><span class='hs-definition'>render</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tensorOutput</span> <span class='hs-varop'>$</span> <span class='hs-varid'>outputOp</span> <span class='hs-varop'>$</span> <span class='hs-varid'>fmap</span> <span class='hs-conid'>Rendered</span> <span class='hs-varop'>.</span> <span class='hs-varid'>resolveOp</span>
+<a name="line-359"></a>
+<a name="line-360"></a><a name="renderNodeName"></a><span class='hs-comment'>-- | Render a 'Tensor' and get its node's name.</span>
+<a name="line-361"></a><span class='hs-definition'>renderNodeName</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-conid'>NodeName</span>
+<a name="line-362"></a><span class='hs-definition'>renderNodeName</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>getOrAddOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>t</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>tensorOutput</span> <span class='hs-varop'>.</span> <span class='hs-varid'>outputOp</span><span class='hs-layout'>)</span>
+<a name="line-363"></a>
+<a name="line-364"></a><a name="addSummary"></a><span class='hs-comment'>-- | Records the given summary action in Build for retrieval with</span>
+<a name="line-365"></a><span class='hs-comment'>-- 'collectAllSummaries'. The summary op is required to produce a</span>
+<a name="line-366"></a><span class='hs-comment'>-- Summary protocol buffer in string form. For safety, use the</span>
+<a name="line-367"></a><span class='hs-comment'>-- pre-composed functions: Logging.scalarSummary and</span>
+<a name="line-368"></a><span class='hs-comment'>-- Logging.histogramSummary.</span>
+<a name="line-369"></a><span class='hs-definition'>addSummary</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>SummaryTensor</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-conid'>()</span>
+<a name="line-370"></a><span class='hs-definition'>addSummary</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>summaries</span> <span class='hs-varop'>%=</span> <span class='hs-layout'>(</span><span class='hs-varid'>t</span> <span class='hs-conop'>:</span><span class='hs-layout'>)</span>
+<a name="line-371"></a>
+<a name="line-372"></a><a name="collectAllSummaries"></a><span class='hs-comment'>-- | Retrieves the summary ops collected thus far. Typically this only</span>
+<a name="line-373"></a><span class='hs-comment'>-- happens once, but if 'TensorFlow.Session.buildWithSummary' is used</span>
+<a name="line-374"></a><span class='hs-comment'>-- repeatedly, the values accumulate.</span>
+<a name="line-375"></a><span class='hs-definition'>collectAllSummaries</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Monad</span> <span class='hs-varid'>m</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>BuildT</span> <span class='hs-varid'>m</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>SummaryTensor</span><span class='hs-keyglyph'>]</span>
+<a name="line-376"></a><span class='hs-definition'>collectAllSummaries</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>use</span> <span class='hs-varid'>summaries</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-BuildOp.html b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-BuildOp.html
new file mode 100644
index 0000000..ff74848
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-BuildOp.html
@@ -0,0 +1,210 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/BuildOp.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE FlexibleInstances #-}</span>
+<a name="line-16"></a><span class='hs-comment'>{-# LANGUAGE TupleSections #-}</span>
+<a name="line-17"></a>
+<a name="line-18"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>BuildOp</span>
+<a name="line-19"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>OpResult</span>
+<a name="line-20"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>BuildOp</span>
+<a name="line-21"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>buildOp</span>
+<a name="line-22"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>buildListOp</span>
+<a name="line-23"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>eqLengthGuard</span>
+<a name="line-24"></a>    <span class='hs-layout'>)</span>
+<a name="line-25"></a>  <span class='hs-keyword'>where</span>
+<a name="line-26"></a>
+<a name="line-27"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Monad</span> <span class='hs-layout'>(</span><span class='hs-varid'>replicateM</span><span class='hs-layout'>)</span>
+<a name="line-28"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Monad</span><span class='hs-varop'>.</span><span class='hs-conid'>Reader</span> <span class='hs-layout'>(</span><span class='hs-conid'>ReaderT</span><span class='hs-layout'>,</span> <span class='hs-varid'>runReaderT</span><span class='hs-layout'>,</span> <span class='hs-varid'>ask</span><span class='hs-layout'>)</span>
+<a name="line-29"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Monad</span><span class='hs-varop'>.</span><span class='hs-conid'>State</span><span class='hs-varop'>.</span><span class='hs-conid'>Strict</span> <span class='hs-layout'>(</span><span class='hs-conid'>State</span><span class='hs-layout'>,</span> <span class='hs-varid'>runState</span><span class='hs-layout'>,</span> <span class='hs-varid'>get</span><span class='hs-layout'>,</span> <span class='hs-varid'>put</span><span class='hs-layout'>)</span>
+<a name="line-30"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span> <span class='hs-layout'>(</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-31"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varop'>&amp;</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>&lt;&gt;~</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>^.</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-32"></a>
+<a name="line-33"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Build</span>
+<a name="line-34"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Output</span>
+<a name="line-35"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span>
+<a name="line-36"></a>
+<a name="line-37"></a><a name="ResultState"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>ResultState</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>ResultState</span> <span class='hs-varop'>!</span><span class='hs-conid'>OutputIx</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-keyword'>deriving</span> <span class='hs-conid'>Show</span>
+<a name="line-38"></a>
+<a name="line-39"></a><a name="Result"></a><span class='hs-keyword'>type</span> <span class='hs-conid'>Result</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>ReaderT</span> <span class='hs-conid'>Op</span> <span class='hs-layout'>(</span><span class='hs-conid'>State</span> <span class='hs-conid'>ResultState</span><span class='hs-layout'>)</span>
+<a name="line-40"></a>
+<a name="line-41"></a><a name="OpResult"></a><span class='hs-comment'>-- | Class of types that can be used as op outputs.</span>
+<a name="line-42"></a><a name="OpResult"></a><span class='hs-keyword'>class</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a</span> <span class='hs-keyword'>where</span>
+<a name="line-43"></a>    <span class='hs-varid'>toResult</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Result</span> <span class='hs-varid'>a</span>
+<a name="line-44"></a>
+<a name="line-45"></a><a name="instance%20OpResult%20(a1,%20a2)"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span><span class='hs-conid'>OpResult</span> <span class='hs-varid'>a1</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a2</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>OpResult</span> <span class='hs-layout'>(</span><span class='hs-varid'>a1</span><span class='hs-layout'>,</span> <span class='hs-varid'>a2</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-46"></a>    <span class='hs-varid'>toResult</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>(,)</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>toResult</span> <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>toResult</span>
+<a name="line-47"></a>
+<a name="line-48"></a><a name="instance%20OpResult%20(a1,%20a2,%20a3)"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span><span class='hs-conid'>OpResult</span> <span class='hs-varid'>a1</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a2</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a3</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>OpResult</span> <span class='hs-layout'>(</span><span class='hs-varid'>a1</span><span class='hs-layout'>,</span> <span class='hs-varid'>a2</span><span class='hs-layout'>,</span> <span class='hs-varid'>a3</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-49"></a>    <span class='hs-varid'>toResult</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>(,,)</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>toResult</span> <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>toResult</span> <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>toResult</span>
+<a name="line-50"></a>
+<a name="line-51"></a><a name="instance%20OpResult%20(a1,%20a2,%20a3,%20a4)"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span><span class='hs-conid'>OpResult</span> <span class='hs-varid'>a1</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a2</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a3</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a4</span><span class='hs-layout'>)</span>
+<a name="line-52"></a>         <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>OpResult</span> <span class='hs-layout'>(</span><span class='hs-varid'>a1</span><span class='hs-layout'>,</span> <span class='hs-varid'>a2</span><span class='hs-layout'>,</span> <span class='hs-varid'>a3</span><span class='hs-layout'>,</span> <span class='hs-varid'>a4</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-53"></a>    <span class='hs-varid'>toResult</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>(,,,)</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>toResult</span> <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>toResult</span> <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>toResult</span> <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>toResult</span>
+<a name="line-54"></a>
+<a name="line-55"></a><a name="instance%20OpResult%20(a1,%20a2,%20a3,%20a4,%20a5)"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span><span class='hs-conid'>OpResult</span> <span class='hs-varid'>a1</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a2</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a3</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a4</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a5</span><span class='hs-layout'>)</span>
+<a name="line-56"></a>         <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>OpResult</span> <span class='hs-layout'>(</span><span class='hs-varid'>a1</span><span class='hs-layout'>,</span> <span class='hs-varid'>a2</span><span class='hs-layout'>,</span> <span class='hs-varid'>a3</span><span class='hs-layout'>,</span> <span class='hs-varid'>a4</span><span class='hs-layout'>,</span> <span class='hs-varid'>a5</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-57"></a>    <span class='hs-varid'>toResult</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>(,,,,)</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>toResult</span>
+<a name="line-58"></a>                      <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>toResult</span>
+<a name="line-59"></a>                      <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>toResult</span>
+<a name="line-60"></a>                      <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>toResult</span>
+<a name="line-61"></a>                      <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>toResult</span>
+<a name="line-62"></a>
+<a name="line-63"></a><a name="instance%20OpResult%20(a1,%20a2,%20a3,%20a4,%20a5,%20a6)"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a1</span>
+<a name="line-64"></a>         <span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a2</span>
+<a name="line-65"></a>         <span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a3</span>
+<a name="line-66"></a>         <span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a4</span>
+<a name="line-67"></a>         <span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a5</span>
+<a name="line-68"></a>         <span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a6</span>
+<a name="line-69"></a>         <span class='hs-layout'>)</span>
+<a name="line-70"></a>         <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>OpResult</span> <span class='hs-layout'>(</span><span class='hs-varid'>a1</span><span class='hs-layout'>,</span> <span class='hs-varid'>a2</span><span class='hs-layout'>,</span> <span class='hs-varid'>a3</span><span class='hs-layout'>,</span> <span class='hs-varid'>a4</span><span class='hs-layout'>,</span> <span class='hs-varid'>a5</span><span class='hs-layout'>,</span> <span class='hs-varid'>a6</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-71"></a>    <span class='hs-varid'>toResult</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>(,,,,,)</span>
+<a name="line-72"></a>               <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>toResult</span>
+<a name="line-73"></a>               <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>toResult</span>
+<a name="line-74"></a>               <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>toResult</span>
+<a name="line-75"></a>               <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>toResult</span>
+<a name="line-76"></a>               <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>toResult</span>
+<a name="line-77"></a>               <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>toResult</span>
+<a name="line-78"></a>
+<a name="line-79"></a><a name="tensorResult"></a><span class='hs-definition'>tensorResult</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>TensorKind</span> <span class='hs-varid'>v</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Result</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-80"></a><span class='hs-definition'>tensorResult</span> <span class='hs-varid'>v</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-81"></a>    <span class='hs-varid'>o</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>ask</span>
+<a name="line-82"></a>    <span class='hs-conid'>ResultState</span> <span class='hs-varid'>i</span> <span class='hs-varid'>ns</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>get</span>
+<a name="line-83"></a>    <span class='hs-varid'>put</span> <span class='hs-varop'>$!</span> <span class='hs-conid'>ResultState</span> <span class='hs-layout'>(</span><span class='hs-varid'>i</span><span class='hs-varop'>+</span><span class='hs-num'>1</span><span class='hs-layout'>)</span> <span class='hs-varid'>ns</span>
+<a name="line-84"></a>    <span class='hs-varid'>return</span> <span class='hs-varop'>$!</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varop'>$</span> <span class='hs-varid'>output</span> <span class='hs-varid'>i</span> <span class='hs-varid'>o</span>
+<a name="line-85"></a>
+<a name="line-86"></a><a name="instance%20OpResult%20(Tensor%20Value%20a)"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>OpResult</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-87"></a>    <span class='hs-varid'>toResult</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tensorResult</span> <span class='hs-conid'>ValueKind</span>
+<a name="line-88"></a>
+<a name="line-89"></a><a name="instance%20OpResult%20(Tensor%20Ref%20a)"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>OpResult</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Ref</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-90"></a>    <span class='hs-varid'>toResult</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tensorResult</span> <span class='hs-conid'>RefKind</span>
+<a name="line-91"></a>
+<a name="line-92"></a><a name="instance%20OpResult%20ControlNode"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>OpResult</span> <span class='hs-conid'>ControlNode</span> <span class='hs-keyword'>where</span>
+<a name="line-93"></a>    <span class='hs-varid'>toResult</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>ControlNode</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>ask</span>
+<a name="line-94"></a>
+<a name="line-95"></a><a name="instance%20OpResult%20%5ba%5d"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>OpResult</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span> <span class='hs-keyword'>where</span>
+<a name="line-96"></a>    <span class='hs-varid'>toResult</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-97"></a>        <span class='hs-conid'>ResultState</span> <span class='hs-varid'>i</span> <span class='hs-varid'>ns</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>get</span>
+<a name="line-98"></a>        <span class='hs-keyword'>case</span> <span class='hs-varid'>ns</span> <span class='hs-keyword'>of</span>
+<a name="line-99"></a>            <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>error</span> <span class='hs-varop'>$</span> <span class='hs-str'>"Ran out of counts in toResult. "</span> <span class='hs-varop'>++</span>
+<a name="line-100"></a>                          <span class='hs-str'>"Likely misuse of buildListOp."</span>
+<a name="line-101"></a>            <span class='hs-layout'>(</span><span class='hs-varid'>n</span> <span class='hs-conop'>:</span> <span class='hs-varid'>rest</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyword'>do</span>
+<a name="line-102"></a>                <span class='hs-varid'>put</span> <span class='hs-varop'>$!</span> <span class='hs-conid'>ResultState</span> <span class='hs-varid'>i</span> <span class='hs-varid'>rest</span>
+<a name="line-103"></a>                <span class='hs-varid'>replicateM</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-varid'>n</span><span class='hs-layout'>)</span> <span class='hs-varid'>toResult</span>
+<a name="line-104"></a>
+<a name="line-105"></a><a name="runResult"></a><span class='hs-definition'>runResult</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Op</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>a</span>
+<a name="line-106"></a><span class='hs-definition'>runResult</span> <span class='hs-varid'>ns</span> <span class='hs-varid'>o</span> <span class='hs-keyglyph'>=</span>
+<a name="line-107"></a>    <span class='hs-keyword'>case</span> <span class='hs-varid'>runState</span> <span class='hs-layout'>(</span><span class='hs-varid'>runReaderT</span> <span class='hs-varid'>toResult</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-conid'>ResultState</span> <span class='hs-num'>0</span> <span class='hs-varid'>ns</span><span class='hs-layout'>)</span> <span class='hs-keyword'>of</span>
+<a name="line-108"></a>        <span class='hs-layout'>(</span><span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-conid'>ResultState</span> <span class='hs-keyword'>_</span> <span class='hs-conid'>[]</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span>
+<a name="line-109"></a>        <span class='hs-layout'>(</span><span class='hs-keyword'>_</span><span class='hs-layout'>,</span> <span class='hs-varid'>ns'</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>error</span> <span class='hs-varop'>$</span> <span class='hs-str'>"Ununsed length in runResult attributes: "</span> <span class='hs-varop'>++</span>
+<a name="line-110"></a>                            <span class='hs-varid'>show</span> <span class='hs-layout'>(</span><span class='hs-varid'>ns</span><span class='hs-layout'>,</span> <span class='hs-varid'>ns'</span><span class='hs-layout'>)</span>
+<a name="line-111"></a>
+<a name="line-112"></a><a name="pureResult"></a><span class='hs-comment'>-- | Make a new "pure" op, which may be deduped with identical ops within</span>
+<a name="line-113"></a><span class='hs-comment'>-- the same scope.</span>
+<a name="line-114"></a><span class='hs-definition'>pureResult</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Output</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>a</span>
+<a name="line-115"></a><span class='hs-definition'>pureResult</span> <span class='hs-varid'>ns</span> <span class='hs-varid'>o</span> <span class='hs-varid'>ts</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>runResult</span> <span class='hs-varid'>ns</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Unrendered</span> <span class='hs-varop'>$</span> <span class='hs-varid'>addReversedInputs</span> <span class='hs-varid'>o</span> <span class='hs-varid'>ts</span>
+<a name="line-116"></a>
+<a name="line-117"></a><a name="buildResult"></a><span class='hs-comment'>-- | Make a new "stateful" op, which will not be deduped with otherwise</span>
+<a name="line-118"></a><span class='hs-comment'>-- identical ops.</span>
+<a name="line-119"></a><span class='hs-definition'>buildResult</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Output</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-varid'>a</span>
+<a name="line-120"></a><span class='hs-definition'>buildResult</span> <span class='hs-varid'>ns</span> <span class='hs-varid'>o</span> <span class='hs-varid'>ts</span>
+<a name="line-121"></a>    <span class='hs-keyglyph'>=</span> <span class='hs-varid'>runResult</span> <span class='hs-varid'>ns</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Rendered</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>addNewOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>addReversedInputs</span> <span class='hs-varid'>o</span> <span class='hs-varid'>ts</span><span class='hs-layout'>)</span>
+<a name="line-122"></a>
+<a name="line-123"></a><a name="addReversedInputs"></a><span class='hs-definition'>addReversedInputs</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Output</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>OpDef</span>
+<a name="line-124"></a><span class='hs-definition'>addReversedInputs</span> <span class='hs-varid'>o</span> <span class='hs-varid'>ts</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>o</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opInputs</span> <span class='hs-varop'>&lt;&gt;~</span> <span class='hs-varid'>reverse</span> <span class='hs-varid'>ts</span>
+<a name="line-125"></a>
+<a name="line-126"></a><a name="OpDef"></a><span class='hs-comment'>-- | Class of types that can be used as op functions.</span>
+<a name="line-127"></a><a name="OpDef"></a><span class='hs-keyword'>class</span> <span class='hs-conid'>BuildOp</span> <span class='hs-varid'>f</span> <span class='hs-keyword'>where</span>
+<a name="line-128"></a>    <span class='hs-varid'>buildOp'</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span>  <span class='hs-comment'>-- ^ Sizes of list results (having number_attr)</span>
+<a name="line-129"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>OpDef</span>
+<a name="line-130"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Output</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ Accumulator for inputs to the op.</span>
+<a name="line-131"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>f</span>
+<a name="line-132"></a>
+<a name="line-133"></a><a name="buildOp"></a><span class='hs-comment'>-- | Starts an operation that returns a structured set of tensors</span>
+<a name="line-134"></a><span class='hs-comment'>-- (singletons or tuples).</span>
+<a name="line-135"></a><span class='hs-definition'>buildOp</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>BuildOp</span> <span class='hs-varid'>f</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>f</span>
+<a name="line-136"></a><span class='hs-definition'>buildOp</span> <span class='hs-varid'>o</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>buildOp'</span> <span class='hs-conid'>[]</span> <span class='hs-varid'>o</span> <span class='hs-conid'>[]</span>
+<a name="line-137"></a>
+<a name="line-138"></a><a name="buildListOp"></a><span class='hs-comment'>-- | Starts an operation that returns a list of tensors.</span>
+<a name="line-139"></a><span class='hs-definition'>buildListOp</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>BuildOp</span> <span class='hs-varid'>f</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span>
+<a name="line-140"></a>               <span class='hs-comment'>-- ^ Cardinality of the corresponding list of tensors output.</span>
+<a name="line-141"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>f</span>
+<a name="line-142"></a><span class='hs-definition'>buildListOp</span> <span class='hs-varid'>counts</span> <span class='hs-varid'>o</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>buildOp'</span> <span class='hs-varid'>counts</span> <span class='hs-varid'>o</span> <span class='hs-conid'>[]</span>
+<a name="line-143"></a>
+<a name="line-144"></a><a name="instance%20BuildOp%20ControlNode"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>BuildOp</span> <span class='hs-conid'>ControlNode</span> <span class='hs-keyword'>where</span>
+<a name="line-145"></a>    <span class='hs-varid'>buildOp'</span> <span class='hs-keyword'>_</span> <span class='hs-varid'>o</span> <span class='hs-varid'>ts</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>ControlNode</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Unrendered</span> <span class='hs-varop'>$</span> <span class='hs-varid'>addReversedInputs</span> <span class='hs-varid'>o</span> <span class='hs-varid'>ts</span>
+<a name="line-146"></a>
+<a name="line-147"></a><a name="instance%20BuildOp%20(Tensor%20Value%20a)"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>BuildOp</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-148"></a>    <span class='hs-varid'>buildOp'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>pureResult</span>
+<a name="line-149"></a>
+<a name="line-150"></a><a name="instance%20BuildOp%20(Tensor%20Ref%20a)"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>BuildOp</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Ref</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-151"></a>    <span class='hs-varid'>buildOp'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>pureResult</span>
+<a name="line-152"></a>
+<a name="line-153"></a><a name="instance%20BuildOp%20%5bTensor%20Value%20a%5d"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>BuildOp</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span> <span class='hs-keyword'>where</span>
+<a name="line-154"></a>    <span class='hs-varid'>buildOp'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>pureResult</span>
+<a name="line-155"></a>
+<a name="line-156"></a><a name="instance%20BuildOp%20(t1,%20t2)"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span><span class='hs-conid'>OpResult</span> <span class='hs-varid'>t1</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>t2</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>BuildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>t1</span><span class='hs-layout'>,</span> <span class='hs-varid'>t2</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-157"></a>    <span class='hs-varid'>buildOp'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>pureResult</span>
+<a name="line-158"></a>
+<a name="line-159"></a><a name="instance%20BuildOp%20(t1,%20t2,%20t3)"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span><span class='hs-conid'>OpResult</span> <span class='hs-varid'>t1</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>t2</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>t3</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>BuildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>t1</span><span class='hs-layout'>,</span> <span class='hs-varid'>t2</span><span class='hs-layout'>,</span> <span class='hs-varid'>t3</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-160"></a>    <span class='hs-varid'>buildOp'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>pureResult</span>
+<a name="line-161"></a>
+<a name="line-162"></a><a name="instance%20BuildOp%20(t1,%20t2,%20t3,%20t4)"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span><span class='hs-conid'>OpResult</span> <span class='hs-varid'>t1</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>t2</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>t3</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>t4</span><span class='hs-layout'>)</span>
+<a name="line-163"></a>         <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>BuildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>t1</span><span class='hs-layout'>,</span> <span class='hs-varid'>t2</span><span class='hs-layout'>,</span> <span class='hs-varid'>t3</span><span class='hs-layout'>,</span> <span class='hs-varid'>t4</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-164"></a>    <span class='hs-varid'>buildOp'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>pureResult</span>
+<a name="line-165"></a>
+<a name="line-166"></a><a name="instance%20BuildOp%20(t1,%20t2,%20t3,%20t4,%20t5)"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span><span class='hs-conid'>OpResult</span> <span class='hs-varid'>t1</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>t2</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>t3</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>t4</span><span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>t5</span><span class='hs-layout'>)</span>
+<a name="line-167"></a>         <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>BuildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>t1</span><span class='hs-layout'>,</span> <span class='hs-varid'>t2</span><span class='hs-layout'>,</span> <span class='hs-varid'>t3</span><span class='hs-layout'>,</span> <span class='hs-varid'>t4</span><span class='hs-layout'>,</span> <span class='hs-varid'>t5</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-168"></a>    <span class='hs-varid'>buildOp'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>pureResult</span>
+<a name="line-169"></a>
+<a name="line-170"></a><a name="instance%20BuildOp%20(t1,%20t2,%20t3,%20t4,%20t5,%20t6)"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>t1</span>
+<a name="line-171"></a>         <span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>t2</span>
+<a name="line-172"></a>         <span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>t3</span>
+<a name="line-173"></a>         <span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>t4</span>
+<a name="line-174"></a>         <span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>t5</span>
+<a name="line-175"></a>         <span class='hs-layout'>,</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>t6</span>
+<a name="line-176"></a>         <span class='hs-layout'>)</span>
+<a name="line-177"></a>         <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>BuildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>t1</span><span class='hs-layout'>,</span> <span class='hs-varid'>t2</span><span class='hs-layout'>,</span> <span class='hs-varid'>t3</span><span class='hs-layout'>,</span> <span class='hs-varid'>t4</span><span class='hs-layout'>,</span> <span class='hs-varid'>t5</span><span class='hs-layout'>,</span> <span class='hs-varid'>t6</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-178"></a>    <span class='hs-varid'>buildOp'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>pureResult</span>
+<a name="line-179"></a>
+<a name="line-180"></a><a name="instance%20BuildOp%20(Build%20a)"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>OpResult</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>BuildOp</span> <span class='hs-layout'>(</span><span class='hs-conid'>Build</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-181"></a>    <span class='hs-varid'>buildOp'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>buildResult</span>
+<a name="line-182"></a>
+<a name="line-183"></a><a name="instance%20BuildOp%20(Tensor%20v%20a%20-%3e%20f)"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>BuildOp</span> <span class='hs-varid'>f</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>BuildOp</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>f</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-184"></a>    <span class='hs-varid'>buildOp'</span> <span class='hs-varid'>rf</span> <span class='hs-varid'>o</span> <span class='hs-varid'>ts</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>buildOp'</span> <span class='hs-varid'>rf</span> <span class='hs-varid'>o</span> <span class='hs-layout'>(</span><span class='hs-varid'>t</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>tensorOutput</span> <span class='hs-conop'>:</span> <span class='hs-varid'>ts</span><span class='hs-layout'>)</span>
+<a name="line-185"></a>
+<a name="line-186"></a><a name="instance%20BuildOp%20(%5bTensor%20v%20a%5d%20-%3e%20f)"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>BuildOp</span> <span class='hs-varid'>f</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>BuildOp</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>f</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-187"></a>    <span class='hs-varid'>buildOp'</span> <span class='hs-varid'>rf</span> <span class='hs-varid'>o</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>ts</span>
+<a name="line-188"></a>        <span class='hs-keyglyph'>=</span> <span class='hs-varid'>buildOp'</span> <span class='hs-varid'>rf</span> <span class='hs-varid'>o</span> <span class='hs-layout'>(</span><span class='hs-varid'>reverse</span> <span class='hs-layout'>(</span><span class='hs-varid'>fmap</span> <span class='hs-layout'>(</span><span class='hs-varop'>^.</span> <span class='hs-varid'>tensorOutput</span><span class='hs-layout'>)</span> <span class='hs-varid'>ts</span><span class='hs-layout'>)</span> <span class='hs-varop'>++</span> <span class='hs-varid'>accum</span><span class='hs-layout'>)</span>
+<a name="line-189"></a>
+<a name="line-190"></a><a name="eqLengthGuard"></a><span class='hs-comment'>-- | Returns true if all the integers in each tuple are identical.</span>
+<a name="line-191"></a><span class='hs-comment'>-- Throws an error with a descriptive message if not.</span>
+<a name="line-192"></a><span class='hs-definition'>eqLengthGuard</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>String</span><span class='hs-layout'>,</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>String</span><span class='hs-layout'>,</span> <span class='hs-conid'>Int</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Bool</span>
+<a name="line-193"></a><span class='hs-definition'>eqLengthGuard</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>all</span> <span class='hs-varid'>eachOk</span>
+<a name="line-194"></a>  <span class='hs-keyword'>where</span>
+<a name="line-195"></a>    <span class='hs-varid'>eachOk</span> <span class='hs-layout'>(</span><span class='hs-keyword'>_</span><span class='hs-layout'>,</span> <span class='hs-conid'>[]</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>True</span>
+<a name="line-196"></a>    <span class='hs-comment'>-- The next line has (== 1) . length . nub in disguise</span>
+<a name="line-197"></a>    <span class='hs-varid'>eachOk</span> <span class='hs-layout'>(</span><span class='hs-varid'>numberAttrName</span><span class='hs-layout'>,</span> <span class='hs-varid'>pairs</span><span class='hs-keyglyph'>@</span><span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-keyword'>_</span><span class='hs-layout'>,</span> <span class='hs-varid'>x</span><span class='hs-layout'>)</span> <span class='hs-conop'>:</span> <span class='hs-varid'>zs</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>all</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>z</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>snd</span> <span class='hs-varid'>z</span> <span class='hs-varop'>==</span> <span class='hs-varid'>x</span><span class='hs-layout'>)</span> <span class='hs-varid'>zs</span> <span class='hs-varop'>||</span>
+<a name="line-198"></a>        <span class='hs-varid'>error</span> <span class='hs-layout'>(</span><span class='hs-str'>"number_attr "</span> <span class='hs-varop'>++</span> <span class='hs-varid'>numberAttrName</span> <span class='hs-varop'>++</span>
+<a name="line-199"></a>               <span class='hs-str'>" contains tensors with different length "</span> <span class='hs-varop'>++</span> <span class='hs-varid'>show</span> <span class='hs-varid'>pairs</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-ControlFlow.html b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-ControlFlow.html
new file mode 100644
index 0000000..091a8cf
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-ControlFlow.html
@@ -0,0 +1,98 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/ControlFlow.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE GADTs #-}</span>
+<a name="line-16"></a><span class='hs-comment'>{-# LANGUAGE OverloadedStrings #-}</span>
+<a name="line-17"></a><span class='hs-comment'>{-# LANGUAGE RankNTypes #-}</span>
+<a name="line-18"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables #-}</span>
+<a name="line-19"></a>
+<a name="line-20"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>ControlFlow</span>
+<a name="line-21"></a>    <span class='hs-layout'>(</span> <span class='hs-comment'>-- * Dependencies</span>
+<a name="line-22"></a>      <span class='hs-varid'>withControlDependencies</span>
+<a name="line-23"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>group</span>
+<a name="line-24"></a>      <span class='hs-comment'>-- * Operations</span>
+<a name="line-25"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>identity</span>
+<a name="line-26"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>noOp</span>
+<a name="line-27"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>named</span>
+<a name="line-28"></a>    <span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-29"></a>
+<a name="line-30"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Set</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Set</span>
+<a name="line-31"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-layout'>(</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-32"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varop'>&amp;</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>^.</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>.~</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-33"></a>
+<a name="line-34"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>BuildOp</span>
+<a name="line-35"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Build</span>
+<a name="line-36"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Nodes</span>
+<a name="line-37"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Output</span>
+<a name="line-38"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span>
+<a name="line-39"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span>
+<a name="line-40"></a>
+<a name="line-41"></a><a name="withControlDependencies"></a><span class='hs-comment'>-- | Modify a 'Build' action, such that all new ops rendered in it will depend</span>
+<a name="line-42"></a><span class='hs-comment'>-- on the nodes in the first argument.</span>
+<a name="line-43"></a><span class='hs-definition'>withControlDependencies</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Nodes</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-varid'>a</span>
+<a name="line-44"></a><span class='hs-definition'>withControlDependencies</span> <span class='hs-varid'>deps</span> <span class='hs-varid'>act</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-45"></a>    <span class='hs-varid'>nodes</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>getNodes</span> <span class='hs-varid'>deps</span>
+<a name="line-46"></a>    <span class='hs-varid'>withNodeDependencies</span> <span class='hs-varid'>nodes</span> <span class='hs-varid'>act</span>
+<a name="line-47"></a>
+<a name="line-48"></a><span class='hs-comment'>-- TODO(judahjacobson): Reimplement withDependencies.</span>
+<a name="line-49"></a>
+<a name="line-50"></a><a name="group"></a><span class='hs-comment'>-- | Create an op that groups multiple operations.</span>
+<a name="line-51"></a><span class='hs-comment'>--</span>
+<a name="line-52"></a><span class='hs-comment'>-- When this op finishes, all ops in the input @n@ have finished.  This op has</span>
+<a name="line-53"></a><span class='hs-comment'>-- no output.</span>
+<a name="line-54"></a><span class='hs-definition'>group</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Nodes</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-55"></a><span class='hs-definition'>group</span> <span class='hs-varid'>deps</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-56"></a>    <span class='hs-varid'>nodes</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>toList</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>getNodes</span> <span class='hs-varid'>deps</span>
+<a name="line-57"></a>    <span class='hs-comment'>-- TODO: slicker way</span>
+<a name="line-58"></a>    <span class='hs-varid'>return</span> <span class='hs-varop'>$</span> <span class='hs-varid'>buildOp</span> <span class='hs-varop'>$</span> <span class='hs-varid'>opDef</span> <span class='hs-str'>"NoOp"</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opControlInputs</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>nodes</span>
+<a name="line-59"></a>
+<a name="line-60"></a>
+<a name="line-61"></a><a name="identity"></a><span class='hs-comment'>-- | Returns a 'Tensor' with the same shape and contents as the input.</span>
+<a name="line-62"></a><span class='hs-definition'>identity</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span>
+<a name="line-63"></a><span class='hs-definition'>identity</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>namedIdentity</span> <span class='hs-varid'>implicitName</span>
+<a name="line-64"></a>
+<a name="line-65"></a><a name="named"></a><span class='hs-comment'>-- | Returns a 'Tensor' with a given name and the same shape and contents as</span>
+<a name="line-66"></a><span class='hs-comment'>-- the input.</span>
+<a name="line-67"></a><span class='hs-comment'>--</span>
+<a name="line-68"></a><span class='hs-comment'>-- TODO(judahjacobson): This breaks when used with uninitialize @Tensor Ref@s,</span>
+<a name="line-69"></a><span class='hs-comment'>-- since @RefIdentity@ doesn't have SetAllowsUninitializedInput().  Look into</span>
+<a name="line-70"></a><span class='hs-comment'>-- whether we can change that op.</span>
+<a name="line-71"></a><span class='hs-definition'>named</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span>
+<a name="line-72"></a><span class='hs-definition'>named</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>namedIdentity</span> <span class='hs-varop'>.</span> <span class='hs-varid'>explicitName</span>
+<a name="line-73"></a>
+<a name="line-74"></a><a name="namedIdentity"></a><span class='hs-comment'>-- | An internal version of "identity" that allows setting the name</span>
+<a name="line-75"></a><span class='hs-comment'>-- of the output Tensor.</span>
+<a name="line-76"></a><span class='hs-definition'>namedIdentity</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varid'>v</span> <span class='hs-varop'>.</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span>
+<a name="line-77"></a>              <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>PendingNodeName</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span>
+<a name="line-78"></a><span class='hs-definition'>namedIdentity</span> <span class='hs-varid'>n</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>case</span> <span class='hs-varid'>t</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>tensorKind</span> <span class='hs-keyword'>of</span>
+<a name="line-79"></a>                      <span class='hs-conid'>ValueKind</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDefWithName</span> <span class='hs-varid'>n</span> <span class='hs-str'>"Identity"</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>setTypeAttr</span><span class='hs-layout'>)</span> <span class='hs-varid'>t</span>
+<a name="line-80"></a>                      <span class='hs-conid'>RefKind</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDefWithName</span> <span class='hs-varid'>n</span> <span class='hs-str'>"RefIdentity"</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>setTypeAttr</span><span class='hs-layout'>)</span> <span class='hs-varid'>t</span>
+<a name="line-81"></a>  <span class='hs-keyword'>where</span>
+<a name="line-82"></a>    <span class='hs-varid'>setTypeAttr</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-83"></a>
+<a name="line-84"></a>
+<a name="line-85"></a><a name="noOp"></a><span class='hs-comment'>-- | Does nothing.  Only useful as a placeholder for control edges.</span>
+<a name="line-86"></a><span class='hs-definition'>noOp</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-87"></a><span class='hs-definition'>noOp</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>buildOp</span> <span class='hs-varop'>$</span> <span class='hs-varid'>opDef</span> <span class='hs-str'>"NoOp"</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Internal-FFI.html b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Internal-FFI.html
new file mode 100644
index 0000000..9ed15ea
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Internal-FFI.html
@@ -0,0 +1,254 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/Internal/FFI.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE DeriveDataTypeable #-}</span>
+<a name="line-16"></a><span class='hs-comment'>{-# LANGUAGE OverloadedStrings #-}</span>
+<a name="line-17"></a>
+<a name="line-18"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Internal</span><span class='hs-varop'>.</span><span class='hs-conid'>FFI</span>
+<a name="line-19"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>TensorFlowException</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-20"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-conid'>Session</span>
+<a name="line-21"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>withSession</span>
+<a name="line-22"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>extendGraph</span>
+<a name="line-23"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>run</span>
+<a name="line-24"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>TensorData</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-25"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>setSessionConfig</span>
+<a name="line-26"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>setSessionTarget</span>
+<a name="line-27"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>getAllOpList</span>
+<a name="line-28"></a>      <span class='hs-comment'>-- * Internal helper.</span>
+<a name="line-29"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>useProtoAsVoidPtrLen</span>
+<a name="line-30"></a>    <span class='hs-layout'>)</span>
+<a name="line-31"></a>    <span class='hs-keyword'>where</span>
+<a name="line-32"></a>
+<a name="line-33"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Concurrent</span><span class='hs-varop'>.</span><span class='hs-conid'>Async</span> <span class='hs-layout'>(</span><span class='hs-conid'>Async</span><span class='hs-layout'>,</span> <span class='hs-varid'>async</span><span class='hs-layout'>,</span> <span class='hs-varid'>cancel</span><span class='hs-layout'>,</span> <span class='hs-varid'>waitCatch</span><span class='hs-layout'>)</span>
+<a name="line-34"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Concurrent</span><span class='hs-varop'>.</span><span class='hs-conid'>MVar</span> <span class='hs-layout'>(</span><span class='hs-conid'>MVar</span><span class='hs-layout'>,</span> <span class='hs-varid'>modifyMVarMasked_</span><span class='hs-layout'>,</span> <span class='hs-varid'>newMVar</span><span class='hs-layout'>,</span> <span class='hs-varid'>takeMVar</span><span class='hs-layout'>)</span>
+<a name="line-35"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Exception</span> <span class='hs-layout'>(</span><span class='hs-conid'>Exception</span><span class='hs-layout'>,</span> <span class='hs-varid'>throwIO</span><span class='hs-layout'>,</span> <span class='hs-varid'>bracket</span><span class='hs-layout'>,</span> <span class='hs-varid'>finally</span><span class='hs-layout'>,</span> <span class='hs-varid'>mask_</span><span class='hs-layout'>)</span>
+<a name="line-36"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Monad</span> <span class='hs-layout'>(</span><span class='hs-varid'>when</span><span class='hs-layout'>)</span>
+<a name="line-37"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span> <span class='hs-layout'>(</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-38"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Typeable</span> <span class='hs-layout'>(</span><span class='hs-conid'>Typeable</span><span class='hs-layout'>)</span>
+<a name="line-39"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span> <span class='hs-layout'>(</span><span class='hs-conid'>Word8</span><span class='hs-layout'>)</span>
+<a name="line-40"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Foreign</span> <span class='hs-layout'>(</span><span class='hs-conid'>Ptr</span><span class='hs-layout'>,</span> <span class='hs-conid'>FunPtr</span><span class='hs-layout'>,</span> <span class='hs-varid'>nullPtr</span><span class='hs-layout'>,</span> <span class='hs-varid'>castPtr</span><span class='hs-layout'>)</span>
+<a name="line-41"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Foreign</span><span class='hs-varop'>.</span><span class='hs-conid'>C</span><span class='hs-varop'>.</span><span class='hs-conid'>String</span> <span class='hs-layout'>(</span><span class='hs-conid'>CString</span><span class='hs-layout'>)</span>
+<a name="line-42"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Foreign</span><span class='hs-varop'>.</span><span class='hs-conid'>ForeignPtr</span> <span class='hs-layout'>(</span><span class='hs-varid'>newForeignPtr</span><span class='hs-layout'>,</span> <span class='hs-varid'>withForeignPtr</span><span class='hs-layout'>)</span>
+<a name="line-43"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Foreign</span><span class='hs-varop'>.</span><span class='hs-conid'>Marshal</span><span class='hs-varop'>.</span><span class='hs-conid'>Alloc</span> <span class='hs-layout'>(</span><span class='hs-varid'>free</span><span class='hs-layout'>)</span>
+<a name="line-44"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Foreign</span><span class='hs-varop'>.</span><span class='hs-conid'>Marshal</span><span class='hs-varop'>.</span><span class='hs-conid'>Array</span> <span class='hs-layout'>(</span><span class='hs-varid'>withArrayLen</span><span class='hs-layout'>,</span> <span class='hs-varid'>peekArray</span><span class='hs-layout'>,</span> <span class='hs-varid'>mallocArray</span><span class='hs-layout'>,</span> <span class='hs-varid'>copyArray</span><span class='hs-layout'>)</span>
+<a name="line-45"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>System</span><span class='hs-varop'>.</span><span class='hs-conid'>IO</span><span class='hs-varop'>.</span><span class='hs-conid'>Unsafe</span> <span class='hs-layout'>(</span><span class='hs-varid'>unsafePerformIO</span><span class='hs-layout'>)</span>
+<a name="line-46"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>B</span>
+<a name="line-47"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>T</span>
+<a name="line-48"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Encoding</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>T</span>
+<a name="line-49"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Encoding</span><span class='hs-varop'>.</span><span class='hs-conid'>Error</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>T</span>
+<a name="line-50"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Vector</span><span class='hs-varop'>.</span><span class='hs-conid'>Storable</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>S</span>
+<a name="line-51"></a>
+<a name="line-52"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span> <span class='hs-layout'>(</span><span class='hs-conid'>Message</span><span class='hs-layout'>,</span> <span class='hs-varid'>encodeMessage</span><span class='hs-layout'>)</span>
+<a name="line-53"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Graph</span> <span class='hs-layout'>(</span><span class='hs-conid'>GraphDef</span><span class='hs-layout'>)</span>
+<a name="line-54"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span> <span class='hs-layout'>(</span><span class='hs-conid'>DataType</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-55"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Protobuf</span><span class='hs-varop'>.</span><span class='hs-conid'>Config</span> <span class='hs-layout'>(</span><span class='hs-conid'>ConfigProto</span><span class='hs-layout'>)</span>
+<a name="line-56"></a>
+<a name="line-57"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Internal</span><span class='hs-varop'>.</span><span class='hs-conid'>Raw</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Raw</span>
+<a name="line-58"></a>
+<a name="line-59"></a><a name="TensorFlowException"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>TensorFlowException</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TensorFlowException</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-conid'>Code</span> <span class='hs-conid'>T</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-60"></a>    <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Eq</span><span class='hs-layout'>,</span> <span class='hs-conid'>Typeable</span><span class='hs-layout'>)</span>
+<a name="line-61"></a>
+<a name="line-62"></a><a name="instance%20Exception%20TensorFlowException"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Exception</span> <span class='hs-conid'>TensorFlowException</span>
+<a name="line-63"></a>
+<a name="line-64"></a><a name="TensorData"></a><span class='hs-comment'>-- | All of the data needed to represent a tensor.</span>
+<a name="line-65"></a><a name="TensorData"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>TensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TensorData</span>
+<a name="line-66"></a>    <span class='hs-layout'>{</span> <span class='hs-varid'>tensorDataDimensions</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span>
+<a name="line-67"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>tensorDataType</span>       <span class='hs-keyglyph'>::</span> <span class='hs-varop'>!</span><span class='hs-conid'>DataType</span>
+<a name="line-68"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>tensorDataBytes</span>      <span class='hs-keyglyph'>::</span> <span class='hs-varop'>!</span><span class='hs-layout'>(</span><span class='hs-conid'>S</span><span class='hs-varop'>.</span><span class='hs-conid'>Vector</span> <span class='hs-conid'>Word8</span><span class='hs-layout'>)</span>
+<a name="line-69"></a>    <span class='hs-layout'>}</span>
+<a name="line-70"></a>  <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-71"></a>
+<a name="line-72"></a><a name="withSession"></a><span class='hs-comment'>-- | Runs the given action after creating a session with options</span>
+<a name="line-73"></a><span class='hs-comment'>-- populated by the given optionSetter.</span>
+<a name="line-74"></a><span class='hs-definition'>withSession</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-conid'>SessionOptions</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span>
+<a name="line-75"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>()</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-conid'>Session</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-76"></a>            <span class='hs-comment'>-- ^ The action can spawn concurrent tasks which will</span>
+<a name="line-77"></a>            <span class='hs-comment'>-- be canceled before withSession returns.</span>
+<a name="line-78"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span>
+<a name="line-79"></a><span class='hs-definition'>withSession</span> <span class='hs-varid'>optionSetter</span> <span class='hs-varid'>action</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-80"></a>    <span class='hs-varid'>drain</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>newMVar</span> <span class='hs-conid'>[]</span>
+<a name="line-81"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>cleanup</span> <span class='hs-varid'>s</span> <span class='hs-keyglyph'>=</span>
+<a name="line-82"></a>        <span class='hs-comment'>-- Closes the session to nudge the pending run calls to fail and exit.</span>
+<a name="line-83"></a>            <span class='hs-varid'>finally</span> <span class='hs-layout'>(</span><span class='hs-varid'>checkStatus</span> <span class='hs-layout'>(</span><span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>closeSession</span> <span class='hs-varid'>s</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-varop'>$</span> <span class='hs-keyword'>do</span>
+<a name="line-84"></a>                <span class='hs-varid'>runners</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>takeMVar</span> <span class='hs-varid'>drain</span>
+<a name="line-85"></a>                <span class='hs-comment'>-- Collects all runners before deleting the session.</span>
+<a name="line-86"></a>                <span class='hs-varid'>mapM_</span> <span class='hs-varid'>shutDownRunner</span> <span class='hs-varid'>runners</span>
+<a name="line-87"></a>                <span class='hs-varid'>checkStatus</span> <span class='hs-layout'>(</span><span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>deleteSession</span> <span class='hs-varid'>s</span><span class='hs-layout'>)</span>
+<a name="line-88"></a>    <span class='hs-varid'>bracket</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>newSessionOptions</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>deleteSessionOptions</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-varid'>options</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyword'>do</span>
+<a name="line-89"></a>        <span class='hs-varid'>optionSetter</span> <span class='hs-varid'>options</span>
+<a name="line-90"></a>        <span class='hs-varid'>bracket</span>
+<a name="line-91"></a>            <span class='hs-layout'>(</span><span class='hs-varid'>checkStatus</span> <span class='hs-layout'>(</span><span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>newSession</span> <span class='hs-varid'>options</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-92"></a>            <span class='hs-varid'>cleanup</span>
+<a name="line-93"></a>            <span class='hs-layout'>(</span><span class='hs-varid'>action</span> <span class='hs-layout'>(</span><span class='hs-varid'>asyncCollector</span> <span class='hs-varid'>drain</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-94"></a>
+<a name="line-95"></a><a name="asyncCollector"></a><span class='hs-definition'>asyncCollector</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>MVar</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Async</span> <span class='hs-conid'>()</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-96"></a><span class='hs-definition'>asyncCollector</span> <span class='hs-varid'>drain</span> <span class='hs-varid'>runner</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>modifyMVarMasked_</span> <span class='hs-varid'>drain</span> <span class='hs-varid'>launchAndRecord</span>
+<a name="line-97"></a>    <span class='hs-keyword'>where</span>
+<a name="line-98"></a>      <span class='hs-varid'>launchAndRecord</span> <span class='hs-varid'>restRunners</span> <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conop'>:</span> <span class='hs-varid'>restRunners</span><span class='hs-layout'>)</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>async</span> <span class='hs-varid'>runner</span>
+<a name="line-99"></a>
+<a name="line-100"></a><a name="shutDownRunner"></a><span class='hs-definition'>shutDownRunner</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Async</span> <span class='hs-conid'>()</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-101"></a><span class='hs-definition'>shutDownRunner</span> <span class='hs-varid'>r</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-102"></a>    <span class='hs-varid'>cancel</span> <span class='hs-varid'>r</span>
+<a name="line-103"></a>    <span class='hs-comment'>-- TODO(gnezdo): manage exceptions better than print.</span>
+<a name="line-104"></a>    <span class='hs-varid'>either</span> <span class='hs-varid'>print</span> <span class='hs-layout'>(</span><span class='hs-varid'>const</span> <span class='hs-layout'>(</span><span class='hs-varid'>return</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-varop'>=&lt;&lt;</span> <span class='hs-varid'>waitCatch</span> <span class='hs-varid'>r</span>
+<a name="line-105"></a>
+<a name="line-106"></a><a name="extendGraph"></a><span class='hs-definition'>extendGraph</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-conid'>Session</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>GraphDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-107"></a><span class='hs-definition'>extendGraph</span> <span class='hs-varid'>session</span> <span class='hs-varid'>pb</span> <span class='hs-keyglyph'>=</span>
+<a name="line-108"></a>    <span class='hs-varid'>useProtoAsVoidPtrLen</span> <span class='hs-varid'>pb</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-varid'>ptr</span> <span class='hs-varid'>len</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-109"></a>        <span class='hs-varid'>checkStatus</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>extendGraph</span> <span class='hs-varid'>session</span> <span class='hs-varid'>ptr</span> <span class='hs-varid'>len</span>
+<a name="line-110"></a>
+<a name="line-111"></a>
+<a name="line-112"></a><a name="run"></a><span class='hs-definition'>run</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-conid'>Session</span>
+<a name="line-113"></a>    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorData</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ Feeds.</span>
+<a name="line-114"></a>    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-keyglyph'>]</span>               <span class='hs-comment'>-- ^ Fetches.</span>
+<a name="line-115"></a>    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-keyglyph'>]</span>               <span class='hs-comment'>-- ^ Targets.</span>
+<a name="line-116"></a>    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>TensorData</span><span class='hs-keyglyph'>]</span>
+<a name="line-117"></a><span class='hs-definition'>run</span> <span class='hs-varid'>session</span> <span class='hs-varid'>feeds</span> <span class='hs-varid'>fetches</span> <span class='hs-varid'>targets</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-118"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>nullTensor</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>nullPtr</span>
+<a name="line-119"></a>    <span class='hs-comment'>-- Use mask to avoid leaking input tensors before they are passed to 'run'</span>
+<a name="line-120"></a>    <span class='hs-comment'>-- and output tensors before they are passed to 'createTensorData'.</span>
+<a name="line-121"></a>    <span class='hs-varid'>mask_</span> <span class='hs-varop'>$</span>
+<a name="line-122"></a>        <span class='hs-comment'>-- Feeds</span>
+<a name="line-123"></a>        <span class='hs-varid'>withStringArrayLen</span> <span class='hs-layout'>(</span><span class='hs-varid'>fst</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>feeds</span><span class='hs-layout'>)</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-varid'>feedsLen</span> <span class='hs-varid'>feedNames</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-124"></a>        <span class='hs-varid'>mapM</span> <span class='hs-layout'>(</span><span class='hs-varid'>createRawTensor</span> <span class='hs-varop'>.</span> <span class='hs-varid'>snd</span><span class='hs-layout'>)</span> <span class='hs-varid'>feeds</span> <span class='hs-varop'>&gt;&gt;=</span> <span class='hs-keyglyph'>\</span><span class='hs-varid'>feedTensors</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-125"></a>        <span class='hs-varid'>withArrayLen</span> <span class='hs-varid'>feedTensors</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-keyword'>_</span> <span class='hs-varid'>cFeedTensors</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-126"></a>        <span class='hs-comment'>-- Fetches.</span>
+<a name="line-127"></a>        <span class='hs-varid'>withStringArrayLen</span> <span class='hs-varid'>fetches</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-varid'>fetchesLen</span> <span class='hs-varid'>fetchNames</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-128"></a>        <span class='hs-comment'>-- tensorOuts is an array of null Tensor pointers that will be filled</span>
+<a name="line-129"></a>        <span class='hs-comment'>-- by the call to Raw.run.</span>
+<a name="line-130"></a>        <span class='hs-varid'>withArrayLen</span> <span class='hs-layout'>(</span><span class='hs-varid'>replicate</span> <span class='hs-varid'>fetchesLen</span> <span class='hs-varid'>nullTensor</span><span class='hs-layout'>)</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-keyword'>_</span> <span class='hs-varid'>tensorOuts</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-131"></a>        <span class='hs-comment'>-- Targets.</span>
+<a name="line-132"></a>        <span class='hs-varid'>withStringArrayLen</span> <span class='hs-varid'>targets</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-varid'>targetsLen</span> <span class='hs-varid'>ctargets</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyword'>do</span>
+<a name="line-133"></a>            <span class='hs-varid'>checkStatus</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>run</span>
+<a name="line-134"></a>                <span class='hs-varid'>session</span>
+<a name="line-135"></a>                <span class='hs-varid'>nullPtr</span>
+<a name="line-136"></a>                <span class='hs-varid'>feedNames</span> <span class='hs-varid'>cFeedTensors</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-varid'>feedsLen</span><span class='hs-layout'>)</span>
+<a name="line-137"></a>                <span class='hs-varid'>fetchNames</span> <span class='hs-varid'>tensorOuts</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-varid'>fetchesLen</span><span class='hs-layout'>)</span>
+<a name="line-138"></a>                <span class='hs-varid'>ctargets</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-varid'>targetsLen</span><span class='hs-layout'>)</span>
+<a name="line-139"></a>                <span class='hs-varid'>nullPtr</span>
+<a name="line-140"></a>            <span class='hs-varid'>outTensors</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>peekArray</span> <span class='hs-varid'>fetchesLen</span> <span class='hs-varid'>tensorOuts</span>
+<a name="line-141"></a>            <span class='hs-varid'>mapM</span> <span class='hs-varid'>createTensorData</span> <span class='hs-varid'>outTensors</span>
+<a name="line-142"></a>
+<a name="line-143"></a>
+<a name="line-144"></a><span class='hs-comment'>-- Internal.</span>
+<a name="line-145"></a>
+<a name="line-146"></a>
+<a name="line-147"></a><a name="withStringList"></a><span class='hs-comment'>-- | Use a list of ByteString as a list of CString.</span>
+<a name="line-148"></a><span class='hs-definition'>withStringList</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>CString</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span>
+<a name="line-149"></a><span class='hs-definition'>withStringList</span> <span class='hs-varid'>strings</span> <span class='hs-varid'>fn</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>go</span> <span class='hs-varid'>strings</span> <span class='hs-conid'>[]</span>
+<a name="line-150"></a>  <span class='hs-keyword'>where</span>
+<a name="line-151"></a>    <span class='hs-varid'>go</span> <span class='hs-conid'>[]</span> <span class='hs-varid'>cs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>fn</span> <span class='hs-layout'>(</span><span class='hs-varid'>reverse</span> <span class='hs-varid'>cs</span><span class='hs-layout'>)</span>
+<a name="line-152"></a>    <span class='hs-comment'>-- TODO(fmayle): Is it worth using unsafeAsCString here?</span>
+<a name="line-153"></a>    <span class='hs-varid'>go</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span><span class='hs-conop'>:</span><span class='hs-varid'>xs</span><span class='hs-layout'>)</span> <span class='hs-varid'>cs</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-varid'>useAsCString</span> <span class='hs-varid'>x</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-varid'>c</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>go</span> <span class='hs-varid'>xs</span> <span class='hs-layout'>(</span><span class='hs-varid'>c</span><span class='hs-conop'>:</span><span class='hs-varid'>cs</span><span class='hs-layout'>)</span>
+<a name="line-154"></a>
+<a name="line-155"></a>
+<a name="line-156"></a><a name="withStringArrayLen"></a><span class='hs-comment'>-- | Use a list of ByteString as an array of CString.</span>
+<a name="line-157"></a><span class='hs-definition'>withStringArrayLen</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Int</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Ptr</span> <span class='hs-conid'>CString</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span>
+<a name="line-158"></a><span class='hs-definition'>withStringArrayLen</span> <span class='hs-varid'>xs</span> <span class='hs-varid'>fn</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>withStringList</span> <span class='hs-varid'>xs</span> <span class='hs-layout'>(</span><span class='hs-varop'>`withArrayLen`</span> <span class='hs-varid'>fn</span><span class='hs-layout'>)</span>
+<a name="line-159"></a>
+<a name="line-160"></a>
+<a name="line-161"></a><a name="createRawTensor"></a><span class='hs-comment'>-- | Create a Raw.Tensor from a TensorData.</span>
+<a name="line-162"></a><span class='hs-definition'>createRawTensor</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>TensorData</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span>
+<a name="line-163"></a><span class='hs-definition'>createRawTensor</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorData</span> <span class='hs-varid'>dims</span> <span class='hs-varid'>dt</span> <span class='hs-varid'>byteVec</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span>
+<a name="line-164"></a>    <span class='hs-varid'>withArrayLen</span> <span class='hs-layout'>(</span><span class='hs-varid'>map</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-varid'>dims</span><span class='hs-layout'>)</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-varid'>cdimsLen</span> <span class='hs-varid'>cdims</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyword'>do</span>
+<a name="line-165"></a>        <span class='hs-keyword'>let</span> <span class='hs-varid'>len</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>S</span><span class='hs-varop'>.</span><span class='hs-varid'>length</span> <span class='hs-varid'>byteVec</span>
+<a name="line-166"></a>        <span class='hs-varid'>dest</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>mallocArray</span> <span class='hs-varid'>len</span>
+<a name="line-167"></a>        <span class='hs-conid'>S</span><span class='hs-varop'>.</span><span class='hs-varid'>unsafeWith</span> <span class='hs-varid'>byteVec</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>copyArray</span> <span class='hs-varid'>dest</span> <span class='hs-varid'>x</span> <span class='hs-varid'>len</span>
+<a name="line-168"></a>        <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>newTensor</span> <span class='hs-layout'>(</span><span class='hs-varid'>toEnum</span> <span class='hs-varop'>$</span> <span class='hs-varid'>fromEnum</span> <span class='hs-varid'>dt</span><span class='hs-layout'>)</span>
+<a name="line-169"></a>                      <span class='hs-varid'>cdims</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-varid'>cdimsLen</span><span class='hs-layout'>)</span>
+<a name="line-170"></a>                      <span class='hs-layout'>(</span><span class='hs-varid'>castPtr</span> <span class='hs-varid'>dest</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-varid'>len</span><span class='hs-layout'>)</span>
+<a name="line-171"></a>                      <span class='hs-varid'>tensorDeallocFunPtr</span> <span class='hs-varid'>nullPtr</span>
+<a name="line-172"></a>
+<a name="line-173"></a><a name="tensorDeallocFunPtr"></a><span class='hs-comment'>{-# NOINLINE tensorDeallocFunPtr #-}</span>
+<a name="line-174"></a><span class='hs-definition'>tensorDeallocFunPtr</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>FunPtr</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorDeallocFn</span>
+<a name="line-175"></a><span class='hs-definition'>tensorDeallocFunPtr</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>unsafePerformIO</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>wrapTensorDealloc</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-varid'>x</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>free</span> <span class='hs-varid'>x</span>
+<a name="line-176"></a>
+<a name="line-177"></a><a name="createTensorData"></a><span class='hs-comment'>-- | Create a TensorData from a Raw.Tensor.</span>
+<a name="line-178"></a><span class='hs-comment'>--</span>
+<a name="line-179"></a><span class='hs-comment'>-- Takes ownership of the Raw.Tensor.</span>
+<a name="line-180"></a><span class='hs-definition'>createTensorData</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>TensorData</span>
+<a name="line-181"></a><span class='hs-definition'>createTensorData</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-182"></a>    <span class='hs-comment'>-- Read dimensions.</span>
+<a name="line-183"></a>    <span class='hs-varid'>numDims</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>numDims</span> <span class='hs-varid'>t</span>
+<a name="line-184"></a>    <span class='hs-varid'>dims</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>mapM</span> <span class='hs-layout'>(</span><span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>dim</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>[</span><span class='hs-num'>0</span><span class='hs-keyglyph'>..</span><span class='hs-varid'>numDims</span><span class='hs-comment'>-</span><span class='hs-num'>1</span><span class='hs-keyglyph'>]</span>
+<a name="line-185"></a>    <span class='hs-comment'>-- Read type.</span>
+<a name="line-186"></a>    <span class='hs-varid'>dtype</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>toEnum</span> <span class='hs-varop'>.</span> <span class='hs-varid'>fromEnum</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>tensorType</span> <span class='hs-varid'>t</span>
+<a name="line-187"></a>    <span class='hs-comment'>-- Read data.</span>
+<a name="line-188"></a>    <span class='hs-varid'>len</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>tensorByteSize</span> <span class='hs-varid'>t</span>
+<a name="line-189"></a>    <span class='hs-varid'>bytes</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>castPtr</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>tensorData</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>IO</span> <span class='hs-layout'>(</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>Word8</span><span class='hs-layout'>)</span>
+<a name="line-190"></a>    <span class='hs-comment'>-- TODO(fmayle): Don't copy the data.</span>
+<a name="line-191"></a>    <span class='hs-varid'>v</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>S</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>peekArray</span> <span class='hs-varid'>len</span> <span class='hs-varid'>bytes</span>
+<a name="line-192"></a>    <span class='hs-comment'>-- Free tensor.</span>
+<a name="line-193"></a>    <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>deleteTensor</span> <span class='hs-varid'>t</span>
+<a name="line-194"></a>    <span class='hs-varid'>return</span> <span class='hs-varop'>$</span> <span class='hs-conid'>TensorData</span> <span class='hs-layout'>(</span><span class='hs-varid'>map</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-varid'>dims</span><span class='hs-layout'>)</span> <span class='hs-varid'>dtype</span> <span class='hs-varid'>v</span>
+<a name="line-195"></a>
+<a name="line-196"></a><a name="checkStatus"></a><span class='hs-comment'>-- | Runs the given action which does FFI calls updating a provided</span>
+<a name="line-197"></a><span class='hs-comment'>-- status object. If the status is not OK it is thrown as</span>
+<a name="line-198"></a><span class='hs-comment'>-- TensorFlowException.</span>
+<a name="line-199"></a><span class='hs-definition'>checkStatus</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-conid'>Status</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span>
+<a name="line-200"></a><span class='hs-definition'>checkStatus</span> <span class='hs-varid'>fn</span> <span class='hs-keyglyph'>=</span>
+<a name="line-201"></a>    <span class='hs-varid'>bracket</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>newStatus</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>deleteStatus</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-varid'>status</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyword'>do</span>
+<a name="line-202"></a>        <span class='hs-varid'>result</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>fn</span> <span class='hs-varid'>status</span>
+<a name="line-203"></a>        <span class='hs-varid'>code</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>getCode</span> <span class='hs-varid'>status</span>
+<a name="line-204"></a>        <span class='hs-varid'>when</span> <span class='hs-layout'>(</span><span class='hs-varid'>code</span> <span class='hs-varop'>/=</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-conid'>TF_OK</span><span class='hs-layout'>)</span> <span class='hs-varop'>$</span> <span class='hs-keyword'>do</span>
+<a name="line-205"></a>            <span class='hs-varid'>msg</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>T</span><span class='hs-varop'>.</span><span class='hs-varid'>decodeUtf8With</span> <span class='hs-conid'>T</span><span class='hs-varop'>.</span><span class='hs-varid'>lenientDecode</span> <span class='hs-varop'>&lt;$&gt;</span>
+<a name="line-206"></a>                   <span class='hs-layout'>(</span><span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>message</span> <span class='hs-varid'>status</span> <span class='hs-varop'>&gt;&gt;=</span> <span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-varid'>packCString</span><span class='hs-layout'>)</span>
+<a name="line-207"></a>            <span class='hs-varid'>throwIO</span> <span class='hs-varop'>$</span> <span class='hs-conid'>TensorFlowException</span> <span class='hs-varid'>code</span> <span class='hs-varid'>msg</span>
+<a name="line-208"></a>        <span class='hs-varid'>return</span> <span class='hs-varid'>result</span>
+<a name="line-209"></a>
+<a name="line-210"></a><a name="setSessionConfig"></a><span class='hs-definition'>setSessionConfig</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-conid'>SessionOptions</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-211"></a><span class='hs-definition'>setSessionConfig</span> <span class='hs-varid'>pb</span> <span class='hs-varid'>opt</span> <span class='hs-keyglyph'>=</span>
+<a name="line-212"></a>    <span class='hs-varid'>useProtoAsVoidPtrLen</span> <span class='hs-varid'>pb</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-varid'>ptr</span> <span class='hs-varid'>len</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-213"></a>        <span class='hs-varid'>checkStatus</span> <span class='hs-layout'>(</span><span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>setConfig</span> <span class='hs-varid'>opt</span> <span class='hs-varid'>ptr</span> <span class='hs-varid'>len</span><span class='hs-layout'>)</span>
+<a name="line-214"></a>
+<a name="line-215"></a><a name="setSessionTarget"></a><span class='hs-definition'>setSessionTarget</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-conid'>SessionOptions</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-216"></a><span class='hs-definition'>setSessionTarget</span> <span class='hs-varid'>target</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-varid'>useAsCString</span> <span class='hs-varid'>target</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>setTarget</span>
+<a name="line-217"></a>
+<a name="line-218"></a><a name="useProtoAsVoidPtrLen"></a><span class='hs-comment'>-- | Serializes the given msg and provides it as (ptr,len) argument</span>
+<a name="line-219"></a><span class='hs-comment'>-- to the given action.</span>
+<a name="line-220"></a><span class='hs-definition'>useProtoAsVoidPtrLen</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>Message</span> <span class='hs-varid'>msg</span><span class='hs-layout'>,</span> <span class='hs-conid'>Num</span> <span class='hs-varid'>c</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-221"></a>                        <span class='hs-varid'>msg</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Ptr</span> <span class='hs-varid'>b</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>c</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span>
+<a name="line-222"></a><span class='hs-definition'>useProtoAsVoidPtrLen</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>f</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-varid'>useAsCStringLen</span> <span class='hs-layout'>(</span><span class='hs-varid'>encodeMessage</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span> <span class='hs-varop'>$</span>
+<a name="line-223"></a>        <span class='hs-keyglyph'>\</span><span class='hs-layout'>(</span><span class='hs-varid'>bytes</span><span class='hs-layout'>,</span> <span class='hs-varid'>len</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>f</span> <span class='hs-layout'>(</span><span class='hs-varid'>castPtr</span> <span class='hs-varid'>bytes</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-varid'>len</span><span class='hs-layout'>)</span>
+<a name="line-224"></a>
+<a name="line-225"></a><a name="getAllOpList"></a><span class='hs-comment'>-- | Returns the serialized OpList of all OpDefs defined in this</span>
+<a name="line-226"></a><span class='hs-comment'>-- address space.</span>
+<a name="line-227"></a><span class='hs-definition'>getAllOpList</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-228"></a><span class='hs-definition'>getAllOpList</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-229"></a>    <span class='hs-varid'>foreignPtr</span> <span class='hs-keyglyph'>&lt;-</span>
+<a name="line-230"></a>        <span class='hs-varid'>mask_</span> <span class='hs-layout'>(</span><span class='hs-varid'>newForeignPtr</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>deleteBuffer</span> <span class='hs-varop'>=&lt;&lt;</span> <span class='hs-varid'>checkCall</span><span class='hs-layout'>)</span>
+<a name="line-231"></a>    <span class='hs-comment'>-- Makes a copy because it is more reliable than eviscerating</span>
+<a name="line-232"></a>    <span class='hs-comment'>-- Buffer to steal its memory (including custom deallocator).</span>
+<a name="line-233"></a>    <span class='hs-varid'>withForeignPtr</span> <span class='hs-varid'>foreignPtr</span> <span class='hs-varop'>$</span>
+<a name="line-234"></a>        <span class='hs-keyglyph'>\</span><span class='hs-varid'>ptr</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-varid'>packCStringLen</span> <span class='hs-varop'>=&lt;&lt;</span> <span class='hs-conid'>(,)</span>
+<a name="line-235"></a>                <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-layout'>(</span><span class='hs-varid'>castPtr</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>getBufferData</span> <span class='hs-varid'>ptr</span><span class='hs-layout'>)</span>
+<a name="line-236"></a>                <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>getBufferLength</span> <span class='hs-varid'>ptr</span><span class='hs-layout'>)</span>
+<a name="line-237"></a>    <span class='hs-keyword'>where</span>
+<a name="line-238"></a>      <span class='hs-varid'>checkCall</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-239"></a>          <span class='hs-varid'>p</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-varid'>getAllOpList</span>
+<a name="line-240"></a>          <span class='hs-varid'>when</span> <span class='hs-layout'>(</span><span class='hs-varid'>p</span> <span class='hs-varop'>==</span> <span class='hs-varid'>nullPtr</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>throwIO</span> <span class='hs-varid'>exception</span><span class='hs-layout'>)</span>
+<a name="line-241"></a>          <span class='hs-varid'>return</span> <span class='hs-varid'>p</span>
+<a name="line-242"></a>      <span class='hs-varid'>exception</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TensorFlowException</span>
+<a name="line-243"></a>                <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-conid'>TF_UNKNOWN</span> <span class='hs-str'>"GetAllOpList failure, check logs"</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Internal-Raw.html b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Internal-Raw.html
new file mode 100644
index 0000000..e72ddd7
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Internal-Raw.html
@@ -0,0 +1,514 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/TensorFlow/Internal/Raw.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- GENERATED by C-&gt;Haskell Compiler, version 0.28.1 Switcheroo, 1 April 2016 (Haskell)</span>
+<a name="line-2"></a><span class='hs-comment'>-- Edit the ORIGNAL .chs file instead!</span>
+<a name="line-3"></a>
+<a name="line-4"></a>
+<a name="line-5"></a><span class='hs-comment'>{-# LINE 1 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-6"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-7"></a><span class='hs-comment'>--</span>
+<a name="line-8"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-9"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-10"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-11"></a><span class='hs-comment'>--</span>
+<a name="line-12"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-13"></a><span class='hs-comment'>--</span>
+<a name="line-14"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-15"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-16"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-17"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-18"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-19"></a>
+<a name="line-20"></a><span class='hs-comment'>{-# LANGUAGE ForeignFunctionInterface #-}</span>
+<a name="line-21"></a>
+<a name="line-22"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Internal</span><span class='hs-varop'>.</span><span class='hs-conid'>Raw</span> <span class='hs-keyword'>where</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Foreign</span><span class='hs-varop'>.</span><span class='hs-conid'>C</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>C2HSImp</span>
+<a name="line-24"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Foreign</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>C2HSImp</span>
+<a name="line-25"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Foreign</span><span class='hs-varop'>.</span><span class='hs-conid'>Storable</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>C2HSImp</span>
+<a name="line-26"></a>
+<a name="line-27"></a>
+<a name="line-28"></a>
+<a name="line-29"></a>
+<a name="line-30"></a>
+<a name="line-31"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Foreign</span>
+<a name="line-32"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Foreign</span><span class='hs-varop'>.</span><span class='hs-conid'>C</span>
+<a name="line-33"></a>
+<a name="line-34"></a><a name="DataType"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>DataType</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_FLOAT</span>
+<a name="line-35"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_DOUBLE</span>
+<a name="line-36"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_INT32</span>
+<a name="line-37"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_UINT8</span>
+<a name="line-38"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_INT16</span>
+<a name="line-39"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_INT8</span>
+<a name="line-40"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_STRING</span>
+<a name="line-41"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_COMPLEX64</span>
+<a name="line-42"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_COMPLEX</span>
+<a name="line-43"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_INT64</span>
+<a name="line-44"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_BOOL</span>
+<a name="line-45"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_QINT8</span>
+<a name="line-46"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_QUINT8</span>
+<a name="line-47"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_QINT32</span>
+<a name="line-48"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_BFLOAT16</span>
+<a name="line-49"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_QINT16</span>
+<a name="line-50"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_QUINT16</span>
+<a name="line-51"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_UINT16</span>
+<a name="line-52"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_COMPLEX128</span>
+<a name="line-53"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_HALF</span>
+<a name="line-54"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_RESOURCE</span>
+<a name="line-55"></a>  <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-56"></a><a name="instance%20Enum%20DataType"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Enum</span> <span class='hs-conid'>DataType</span> <span class='hs-keyword'>where</span>
+<a name="line-57"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_FLOAT</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_DOUBLE</span>
+<a name="line-58"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_DOUBLE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INT32</span>
+<a name="line-59"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_INT32</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UINT8</span>
+<a name="line-60"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_UINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INT16</span>
+<a name="line-61"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_INT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INT8</span>
+<a name="line-62"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_INT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_STRING</span>
+<a name="line-63"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_STRING</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_COMPLEX64</span>
+<a name="line-64"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_COMPLEX64</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INT64</span>
+<a name="line-65"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_COMPLEX</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INT64</span>
+<a name="line-66"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_INT64</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_BOOL</span>
+<a name="line-67"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_BOOL</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_QINT8</span>
+<a name="line-68"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_QINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_QUINT8</span>
+<a name="line-69"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_QUINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_QINT32</span>
+<a name="line-70"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_QINT32</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_BFLOAT16</span>
+<a name="line-71"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_BFLOAT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_QINT16</span>
+<a name="line-72"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_QINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_QUINT16</span>
+<a name="line-73"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_QUINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UINT16</span>
+<a name="line-74"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_UINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_COMPLEX128</span>
+<a name="line-75"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_COMPLEX128</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_HALF</span>
+<a name="line-76"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_HALF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_RESOURCE</span>
+<a name="line-77"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_RESOURCE</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-str'>"DataType.succ: TF_RESOURCE has no successor"</span>
+<a name="line-78"></a>
+<a name="line-79"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_DOUBLE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_FLOAT</span>
+<a name="line-80"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_INT32</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_DOUBLE</span>
+<a name="line-81"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_UINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INT32</span>
+<a name="line-82"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_INT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UINT8</span>
+<a name="line-83"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_INT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INT16</span>
+<a name="line-84"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_STRING</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INT8</span>
+<a name="line-85"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_COMPLEX64</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_STRING</span>
+<a name="line-86"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_COMPLEX</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_STRING</span>
+<a name="line-87"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_INT64</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_COMPLEX64</span>
+<a name="line-88"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_BOOL</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INT64</span>
+<a name="line-89"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_QINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_BOOL</span>
+<a name="line-90"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_QUINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_QINT8</span>
+<a name="line-91"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_QINT32</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_QUINT8</span>
+<a name="line-92"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_BFLOAT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_QINT32</span>
+<a name="line-93"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_QINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_BFLOAT16</span>
+<a name="line-94"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_QUINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_QINT16</span>
+<a name="line-95"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_UINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_QUINT16</span>
+<a name="line-96"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_COMPLEX128</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UINT16</span>
+<a name="line-97"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_HALF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_COMPLEX128</span>
+<a name="line-98"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_RESOURCE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_HALF</span>
+<a name="line-99"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_FLOAT</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-str'>"DataType.pred: TF_FLOAT has no predecessor"</span>
+<a name="line-100"></a>
+<a name="line-101"></a>  <span class='hs-varid'>enumFromTo</span> <span class='hs-varid'>from</span> <span class='hs-varid'>to</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>go</span> <span class='hs-varid'>from</span>
+<a name="line-102"></a>    <span class='hs-keyword'>where</span>
+<a name="line-103"></a>      <span class='hs-varid'>end</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>fromEnum</span> <span class='hs-varid'>to</span>
+<a name="line-104"></a>      <span class='hs-varid'>go</span> <span class='hs-varid'>v</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>case</span> <span class='hs-varid'>compare</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromEnum</span> <span class='hs-varid'>v</span><span class='hs-layout'>)</span> <span class='hs-varid'>end</span> <span class='hs-keyword'>of</span>
+<a name="line-105"></a>                 <span class='hs-conid'>LT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>v</span> <span class='hs-conop'>:</span> <span class='hs-varid'>go</span> <span class='hs-layout'>(</span><span class='hs-varid'>succ</span> <span class='hs-varid'>v</span><span class='hs-layout'>)</span>
+<a name="line-106"></a>                 <span class='hs-conid'>EQ</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>v</span><span class='hs-keyglyph'>]</span>
+<a name="line-107"></a>                 <span class='hs-conid'>GT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>[]</span>
+<a name="line-108"></a>
+<a name="line-109"></a>  <span class='hs-varid'>enumFrom</span> <span class='hs-varid'>from</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>enumFromTo</span> <span class='hs-varid'>from</span> <span class='hs-conid'>TF_RESOURCE</span>
+<a name="line-110"></a>
+<a name="line-111"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_FLOAT</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>1</span>
+<a name="line-112"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_DOUBLE</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>2</span>
+<a name="line-113"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_INT32</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>3</span>
+<a name="line-114"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_UINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>4</span>
+<a name="line-115"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_INT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>5</span>
+<a name="line-116"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_INT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>6</span>
+<a name="line-117"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_STRING</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>7</span>
+<a name="line-118"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_COMPLEX64</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>8</span>
+<a name="line-119"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_COMPLEX</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>8</span>
+<a name="line-120"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_INT64</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>9</span>
+<a name="line-121"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_BOOL</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>10</span>
+<a name="line-122"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_QINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>11</span>
+<a name="line-123"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_QUINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>12</span>
+<a name="line-124"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_QINT32</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>13</span>
+<a name="line-125"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_BFLOAT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>14</span>
+<a name="line-126"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_QINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>15</span>
+<a name="line-127"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_QUINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>16</span>
+<a name="line-128"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_UINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>17</span>
+<a name="line-129"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_COMPLEX128</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>18</span>
+<a name="line-130"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_HALF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>19</span>
+<a name="line-131"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_RESOURCE</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>20</span>
+<a name="line-132"></a>
+<a name="line-133"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>1</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_FLOAT</span>
+<a name="line-134"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>2</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_DOUBLE</span>
+<a name="line-135"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>3</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INT32</span>
+<a name="line-136"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>4</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UINT8</span>
+<a name="line-137"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>5</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INT16</span>
+<a name="line-138"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>6</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INT8</span>
+<a name="line-139"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>7</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_STRING</span>
+<a name="line-140"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_COMPLEX64</span>
+<a name="line-141"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>9</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INT64</span>
+<a name="line-142"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>10</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_BOOL</span>
+<a name="line-143"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>11</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_QINT8</span>
+<a name="line-144"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>12</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_QUINT8</span>
+<a name="line-145"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>13</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_QINT32</span>
+<a name="line-146"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>14</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_BFLOAT16</span>
+<a name="line-147"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>15</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_QINT16</span>
+<a name="line-148"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_QUINT16</span>
+<a name="line-149"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>17</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UINT16</span>
+<a name="line-150"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>18</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_COMPLEX128</span>
+<a name="line-151"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>19</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_HALF</span>
+<a name="line-152"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>20</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_RESOURCE</span>
+<a name="line-153"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-varid'>unmatched</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-layout'>(</span><span class='hs-str'>"DataType.toEnum: Cannot match "</span> <span class='hs-varop'>++</span> <span class='hs-varid'>show</span> <span class='hs-varid'>unmatched</span><span class='hs-layout'>)</span>
+<a name="line-154"></a>
+<a name="line-155"></a><span class='hs-comment'>{-# LINE 24 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-156"></a>
+<a name="line-157"></a><a name="Code"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>Code</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_OK</span>
+<a name="line-158"></a>          <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_CANCELLED</span>
+<a name="line-159"></a>          <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_UNKNOWN</span>
+<a name="line-160"></a>          <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_INVALID_ARGUMENT</span>
+<a name="line-161"></a>          <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_DEADLINE_EXCEEDED</span>
+<a name="line-162"></a>          <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_NOT_FOUND</span>
+<a name="line-163"></a>          <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_ALREADY_EXISTS</span>
+<a name="line-164"></a>          <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_PERMISSION_DENIED</span>
+<a name="line-165"></a>          <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_RESOURCE_EXHAUSTED</span>
+<a name="line-166"></a>          <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_FAILED_PRECONDITION</span>
+<a name="line-167"></a>          <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_ABORTED</span>
+<a name="line-168"></a>          <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_OUT_OF_RANGE</span>
+<a name="line-169"></a>          <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_UNIMPLEMENTED</span>
+<a name="line-170"></a>          <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_INTERNAL</span>
+<a name="line-171"></a>          <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_UNAVAILABLE</span>
+<a name="line-172"></a>          <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_DATA_LOSS</span>
+<a name="line-173"></a>          <span class='hs-keyglyph'>|</span> <span class='hs-conid'>TF_UNAUTHENTICATED</span>
+<a name="line-174"></a>  <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-175"></a><a name="instance%20Enum%20Code"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Enum</span> <span class='hs-conid'>Code</span> <span class='hs-keyword'>where</span>
+<a name="line-176"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_OK</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_CANCELLED</span>
+<a name="line-177"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_CANCELLED</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UNKNOWN</span>
+<a name="line-178"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_UNKNOWN</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INVALID_ARGUMENT</span>
+<a name="line-179"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_INVALID_ARGUMENT</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_DEADLINE_EXCEEDED</span>
+<a name="line-180"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_DEADLINE_EXCEEDED</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_NOT_FOUND</span>
+<a name="line-181"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_NOT_FOUND</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_ALREADY_EXISTS</span>
+<a name="line-182"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_ALREADY_EXISTS</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_PERMISSION_DENIED</span>
+<a name="line-183"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_PERMISSION_DENIED</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_RESOURCE_EXHAUSTED</span>
+<a name="line-184"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_RESOURCE_EXHAUSTED</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_FAILED_PRECONDITION</span>
+<a name="line-185"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_FAILED_PRECONDITION</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_ABORTED</span>
+<a name="line-186"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_ABORTED</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_OUT_OF_RANGE</span>
+<a name="line-187"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_OUT_OF_RANGE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UNIMPLEMENTED</span>
+<a name="line-188"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_UNIMPLEMENTED</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INTERNAL</span>
+<a name="line-189"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_INTERNAL</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UNAVAILABLE</span>
+<a name="line-190"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_UNAVAILABLE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_DATA_LOSS</span>
+<a name="line-191"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_DATA_LOSS</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UNAUTHENTICATED</span>
+<a name="line-192"></a>  <span class='hs-varid'>succ</span> <span class='hs-conid'>TF_UNAUTHENTICATED</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-str'>"Code.succ: TF_UNAUTHENTICATED has no successor"</span>
+<a name="line-193"></a>
+<a name="line-194"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_CANCELLED</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_OK</span>
+<a name="line-195"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_UNKNOWN</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_CANCELLED</span>
+<a name="line-196"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_INVALID_ARGUMENT</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UNKNOWN</span>
+<a name="line-197"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_DEADLINE_EXCEEDED</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INVALID_ARGUMENT</span>
+<a name="line-198"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_NOT_FOUND</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_DEADLINE_EXCEEDED</span>
+<a name="line-199"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_ALREADY_EXISTS</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_NOT_FOUND</span>
+<a name="line-200"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_PERMISSION_DENIED</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_ALREADY_EXISTS</span>
+<a name="line-201"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_RESOURCE_EXHAUSTED</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_PERMISSION_DENIED</span>
+<a name="line-202"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_FAILED_PRECONDITION</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_RESOURCE_EXHAUSTED</span>
+<a name="line-203"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_ABORTED</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_FAILED_PRECONDITION</span>
+<a name="line-204"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_OUT_OF_RANGE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_ABORTED</span>
+<a name="line-205"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_UNIMPLEMENTED</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_OUT_OF_RANGE</span>
+<a name="line-206"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_INTERNAL</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UNIMPLEMENTED</span>
+<a name="line-207"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_UNAVAILABLE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INTERNAL</span>
+<a name="line-208"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_DATA_LOSS</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UNAVAILABLE</span>
+<a name="line-209"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_UNAUTHENTICATED</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_DATA_LOSS</span>
+<a name="line-210"></a>  <span class='hs-varid'>pred</span> <span class='hs-conid'>TF_OK</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-str'>"Code.pred: TF_OK has no predecessor"</span>
+<a name="line-211"></a>
+<a name="line-212"></a>  <span class='hs-varid'>enumFromTo</span> <span class='hs-varid'>from</span> <span class='hs-varid'>to</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>go</span> <span class='hs-varid'>from</span>
+<a name="line-213"></a>    <span class='hs-keyword'>where</span>
+<a name="line-214"></a>      <span class='hs-varid'>end</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>fromEnum</span> <span class='hs-varid'>to</span>
+<a name="line-215"></a>      <span class='hs-varid'>go</span> <span class='hs-varid'>v</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>case</span> <span class='hs-varid'>compare</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromEnum</span> <span class='hs-varid'>v</span><span class='hs-layout'>)</span> <span class='hs-varid'>end</span> <span class='hs-keyword'>of</span>
+<a name="line-216"></a>                 <span class='hs-conid'>LT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>v</span> <span class='hs-conop'>:</span> <span class='hs-varid'>go</span> <span class='hs-layout'>(</span><span class='hs-varid'>succ</span> <span class='hs-varid'>v</span><span class='hs-layout'>)</span>
+<a name="line-217"></a>                 <span class='hs-conid'>EQ</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>v</span><span class='hs-keyglyph'>]</span>
+<a name="line-218"></a>                 <span class='hs-conid'>GT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>[]</span>
+<a name="line-219"></a>
+<a name="line-220"></a>  <span class='hs-varid'>enumFrom</span> <span class='hs-varid'>from</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>enumFromTo</span> <span class='hs-varid'>from</span> <span class='hs-conid'>TF_UNAUTHENTICATED</span>
+<a name="line-221"></a>
+<a name="line-222"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_OK</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>0</span>
+<a name="line-223"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_CANCELLED</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>1</span>
+<a name="line-224"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_UNKNOWN</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>2</span>
+<a name="line-225"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_INVALID_ARGUMENT</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>3</span>
+<a name="line-226"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_DEADLINE_EXCEEDED</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>4</span>
+<a name="line-227"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_NOT_FOUND</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>5</span>
+<a name="line-228"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_ALREADY_EXISTS</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>6</span>
+<a name="line-229"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_PERMISSION_DENIED</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>7</span>
+<a name="line-230"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_RESOURCE_EXHAUSTED</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>8</span>
+<a name="line-231"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_FAILED_PRECONDITION</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>9</span>
+<a name="line-232"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_ABORTED</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>10</span>
+<a name="line-233"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_OUT_OF_RANGE</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>11</span>
+<a name="line-234"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_UNIMPLEMENTED</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>12</span>
+<a name="line-235"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_INTERNAL</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>13</span>
+<a name="line-236"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_UNAVAILABLE</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>14</span>
+<a name="line-237"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_DATA_LOSS</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>15</span>
+<a name="line-238"></a>  <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>TF_UNAUTHENTICATED</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>16</span>
+<a name="line-239"></a>
+<a name="line-240"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>0</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_OK</span>
+<a name="line-241"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>1</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_CANCELLED</span>
+<a name="line-242"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>2</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UNKNOWN</span>
+<a name="line-243"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>3</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INVALID_ARGUMENT</span>
+<a name="line-244"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>4</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_DEADLINE_EXCEEDED</span>
+<a name="line-245"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>5</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_NOT_FOUND</span>
+<a name="line-246"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>6</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_ALREADY_EXISTS</span>
+<a name="line-247"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>7</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_PERMISSION_DENIED</span>
+<a name="line-248"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_RESOURCE_EXHAUSTED</span>
+<a name="line-249"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>9</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_FAILED_PRECONDITION</span>
+<a name="line-250"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>10</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_ABORTED</span>
+<a name="line-251"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>11</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_OUT_OF_RANGE</span>
+<a name="line-252"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>12</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UNIMPLEMENTED</span>
+<a name="line-253"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>13</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_INTERNAL</span>
+<a name="line-254"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>14</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UNAVAILABLE</span>
+<a name="line-255"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>15</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_DATA_LOSS</span>
+<a name="line-256"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-num'>16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TF_UNAUTHENTICATED</span>
+<a name="line-257"></a>  <span class='hs-varid'>toEnum</span> <span class='hs-varid'>unmatched</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-layout'>(</span><span class='hs-str'>"Code.toEnum: Cannot match "</span> <span class='hs-varop'>++</span> <span class='hs-varid'>show</span> <span class='hs-varid'>unmatched</span><span class='hs-layout'>)</span>
+<a name="line-258"></a>
+<a name="line-259"></a><span class='hs-comment'>{-# LINE 25 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-260"></a>
+<a name="line-261"></a>
+<a name="line-262"></a>
+<a name="line-263"></a><a name="Status"></a><span class='hs-comment'>-- Status.</span>
+<a name="line-264"></a><a name="Status"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>Status</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Status</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-layout'>(</span><span class='hs-conid'>Status</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-265"></a><span class='hs-comment'>{-# LINE 29 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-266"></a>
+<a name="line-267"></a>
+<a name="line-268"></a><a name="newStatus"></a><span class='hs-definition'>newStatus</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>Status</span>
+<a name="line-269"></a><span class='hs-definition'>newStatus</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFNewStatus</span>
+<a name="line-270"></a><span class='hs-comment'>{-# LINE 32 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-271"></a>
+<a name="line-272"></a>
+<a name="line-273"></a><a name="deleteStatus"></a><span class='hs-definition'>deleteStatus</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Status</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-274"></a><span class='hs-definition'>deleteStatus</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFDeleteStatus</span>
+<a name="line-275"></a><span class='hs-comment'>{-# LINE 35 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-276"></a>
+<a name="line-277"></a>
+<a name="line-278"></a><a name="setStatus"></a><span class='hs-definition'>setStatus</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Status</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Code</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>CString</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-279"></a><span class='hs-definition'>setStatus</span> <span class='hs-varid'>s</span> <span class='hs-varid'>c</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFSetStatus</span> <span class='hs-varid'>s</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-varop'>$</span> <span class='hs-varid'>fromEnum</span> <span class='hs-varid'>c</span><span class='hs-layout'>)</span>
+<a name="line-280"></a>
+<a name="line-281"></a><a name="getCode"></a><span class='hs-definition'>getCode</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Status</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>Code</span>
+<a name="line-282"></a><span class='hs-definition'>getCode</span> <span class='hs-varid'>s</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>toEnum</span> <span class='hs-varop'>.</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>tFGetCode</span> <span class='hs-varid'>s</span>
+<a name="line-283"></a>
+<a name="line-284"></a><a name="message"></a><span class='hs-definition'>message</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Status</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>CString</span>
+<a name="line-285"></a><span class='hs-definition'>message</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFMessage</span>
+<a name="line-286"></a><span class='hs-comment'>{-# LINE 44 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-287"></a>
+<a name="line-288"></a>
+<a name="line-289"></a>
+<a name="line-290"></a><a name="Buffer"></a><span class='hs-comment'>-- Buffer.</span>
+<a name="line-291"></a><a name="Buffer"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>Buffer</span>
+<a name="line-292"></a><a name="BufferPtr"></a><span class='hs-keyword'>type</span> <span class='hs-conid'>BufferPtr</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-layout'>(</span><span class='hs-conid'>Buffer</span><span class='hs-layout'>)</span>
+<a name="line-293"></a><span class='hs-comment'>{-# LINE 49 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-294"></a>
+<a name="line-295"></a>
+<a name="line-296"></a><a name="getBufferData"></a><span class='hs-definition'>getBufferData</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>BufferPtr</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-layout'>(</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span>
+<a name="line-297"></a><span class='hs-definition'>getBufferData</span> <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>ptr</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyword'>do</span> <span class='hs-layout'>{</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-varid'>peekByteOff</span> <span class='hs-varid'>ptr</span> <span class='hs-num'>0</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>IO</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-298"></a><span class='hs-comment'>{-# LINE 52 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-299"></a>
+<a name="line-300"></a>
+<a name="line-301"></a><a name="getBufferLength"></a><span class='hs-definition'>getBufferLength</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>BufferPtr</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>CULong</span>
+<a name="line-302"></a><span class='hs-definition'>getBufferLength</span> <span class='hs-keyglyph'>=</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>ptr</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyword'>do</span> <span class='hs-layout'>{</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-varid'>peekByteOff</span> <span class='hs-varid'>ptr</span> <span class='hs-num'>8</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CULong</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-303"></a><span class='hs-comment'>{-# LINE 55 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-304"></a>
+<a name="line-305"></a>
+<a name="line-306"></a><a name="Tensor"></a><span class='hs-comment'>-- Tensor.</span>
+<a name="line-307"></a><a name="Tensor"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>Tensor</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Tensor</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-308"></a><span class='hs-comment'>{-# LINE 58 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-309"></a>
+<a name="line-310"></a>
+<a name="line-311"></a><a name="instance%20Storable%20Tensor"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Storable</span> <span class='hs-conid'>Tensor</span> <span class='hs-keyword'>where</span>
+<a name="line-312"></a>    <span class='hs-varid'>sizeOf</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>sizeOf</span> <span class='hs-varid'>t</span>
+<a name="line-313"></a>    <span class='hs-varid'>alignment</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>alignment</span> <span class='hs-varid'>t</span>
+<a name="line-314"></a>    <span class='hs-varid'>peek</span> <span class='hs-varid'>p</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>fmap</span> <span class='hs-conid'>Tensor</span> <span class='hs-layout'>(</span><span class='hs-varid'>peek</span> <span class='hs-layout'>(</span><span class='hs-varid'>castPtr</span> <span class='hs-varid'>p</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-315"></a>    <span class='hs-varid'>poke</span> <span class='hs-varid'>p</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>poke</span> <span class='hs-layout'>(</span><span class='hs-varid'>castPtr</span> <span class='hs-varid'>p</span><span class='hs-layout'>)</span> <span class='hs-varid'>t</span>
+<a name="line-316"></a>
+<a name="line-317"></a><a name="newTensor"></a><span class='hs-definition'>newTensor</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>DataType</span>
+<a name="line-318"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Ptr</span> <span class='hs-conid'>CLong</span>    <span class='hs-comment'>-- dimensions array</span>
+<a name="line-319"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>CInt</span>         <span class='hs-comment'>-- num dimensions</span>
+<a name="line-320"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span>       <span class='hs-comment'>-- data</span>
+<a name="line-321"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>CULong</span>       <span class='hs-comment'>-- data len</span>
+<a name="line-322"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>FunPtr</span> <span class='hs-layout'>(</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>CULong</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span>  <span class='hs-comment'>-- deallocator</span>
+<a name="line-323"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span>       <span class='hs-comment'>-- deallocator arg</span>
+<a name="line-324"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>Tensor</span>
+<a name="line-325"></a><span class='hs-definition'>newTensor</span> <span class='hs-varid'>dt</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFNewTensor</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-varop'>$</span> <span class='hs-varid'>fromEnum</span> <span class='hs-varid'>dt</span><span class='hs-layout'>)</span>
+<a name="line-326"></a>
+<a name="line-327"></a><a name="deleteTensor"></a><span class='hs-definition'>deleteTensor</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-328"></a><span class='hs-definition'>deleteTensor</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFDeleteTensor</span>
+<a name="line-329"></a><span class='hs-comment'>{-# LINE 77 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-330"></a>
+<a name="line-331"></a>
+<a name="line-332"></a><a name="tensorType"></a><span class='hs-definition'>tensorType</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>DataType</span>
+<a name="line-333"></a><span class='hs-definition'>tensorType</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>toEnum</span> <span class='hs-varop'>.</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>tFTensorType</span> <span class='hs-varid'>t</span>
+<a name="line-334"></a>
+<a name="line-335"></a><a name="numDims"></a><span class='hs-definition'>numDims</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>CInt</span>
+<a name="line-336"></a><span class='hs-definition'>numDims</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFNumDims</span>
+<a name="line-337"></a><span class='hs-comment'>{-# LINE 83 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-338"></a>
+<a name="line-339"></a>
+<a name="line-340"></a><a name="dim"></a><span class='hs-definition'>dim</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>CInt</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>CLong</span>
+<a name="line-341"></a><span class='hs-definition'>dim</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFDim</span>
+<a name="line-342"></a><span class='hs-comment'>{-# LINE 86 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-343"></a>
+<a name="line-344"></a>
+<a name="line-345"></a><a name="tensorByteSize"></a><span class='hs-definition'>tensorByteSize</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>CULong</span>
+<a name="line-346"></a><span class='hs-definition'>tensorByteSize</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFTensorByteSize</span>
+<a name="line-347"></a><span class='hs-comment'>{-# LINE 89 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-348"></a>
+<a name="line-349"></a>
+<a name="line-350"></a><a name="tensorData"></a><span class='hs-definition'>tensorData</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-layout'>(</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span>
+<a name="line-351"></a><span class='hs-definition'>tensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFTensorData</span>
+<a name="line-352"></a><span class='hs-comment'>{-# LINE 92 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-353"></a>
+<a name="line-354"></a>
+<a name="line-355"></a>
+<a name="line-356"></a><a name="SessionOptions"></a><span class='hs-comment'>-- Session Options.</span>
+<a name="line-357"></a><a name="SessionOptions"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>SessionOptions</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>SessionOptions</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-layout'>(</span><span class='hs-conid'>SessionOptions</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-358"></a><span class='hs-comment'>{-# LINE 96 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-359"></a>
+<a name="line-360"></a>
+<a name="line-361"></a><a name="newSessionOptions"></a><span class='hs-definition'>newSessionOptions</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>SessionOptions</span>
+<a name="line-362"></a><span class='hs-definition'>newSessionOptions</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFNewSessionOptions</span>
+<a name="line-363"></a><span class='hs-comment'>{-# LINE 99 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-364"></a>
+<a name="line-365"></a>
+<a name="line-366"></a><a name="setTarget"></a><span class='hs-definition'>setTarget</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>SessionOptions</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>CString</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-367"></a><span class='hs-definition'>setTarget</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFSetTarget</span>
+<a name="line-368"></a><span class='hs-comment'>{-# LINE 102 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-369"></a>
+<a name="line-370"></a>
+<a name="line-371"></a><a name="setConfig"></a><span class='hs-definition'>setConfig</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>SessionOptions</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>CULong</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Status</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-372"></a><span class='hs-definition'>setConfig</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFSetConfig</span>
+<a name="line-373"></a><span class='hs-comment'>{-# LINE 105 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-374"></a>
+<a name="line-375"></a>
+<a name="line-376"></a><a name="deleteSessionOptions"></a><span class='hs-definition'>deleteSessionOptions</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>SessionOptions</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-377"></a><span class='hs-definition'>deleteSessionOptions</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFDeleteSessionOptions</span>
+<a name="line-378"></a><span class='hs-comment'>{-# LINE 108 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-379"></a>
+<a name="line-380"></a>
+<a name="line-381"></a>
+<a name="line-382"></a><a name="Session"></a><span class='hs-comment'>-- Session.</span>
+<a name="line-383"></a><a name="Session"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>Session</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Session</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-layout'>(</span><span class='hs-conid'>Session</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-384"></a><span class='hs-comment'>{-# LINE 112 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-385"></a>
+<a name="line-386"></a>
+<a name="line-387"></a><a name="newSession"></a><span class='hs-definition'>newSession</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>SessionOptions</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Status</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>Session</span>
+<a name="line-388"></a><span class='hs-definition'>newSession</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFNewSession</span>
+<a name="line-389"></a><span class='hs-comment'>{-# LINE 115 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-390"></a>
+<a name="line-391"></a>
+<a name="line-392"></a><a name="closeSession"></a><span class='hs-definition'>closeSession</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Session</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Status</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-393"></a><span class='hs-definition'>closeSession</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFCloseSession</span>
+<a name="line-394"></a><span class='hs-comment'>{-# LINE 118 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-395"></a>
+<a name="line-396"></a>
+<a name="line-397"></a><a name="deleteSession"></a><span class='hs-definition'>deleteSession</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Session</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Status</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-398"></a><span class='hs-definition'>deleteSession</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFDeleteSession</span>
+<a name="line-399"></a><span class='hs-comment'>{-# LINE 121 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-400"></a>
+<a name="line-401"></a>
+<a name="line-402"></a><a name="extendGraph"></a><span class='hs-definition'>extendGraph</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Session</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>CULong</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Status</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-403"></a><span class='hs-definition'>extendGraph</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFExtendGraph</span>
+<a name="line-404"></a><span class='hs-comment'>{-# LINE 124 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-405"></a>
+<a name="line-406"></a>
+<a name="line-407"></a><a name="run"></a><span class='hs-definition'>run</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Session</span>
+<a name="line-408"></a>    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>BufferPtr</span>                          <span class='hs-comment'>-- RunOptions proto.</span>
+<a name="line-409"></a>    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Ptr</span> <span class='hs-conid'>CString</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Ptr</span> <span class='hs-conid'>Tensor</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>CInt</span>  <span class='hs-comment'>-- Input (names, tensors, count).</span>
+<a name="line-410"></a>    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Ptr</span> <span class='hs-conid'>CString</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Ptr</span> <span class='hs-conid'>Tensor</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>CInt</span>  <span class='hs-comment'>-- Output (names, tensors, count).</span>
+<a name="line-411"></a>    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Ptr</span> <span class='hs-conid'>CString</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>CInt</span>                <span class='hs-comment'>-- Target nodes (names, count).</span>
+<a name="line-412"></a>    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>BufferPtr</span>                          <span class='hs-comment'>-- RunMetadata proto.</span>
+<a name="line-413"></a>    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Status</span>
+<a name="line-414"></a>    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-415"></a><span class='hs-definition'>run</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFRun</span>
+<a name="line-416"></a><span class='hs-comment'>{-# LINE 134 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-417"></a>
+<a name="line-418"></a>
+<a name="line-419"></a><a name="TensorDeallocFn"></a><span class='hs-comment'>-- FFI helpers.</span>
+<a name="line-420"></a><a name="TensorDeallocFn"></a><span class='hs-keyword'>type</span> <span class='hs-conid'>TensorDeallocFn</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>CULong</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-421"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-str'>"wrapper"</span>
+<a name="line-422"></a>    <span class='hs-varid'>wrapTensorDealloc</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>TensorDeallocFn</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-layout'>(</span><span class='hs-conid'>FunPtr</span> <span class='hs-conid'>TensorDeallocFn</span><span class='hs-layout'>)</span>
+<a name="line-423"></a>
+<a name="line-424"></a>
+<a name="line-425"></a><a name="getAllOpList"></a><span class='hs-comment'>-- | Get the OpList of all OpDefs defined in this address space.</span>
+<a name="line-426"></a><span class='hs-comment'>-- Returns a BufferPtr, ownership of which is transferred to the caller</span>
+<a name="line-427"></a><span class='hs-comment'>-- (and can be freed using deleteBuffer).</span>
+<a name="line-428"></a><span class='hs-comment'>--</span>
+<a name="line-429"></a><span class='hs-comment'>-- The data in the buffer will be the serialized OpList proto for ops registered</span>
+<a name="line-430"></a><span class='hs-comment'>-- in this address space.</span>
+<a name="line-431"></a><span class='hs-definition'>getAllOpList</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>BufferPtr</span>
+<a name="line-432"></a><span class='hs-definition'>getAllOpList</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tFGetAllOpList</span>
+<a name="line-433"></a><span class='hs-comment'>{-# LINE 149 "src/TensorFlow/Internal/Raw.chs" #-}</span>
+<a name="line-434"></a>
+<a name="line-435"></a>
+<a name="line-436"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-str'>"&amp;TF_DeleteBuffer"</span>
+<a name="line-437"></a>  <span class='hs-varid'>deleteBuffer</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>FunPtr</span> <span class='hs-layout'>(</span><span class='hs-conid'>BufferPtr</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span>
+<a name="line-438"></a>
+<a name="line-439"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_NewStatus"</span>
+<a name="line-440"></a>  <span class='hs-varid'>tFNewStatus</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-layout'>(</span><span class='hs-conid'>Status</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-441"></a>
+<a name="line-442"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_DeleteStatus"</span>
+<a name="line-443"></a>  <span class='hs-varid'>tFDeleteStatus</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Status</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-444"></a>
+<a name="line-445"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_SetStatus"</span>
+<a name="line-446"></a>  <span class='hs-varid'>tFSetStatus</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Status</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CInt</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CChar</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-447"></a>
+<a name="line-448"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_GetCode"</span>
+<a name="line-449"></a>  <span class='hs-varid'>tFGetCode</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Status</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CInt</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-450"></a>
+<a name="line-451"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_Message"</span>
+<a name="line-452"></a>  <span class='hs-varid'>tFMessage</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Status</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CChar</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-453"></a>
+<a name="line-454"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_NewTensor"</span>
+<a name="line-455"></a>  <span class='hs-varid'>tFNewTensor</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CInt</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CLong</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CInt</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CULong</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>FunPtr</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CULong</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-456"></a>
+<a name="line-457"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_DeleteTensor"</span>
+<a name="line-458"></a>  <span class='hs-varid'>tFDeleteTensor</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-459"></a>
+<a name="line-460"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_TensorType"</span>
+<a name="line-461"></a>  <span class='hs-varid'>tFTensorType</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CInt</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-462"></a>
+<a name="line-463"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_NumDims"</span>
+<a name="line-464"></a>  <span class='hs-varid'>tFNumDims</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CInt</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-465"></a>
+<a name="line-466"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_Dim"</span>
+<a name="line-467"></a>  <span class='hs-varid'>tFDim</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CInt</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CLong</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-468"></a>
+<a name="line-469"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_TensorByteSize"</span>
+<a name="line-470"></a>  <span class='hs-varid'>tFTensorByteSize</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CULong</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-471"></a>
+<a name="line-472"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_TensorData"</span>
+<a name="line-473"></a>  <span class='hs-varid'>tFTensorData</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-474"></a>
+<a name="line-475"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_NewSessionOptions"</span>
+<a name="line-476"></a>  <span class='hs-varid'>tFNewSessionOptions</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-layout'>(</span><span class='hs-conid'>SessionOptions</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-477"></a>
+<a name="line-478"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_SetTarget"</span>
+<a name="line-479"></a>  <span class='hs-varid'>tFSetTarget</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>SessionOptions</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CChar</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-480"></a>
+<a name="line-481"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_SetConfig"</span>
+<a name="line-482"></a>  <span class='hs-varid'>tFSetConfig</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>SessionOptions</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CULong</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Status</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-483"></a>
+<a name="line-484"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_DeleteSessionOptions"</span>
+<a name="line-485"></a>  <span class='hs-varid'>tFDeleteSessionOptions</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>SessionOptions</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-486"></a>
+<a name="line-487"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_NewSession"</span>
+<a name="line-488"></a>  <span class='hs-varid'>tFNewSession</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>SessionOptions</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Status</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-layout'>(</span><span class='hs-conid'>Session</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-489"></a>
+<a name="line-490"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_CloseSession"</span>
+<a name="line-491"></a>  <span class='hs-varid'>tFCloseSession</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Session</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Status</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-492"></a>
+<a name="line-493"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_DeleteSession"</span>
+<a name="line-494"></a>  <span class='hs-varid'>tFDeleteSession</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Session</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Status</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-495"></a>
+<a name="line-496"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_ExtendGraph"</span>
+<a name="line-497"></a>  <span class='hs-varid'>tFExtendGraph</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Session</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CULong</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Status</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-498"></a>
+<a name="line-499"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_Run"</span>
+<a name="line-500"></a>  <span class='hs-varid'>tFRun</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Session</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>BufferPtr</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CChar</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CInt</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CChar</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CInt</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>Ptr</span> <span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CChar</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>C2HSImp</span><span class='hs-varop'>.</span><span class='hs-conid'>CInt</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>BufferPtr</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Status</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-501"></a>
+<a name="line-502"></a><span class='hs-keyword'>foreign</span> <span class='hs-keyword'>import</span> <span class='hs-keyword'>ccall</span> <span class='hs-keyword'>safe</span> <span class='hs-str'>"TensorFlow/Internal/Raw.chs.h TF_GetAllOpList"</span>
+<a name="line-503"></a>  <span class='hs-varid'>tFGetAllOpList</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>IO</span> <span class='hs-layout'>(</span><span class='hs-conid'>BufferPtr</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Internal-VarInt.html b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Internal-VarInt.html
new file mode 100644
index 0000000..3f4826d
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Internal-VarInt.html
@@ -0,0 +1,61 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/Internal/VarInt.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE BangPatterns #-}</span>
+<a name="line-16"></a>
+<a name="line-17"></a><span class='hs-comment'>{-|
+<a name="line-18"></a>Module      : TensorFlow.Internal.VarInt
+<a name="line-19"></a>Description : Encoders and decoders for varint types.
+<a name="line-20"></a>
+<a name="line-21"></a>Originally taken from internal proto-lens code.
+<a name="line-22"></a>-}</span>
+<a name="line-23"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Internal</span><span class='hs-varop'>.</span><span class='hs-conid'>VarInt</span>
+<a name="line-24"></a>    <span class='hs-layout'>(</span> <span class='hs-varid'>getVarInt</span>
+<a name="line-25"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>putVarInt</span>
+<a name="line-26"></a>    <span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-27"></a>
+<a name="line-28"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Attoparsec</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Parse</span>
+<a name="line-29"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Bits</span>
+<a name="line-30"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>Lazy</span><span class='hs-varop'>.</span><span class='hs-conid'>Builder</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Builder</span>
+<a name="line-31"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Monoid</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varop'>&lt;&gt;</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-32"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span> <span class='hs-layout'>(</span><span class='hs-conid'>Word64</span><span class='hs-layout'>)</span>
+<a name="line-33"></a>
+<a name="line-34"></a><a name="getVarInt"></a><span class='hs-comment'>-- | Decode an unsigned varint.</span>
+<a name="line-35"></a><span class='hs-definition'>getVarInt</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Parser</span> <span class='hs-conid'>Word64</span>
+<a name="line-36"></a><span class='hs-definition'>getVarInt</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>loop</span> <span class='hs-num'>1</span> <span class='hs-num'>0</span>
+<a name="line-37"></a>  <span class='hs-keyword'>where</span>
+<a name="line-38"></a>    <span class='hs-varid'>loop</span> <span class='hs-varop'>!</span><span class='hs-varid'>s</span> <span class='hs-varop'>!</span><span class='hs-varid'>n</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-39"></a>        <span class='hs-varid'>b</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>anyWord8</span>
+<a name="line-40"></a>        <span class='hs-keyword'>let</span> <span class='hs-varid'>n'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>n</span> <span class='hs-varop'>+</span> <span class='hs-varid'>s</span> <span class='hs-varop'>*</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-layout'>(</span><span class='hs-varid'>b</span> <span class='hs-varop'>.&amp;.</span> <span class='hs-num'>127</span><span class='hs-layout'>)</span>
+<a name="line-41"></a>        <span class='hs-keyword'>if</span> <span class='hs-layout'>(</span><span class='hs-varid'>b</span> <span class='hs-varop'>.&amp;.</span> <span class='hs-num'>128</span><span class='hs-layout'>)</span> <span class='hs-varop'>==</span> <span class='hs-num'>0</span>
+<a name="line-42"></a>            <span class='hs-keyword'>then</span> <span class='hs-varid'>return</span> <span class='hs-varid'>n'</span>
+<a name="line-43"></a>            <span class='hs-keyword'>else</span> <span class='hs-varid'>loop</span> <span class='hs-layout'>(</span><span class='hs-num'>128</span><span class='hs-varop'>*</span><span class='hs-varid'>s</span><span class='hs-layout'>)</span> <span class='hs-varid'>n'</span>
+<a name="line-44"></a>
+<a name="line-45"></a><a name="putVarInt"></a><span class='hs-comment'>-- | Encode a Word64.</span>
+<a name="line-46"></a><span class='hs-definition'>putVarInt</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Word64</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Builder</span>
+<a name="line-47"></a><span class='hs-definition'>putVarInt</span> <span class='hs-varid'>n</span>
+<a name="line-48"></a>    <span class='hs-keyglyph'>|</span> <span class='hs-varid'>n</span> <span class='hs-varop'>&lt;</span> <span class='hs-num'>128</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Builder</span><span class='hs-varop'>.</span><span class='hs-varid'>word8</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-varid'>n</span><span class='hs-layout'>)</span>
+<a name="line-49"></a>    <span class='hs-keyglyph'>|</span> <span class='hs-varid'>otherwise</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Builder</span><span class='hs-varop'>.</span><span class='hs-varid'>word8</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-varop'>$</span> <span class='hs-varid'>n</span> <span class='hs-varop'>.&amp;.</span> <span class='hs-num'>127</span> <span class='hs-varop'>.|.</span> <span class='hs-num'>128</span><span class='hs-layout'>)</span>
+<a name="line-50"></a>                      <span class='hs-varop'>&lt;&gt;</span> <span class='hs-varid'>putVarInt</span> <span class='hs-layout'>(</span><span class='hs-varid'>n</span> <span class='hs-varop'>`shiftR`</span> <span class='hs-num'>7</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Nodes.html b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Nodes.html
new file mode 100644
index 0000000..c5a5d69
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Nodes.html
@@ -0,0 +1,152 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/Nodes.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE FlexibleInstances #-}</span>
+<a name="line-16"></a><span class='hs-comment'>{-# LANGUAGE GeneralizedNewtypeDeriving #-}</span>
+<a name="line-17"></a><span class='hs-comment'>{-# LANGUAGE MultiParamTypeClasses #-}</span>
+<a name="line-18"></a><span class='hs-comment'>{-# LANGUAGE RankNTypes #-}</span>
+<a name="line-19"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables #-}</span>
+<a name="line-20"></a><span class='hs-comment'>{-# LANGUAGE TypeFamilies #-}</span>
+<a name="line-21"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Nodes</span> <span class='hs-keyword'>where</span>
+<a name="line-22"></a>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Applicative</span> <span class='hs-layout'>(</span><span class='hs-varid'>liftA2</span><span class='hs-layout'>,</span> <span class='hs-varid'>liftA3</span><span class='hs-layout'>)</span>
+<a name="line-24"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Strict</span> <span class='hs-layout'>(</span><span class='hs-conid'>Map</span><span class='hs-layout'>)</span>
+<a name="line-25"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Monoid</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varop'>&lt;&gt;</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-26"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Set</span> <span class='hs-layout'>(</span><span class='hs-conid'>Set</span><span class='hs-layout'>)</span>
+<a name="line-27"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>String</span> <span class='hs-layout'>(</span><span class='hs-conid'>IsString</span><span class='hs-layout'>)</span>
+<a name="line-28"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-layout'>(</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-29"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varop'>^.</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-30"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Strict</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Map</span>
+<a name="line-31"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Set</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Set</span>
+<a name="line-32"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Vector</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>V</span>
+<a name="line-33"></a>
+<a name="line-34"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Build</span>
+<a name="line-35"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Output</span>
+<a name="line-36"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span>
+<a name="line-37"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span>
+<a name="line-38"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Internal</span><span class='hs-varop'>.</span><span class='hs-conid'>FFI</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>FFI</span>
+<a name="line-39"></a>
+<a name="line-40"></a><a name="Fetchable"></a><span class='hs-comment'>-- | Types that contain ops which can be run.</span>
+<a name="line-41"></a><a name="Fetchable"></a><span class='hs-keyword'>class</span> <span class='hs-conid'>Nodes</span> <span class='hs-varid'>t</span> <span class='hs-keyword'>where</span>
+<a name="line-42"></a>    <span class='hs-varid'>getNodes</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>Set</span> <span class='hs-conid'>NodeName</span><span class='hs-layout'>)</span>
+<a name="line-43"></a>
+<a name="line-44"></a><a name="Fetchable"></a><span class='hs-comment'>-- | Types that tensor representations (e.g. 'Tensor', 'ControlNode') can be</span>
+<a name="line-45"></a><a name="Fetchable"></a><span class='hs-comment'>-- fetched into.</span>
+<a name="line-46"></a><a name="Fetchable"></a><span class='hs-comment'>--</span>
+<a name="line-47"></a><a name="Fetchable"></a><span class='hs-comment'>-- Includes collections of tensors (e.g. tuples).</span>
+<a name="line-48"></a><a name="Fetchable"></a><span class='hs-keyword'>class</span> <span class='hs-conid'>Nodes</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Fetchable</span> <span class='hs-varid'>t</span> <span class='hs-varid'>a</span> <span class='hs-keyword'>where</span>
+<a name="line-49"></a>    <span class='hs-varid'>getFetch</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>Fetch</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-50"></a>
+<a name="line-51"></a><a name="Fetch"></a><span class='hs-comment'>-- | Fetch action. Keeps track of what needs to be fetched and how to decode</span>
+<a name="line-52"></a><a name="Fetch"></a><span class='hs-comment'>-- the fetched data.</span>
+<a name="line-53"></a><a name="Fetch"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>Fetch</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Fetch</span>
+<a name="line-54"></a>          <span class='hs-layout'>{</span> <span class='hs-comment'>-- | Nodes to fetch</span>
+<a name="line-55"></a>            <span class='hs-varid'>fetches</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Set</span> <span class='hs-conid'>Text</span>
+<a name="line-56"></a>            <span class='hs-comment'>-- | Function to create an 'a' from the fetched data.</span>
+<a name="line-57"></a>          <span class='hs-layout'>,</span> <span class='hs-varid'>fetchRestore</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Map</span> <span class='hs-conid'>Text</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorData</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>a</span>
+<a name="line-58"></a>          <span class='hs-layout'>}</span>
+<a name="line-59"></a>
+<a name="line-60"></a><a name="instance%20Functor%20Fetch"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Functor</span> <span class='hs-conid'>Fetch</span> <span class='hs-keyword'>where</span>
+<a name="line-61"></a>    <span class='hs-varid'>fmap</span> <span class='hs-varid'>f</span> <span class='hs-layout'>(</span><span class='hs-conid'>Fetch</span> <span class='hs-varid'>fetch</span> <span class='hs-varid'>restore</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Fetch</span> <span class='hs-varid'>fetch</span> <span class='hs-layout'>(</span><span class='hs-varid'>f</span> <span class='hs-varop'>.</span> <span class='hs-varid'>restore</span><span class='hs-layout'>)</span>
+<a name="line-62"></a>
+<a name="line-63"></a><a name="instance%20Applicative%20Fetch"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Applicative</span> <span class='hs-conid'>Fetch</span> <span class='hs-keyword'>where</span>
+<a name="line-64"></a>    <span class='hs-varid'>pure</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Fetch</span> <span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>empty</span> <span class='hs-layout'>(</span><span class='hs-varid'>const</span> <span class='hs-varid'>x</span><span class='hs-layout'>)</span>
+<a name="line-65"></a>    <span class='hs-conid'>Fetch</span> <span class='hs-varid'>fetch</span> <span class='hs-varid'>restore</span> <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-conid'>Fetch</span> <span class='hs-varid'>fetch'</span> <span class='hs-varid'>restore'</span> <span class='hs-keyglyph'>=</span>
+<a name="line-66"></a>        <span class='hs-conid'>Fetch</span> <span class='hs-layout'>(</span><span class='hs-varid'>fetch</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-varid'>fetch'</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>restore</span> <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>restore'</span><span class='hs-layout'>)</span>
+<a name="line-67"></a>
+<a name="line-68"></a><a name="nodesUnion"></a><span class='hs-definition'>nodesUnion</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>Monoid</span> <span class='hs-varid'>b</span><span class='hs-layout'>,</span> <span class='hs-conid'>Traversable</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Applicative</span> <span class='hs-varid'>f</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-varid'>t</span> <span class='hs-layout'>(</span><span class='hs-varid'>f</span> <span class='hs-varid'>b</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>f</span> <span class='hs-varid'>b</span>
+<a name="line-69"></a><span class='hs-definition'>nodesUnion</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>fmap</span> <span class='hs-layout'>(</span><span class='hs-varid'>foldMap</span> <span class='hs-varid'>id</span><span class='hs-layout'>)</span> <span class='hs-varop'>.</span> <span class='hs-varid'>sequenceA</span>
+<a name="line-70"></a>
+<a name="line-71"></a><a name="instance%20Nodes%20(t1,%20t2)"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span><span class='hs-conid'>Nodes</span> <span class='hs-varid'>t1</span><span class='hs-layout'>,</span> <span class='hs-conid'>Nodes</span> <span class='hs-varid'>t2</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Nodes</span> <span class='hs-layout'>(</span><span class='hs-varid'>t1</span><span class='hs-layout'>,</span> <span class='hs-varid'>t2</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-72"></a>    <span class='hs-varid'>getNodes</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-varid'>y</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>nodesUnion</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>getNodes</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-varid'>getNodes</span> <span class='hs-varid'>y</span><span class='hs-keyglyph'>]</span>
+<a name="line-73"></a>
+<a name="line-74"></a><a name="instance%20Nodes%20(t1,%20t2,%20t3)"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span><span class='hs-conid'>Nodes</span> <span class='hs-varid'>t1</span><span class='hs-layout'>,</span> <span class='hs-conid'>Nodes</span> <span class='hs-varid'>t2</span><span class='hs-layout'>,</span> <span class='hs-conid'>Nodes</span> <span class='hs-varid'>t3</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Nodes</span> <span class='hs-layout'>(</span><span class='hs-varid'>t1</span><span class='hs-layout'>,</span> <span class='hs-varid'>t2</span><span class='hs-layout'>,</span> <span class='hs-varid'>t3</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-75"></a>    <span class='hs-varid'>getNodes</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-varid'>y</span><span class='hs-layout'>,</span> <span class='hs-varid'>z</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>nodesUnion</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>getNodes</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-varid'>getNodes</span> <span class='hs-varid'>y</span><span class='hs-layout'>,</span> <span class='hs-varid'>getNodes</span> <span class='hs-varid'>z</span><span class='hs-keyglyph'>]</span>
+<a name="line-76"></a>
+<a name="line-77"></a><a name="instance%20Fetchable%20(t1,%20t2)%20(a1,%20a2)"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span><span class='hs-conid'>Fetchable</span> <span class='hs-varid'>t1</span> <span class='hs-varid'>a1</span><span class='hs-layout'>,</span> <span class='hs-conid'>Fetchable</span> <span class='hs-varid'>t2</span> <span class='hs-varid'>a2</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Fetchable</span> <span class='hs-layout'>(</span><span class='hs-varid'>t1</span><span class='hs-layout'>,</span> <span class='hs-varid'>t2</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>a1</span><span class='hs-layout'>,</span> <span class='hs-varid'>a2</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-78"></a>    <span class='hs-varid'>getFetch</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-varid'>y</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>liftA2</span> <span class='hs-conid'>(,)</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>getFetch</span> <span class='hs-varid'>x</span> <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>getFetch</span> <span class='hs-varid'>y</span>
+<a name="line-79"></a>
+<a name="line-80"></a><a name="instance%20Fetchable%20(t1,%20t2,%20t3)%20(a1,%20a2,%20a3)"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span><span class='hs-conid'>Fetchable</span> <span class='hs-varid'>t1</span> <span class='hs-varid'>a1</span><span class='hs-layout'>,</span> <span class='hs-conid'>Fetchable</span> <span class='hs-varid'>t2</span> <span class='hs-varid'>a2</span><span class='hs-layout'>,</span> <span class='hs-conid'>Fetchable</span> <span class='hs-varid'>t3</span> <span class='hs-varid'>a3</span><span class='hs-layout'>)</span>
+<a name="line-81"></a>         <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Fetchable</span> <span class='hs-layout'>(</span><span class='hs-varid'>t1</span><span class='hs-layout'>,</span> <span class='hs-varid'>t2</span><span class='hs-layout'>,</span> <span class='hs-varid'>t3</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>a1</span><span class='hs-layout'>,</span> <span class='hs-varid'>a2</span><span class='hs-layout'>,</span> <span class='hs-varid'>a3</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-82"></a>    <span class='hs-varid'>getFetch</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-varid'>y</span><span class='hs-layout'>,</span> <span class='hs-varid'>z</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span>
+<a name="line-83"></a>        <span class='hs-varid'>liftA3</span> <span class='hs-conid'>(,,)</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>getFetch</span> <span class='hs-varid'>x</span> <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>getFetch</span> <span class='hs-varid'>y</span> <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>getFetch</span> <span class='hs-varid'>z</span>
+<a name="line-84"></a>
+<a name="line-85"></a><a name="instance%20Nodes%20%5bt%5d"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Nodes</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Nodes</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>t</span><span class='hs-keyglyph'>]</span> <span class='hs-keyword'>where</span>
+<a name="line-86"></a>    <span class='hs-varid'>getNodes</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>nodesUnion</span> <span class='hs-varop'>.</span> <span class='hs-varid'>map</span> <span class='hs-varid'>getNodes</span>
+<a name="line-87"></a>
+<a name="line-88"></a><a name="instance%20Fetchable%20%5bt%5d%20%5ba%5d"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Fetchable</span> <span class='hs-varid'>t</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Fetchable</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>t</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span> <span class='hs-keyword'>where</span>
+<a name="line-89"></a>    <span class='hs-varid'>getFetch</span> <span class='hs-varid'>ts</span>  <span class='hs-keyglyph'>=</span> <span class='hs-varid'>sequenceA</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>mapM</span> <span class='hs-varid'>getFetch</span> <span class='hs-varid'>ts</span>
+<a name="line-90"></a>
+<a name="line-91"></a><a name="instance%20Nodes%20ControlNode"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Nodes</span> <span class='hs-conid'>ControlNode</span> <span class='hs-keyword'>where</span>
+<a name="line-92"></a>    <span class='hs-varid'>getNodes</span> <span class='hs-layout'>(</span><span class='hs-conid'>ControlNode</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>singleton</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>getOrAddOp</span> <span class='hs-varid'>o</span>
+<a name="line-93"></a>
+<a name="line-94"></a><a name="instance%20Fetchable%20ControlNode%20a"></a><span class='hs-comment'>-- We use the constraint @(a ~ ())@ to help with type inference.  For example,</span>
+<a name="line-95"></a><a name="instance%20Fetchable%20ControlNode%20a"></a><span class='hs-comment'>-- if @t :: ControlNode@, then this constraint ensures that @run t :: Session</span>
+<a name="line-96"></a><a name="instance%20Fetchable%20ControlNode%20a"></a><span class='hs-comment'>-- ()@.  If we used @instance Fetchable ControlNode ()@ instead, then that</span>
+<a name="line-97"></a><a name="instance%20Fetchable%20ControlNode%20a"></a><span class='hs-comment'>-- expression would be ambiguous without explicitly specifying the return type.</span>
+<a name="line-98"></a><a name="instance%20Fetchable%20ControlNode%20a"></a><span class='hs-keyword'>instance</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>~</span> <span class='hs-conid'>()</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Fetchable</span> <span class='hs-conid'>ControlNode</span> <span class='hs-varid'>a</span> <span class='hs-keyword'>where</span>
+<a name="line-99"></a>    <span class='hs-varid'>getFetch</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>return</span> <span class='hs-varop'>$</span> <span class='hs-varid'>pure</span> <span class='hs-conid'>()</span>
+<a name="line-100"></a>
+<a name="line-101"></a><a name="instance%20Nodes%20(Tensor%20v%20a)"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Nodes</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-102"></a>    <span class='hs-varid'>getNodes</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>singleton</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>getOrAddOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>t</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>tensorOutput</span> <span class='hs-varop'>.</span> <span class='hs-varid'>outputOp</span><span class='hs-layout'>)</span>
+<a name="line-103"></a>
+<a name="line-104"></a><a name="fetchTensorList"></a><span class='hs-definition'>fetchTensorList</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>Fetch</span> <span class='hs-layout'>(</span><span class='hs-conid'>Shape</span><span class='hs-layout'>,</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-105"></a><span class='hs-definition'>fetchTensorList</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>fmap</span> <span class='hs-layout'>(</span><span class='hs-varid'>fmap</span> <span class='hs-conid'>V</span><span class='hs-varop'>.</span><span class='hs-varid'>toList</span><span class='hs-layout'>)</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>fetchTensorVector</span> <span class='hs-varid'>t</span>
+<a name="line-106"></a>
+<a name="line-107"></a><a name="fetchTensorVector"></a><span class='hs-definition'>fetchTensorVector</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varid'>v</span> <span class='hs-varop'>.</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span>
+<a name="line-108"></a>                  <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>Fetch</span> <span class='hs-layout'>(</span><span class='hs-conid'>Shape</span><span class='hs-layout'>,</span> <span class='hs-conid'>V</span><span class='hs-varop'>.</span><span class='hs-conid'>Vector</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-109"></a><span class='hs-definition'>fetchTensorVector</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-keyword'>_</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-110"></a>    <span class='hs-varid'>outputName</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>renderOutput</span> <span class='hs-varid'>o</span>
+<a name="line-111"></a>    <span class='hs-varid'>return</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Fetch</span> <span class='hs-layout'>(</span><span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>singleton</span> <span class='hs-varid'>outputName</span><span class='hs-layout'>)</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-varid'>tensors</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-112"></a>        <span class='hs-keyword'>let</span> <span class='hs-varid'>tensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tensors</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.!</span> <span class='hs-varid'>outputName</span>
+<a name="line-113"></a>            <span class='hs-varid'>shape</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Shape</span> <span class='hs-varop'>$</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-varid'>tensorDataDimensions</span> <span class='hs-varid'>tensorData</span>
+<a name="line-114"></a>            <span class='hs-varid'>vec</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>decodeTensorData</span> <span class='hs-varop'>$</span> <span class='hs-conid'>TensorData</span> <span class='hs-varid'>tensorData</span>
+<a name="line-115"></a>
+<a name="line-116"></a>            <span class='hs-varid'>expectedType</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-117"></a>            <span class='hs-varid'>actualType</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-varid'>tensorDataType</span> <span class='hs-varid'>tensorData</span>
+<a name="line-118"></a>            <span class='hs-varid'>badTypeError</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-varop'>$</span> <span class='hs-str'>"Bad tensor type: expected "</span>
+<a name="line-119"></a>                                   <span class='hs-varop'>++</span> <span class='hs-varid'>show</span> <span class='hs-varid'>expectedType</span>
+<a name="line-120"></a>                                   <span class='hs-varop'>++</span> <span class='hs-str'>", got "</span>
+<a name="line-121"></a>                                   <span class='hs-varop'>++</span> <span class='hs-varid'>show</span> <span class='hs-varid'>actualType</span>
+<a name="line-122"></a>        <span class='hs-keyword'>in</span> <span class='hs-keyword'>if</span> <span class='hs-varid'>expectedType</span> <span class='hs-varop'>/=</span> <span class='hs-varid'>actualType</span>
+<a name="line-123"></a>               <span class='hs-keyword'>then</span> <span class='hs-varid'>badTypeError</span>
+<a name="line-124"></a>               <span class='hs-keyword'>else</span> <span class='hs-layout'>(</span><span class='hs-varid'>shape</span><span class='hs-layout'>,</span> <span class='hs-varid'>vec</span><span class='hs-layout'>)</span>
+<a name="line-125"></a>
+<a name="line-126"></a><a name="instance%20Fetchable%20(Tensor%20v%20a)%20(V.Vector%20a')"></a><span class='hs-comment'>-- The constraint "a ~ a'" means that the input/output of fetch can constrain</span>
+<a name="line-127"></a><a name="instance%20Fetchable%20(Tensor%20v%20a)%20(V.Vector%20a')"></a><span class='hs-comment'>-- the TensorType of each other.</span>
+<a name="line-128"></a><a name="instance%20Fetchable%20(Tensor%20v%20a)%20(V.Vector%20a')"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span><span class='hs-layout'>,</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>~</span> <span class='hs-varid'>a'</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Fetchable</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-conid'>V</span><span class='hs-varop'>.</span><span class='hs-conid'>Vector</span> <span class='hs-varid'>a'</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-129"></a>    <span class='hs-varid'>getFetch</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>fmap</span> <span class='hs-varid'>snd</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>fetchTensorVector</span> <span class='hs-varid'>t</span>
+<a name="line-130"></a>
+<a name="line-131"></a><a name="Scalar"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>Scalar</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Scalar</span> <span class='hs-layout'>{</span><span class='hs-varid'>unScalar</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span><span class='hs-layout'>}</span>
+<a name="line-132"></a>    <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Eq</span><span class='hs-layout'>,</span> <span class='hs-conid'>Ord</span><span class='hs-layout'>,</span> <span class='hs-conid'>Num</span><span class='hs-layout'>,</span> <span class='hs-conid'>Fractional</span><span class='hs-layout'>,</span> <span class='hs-conid'>Floating</span><span class='hs-layout'>,</span> <span class='hs-conid'>Real</span><span class='hs-layout'>,</span> <span class='hs-conid'>RealFloat</span><span class='hs-layout'>,</span>
+<a name="line-133"></a>              <span class='hs-conid'>RealFrac</span><span class='hs-layout'>,</span> <span class='hs-conid'>IsString</span><span class='hs-layout'>)</span>
+<a name="line-134"></a>
+<a name="line-135"></a><a name="instance%20Fetchable%20(Tensor%20v%20a)%20(Scalar%20a')"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span><span class='hs-layout'>,</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>~</span> <span class='hs-varid'>a'</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Fetchable</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-conid'>Scalar</span> <span class='hs-varid'>a'</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-136"></a>    <span class='hs-varid'>getFetch</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>fmap</span> <span class='hs-layout'>(</span><span class='hs-conid'>Scalar</span> <span class='hs-varop'>.</span> <span class='hs-varid'>headFromSingleton</span> <span class='hs-varop'>.</span> <span class='hs-varid'>snd</span><span class='hs-layout'>)</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>fetchTensorList</span> <span class='hs-varid'>t</span>
+<a name="line-137"></a>      <span class='hs-keyword'>where</span>
+<a name="line-138"></a>        <span class='hs-varid'>headFromSingleton</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>x</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span>
+<a name="line-139"></a>        <span class='hs-varid'>headFromSingleton</span> <span class='hs-varid'>xs</span>
+<a name="line-140"></a>            <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-varop'>$</span> <span class='hs-str'>"Unable to extract singleton from tensor of length "</span>
+<a name="line-141"></a>                          <span class='hs-varop'>++</span> <span class='hs-varid'>show</span> <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>xs</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Orphans.html b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Orphans.html
new file mode 100644
index 0000000..a0f10f0
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Orphans.html
@@ -0,0 +1,57 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/Orphans.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a>
+<a name="line-16"></a><span class='hs-comment'>{-# LANGUAGE StandaloneDeriving #-}</span>
+<a name="line-17"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-orphans #-}</span>
+<a name="line-18"></a><span class='hs-comment'>-- Orphan instances for certain proto messages/enums, used internally.</span>
+<a name="line-19"></a><span class='hs-comment'>-- TODO(judahjacobson): consider making proto-lens generate some or all of</span>
+<a name="line-20"></a><span class='hs-comment'>-- these automatically; or, alternately, make new Haskell datatypes.</span>
+<a name="line-21"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Orphans</span><span class='hs-conid'>()</span> <span class='hs-keyword'>where</span>
+<a name="line-22"></a>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span>
+<a name="line-24"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>AttrValue</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-25"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>AttrValue'ListValue</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-26"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>NameAttrList</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-27"></a>    <span class='hs-layout'>)</span>
+<a name="line-28"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span>
+<a name="line-29"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>NodeDef</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-30"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>ResourceHandle</span>
+<a name="line-31"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>ResourceHandle</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-32"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span>
+<a name="line-33"></a>    <span class='hs-layout'>(</span><span class='hs-conid'>TensorProto</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-34"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span>
+<a name="line-35"></a>    <span class='hs-layout'>(</span><span class='hs-conid'>TensorShapeProto</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorShapeProto'Dim</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-36"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span> <span class='hs-layout'>(</span><span class='hs-conid'>DataType</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-37"></a>
+<a name="line-38"></a><span class='hs-keyword'>deriving</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Ord</span> <span class='hs-conid'>AttrValue</span>
+<a name="line-39"></a><span class='hs-keyword'>deriving</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Ord</span> <span class='hs-conid'>AttrValue'ListValue</span>
+<a name="line-40"></a><span class='hs-keyword'>deriving</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Ord</span> <span class='hs-conid'>DataType</span>
+<a name="line-41"></a><span class='hs-keyword'>deriving</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Ord</span> <span class='hs-conid'>NameAttrList</span>
+<a name="line-42"></a><span class='hs-keyword'>deriving</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Ord</span> <span class='hs-conid'>NodeDef</span>
+<a name="line-43"></a><span class='hs-keyword'>deriving</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Ord</span> <span class='hs-conid'>ResourceHandle</span>
+<a name="line-44"></a><span class='hs-keyword'>deriving</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Ord</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-45"></a><span class='hs-keyword'>deriving</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Ord</span> <span class='hs-conid'>TensorShapeProto</span>
+<a name="line-46"></a><span class='hs-keyword'>deriving</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Ord</span> <span class='hs-conid'>TensorShapeProto'Dim</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Output.html b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Output.html
new file mode 100644
index 0000000..6a8759a
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Output.html
@@ -0,0 +1,167 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/Output.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE GeneralizedNewtypeDeriving #-}</span>
+<a name="line-16"></a><span class='hs-comment'>{-# LANGUAGE Rank2Types #-}</span>
+<a name="line-17"></a><span class='hs-comment'>{-# LANGUAGE OverloadedStrings #-}</span>
+<a name="line-18"></a>
+<a name="line-19"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Output</span>
+<a name="line-20"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>ControlNode</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-21"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Device</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-22"></a>    <span class='hs-comment'>-- * Ops</span>
+<a name="line-23"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>NodeName</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-24"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Op</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-25"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>opUnrendered</span>
+<a name="line-26"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>OpDef</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-27"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>opName</span>
+<a name="line-28"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>opType</span>
+<a name="line-29"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>opAttr</span>
+<a name="line-30"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>opInputs</span>
+<a name="line-31"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>opControlInputs</span>
+<a name="line-32"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>OpType</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-33"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>OutputIx</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-34"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Output</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-35"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>output</span>
+<a name="line-36"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>outputIndex</span>
+<a name="line-37"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>outputOp</span>
+<a name="line-38"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>PendingNodeName</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-39"></a>    <span class='hs-layout'>)</span>  <span class='hs-keyword'>where</span>
+<a name="line-40"></a>
+<a name="line-41"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Strict</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Map</span>
+<a name="line-42"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>TextFormat</span> <span class='hs-layout'>(</span><span class='hs-varid'>showMessage</span><span class='hs-layout'>)</span>
+<a name="line-43"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>String</span> <span class='hs-layout'>(</span><span class='hs-conid'>IsString</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-44"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-layout'>(</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-45"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Text</span>
+<a name="line-46"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span> <span class='hs-layout'>(</span><span class='hs-conid'>Lens'</span><span class='hs-layout'>,</span> <span class='hs-conid'>Traversal'</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>.~</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>&amp;</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>^.</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-47"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span> <span class='hs-layout'>(</span><span class='hs-varid'>lens</span><span class='hs-layout'>)</span>
+<a name="line-48"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span> <span class='hs-layout'>(</span><span class='hs-conid'>AttrValue</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-49"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span> <span class='hs-layout'>(</span><span class='hs-conid'>NodeDef</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span>
+<a name="line-50"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-layout'>(</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-51"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span> <span class='hs-layout'>(</span><span class='hs-conid'>Attribute</span><span class='hs-layout'>,</span> <span class='hs-varid'>attrLens</span><span class='hs-layout'>)</span>
+<a name="line-52"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Orphans</span> <span class='hs-conid'>()</span>
+<a name="line-53"></a>
+<a name="line-54"></a><a name="ControlNode"></a><span class='hs-comment'>-- | A type of graph node which has no outputs. These nodes are</span>
+<a name="line-55"></a><a name="ControlNode"></a><span class='hs-comment'>-- valuable for causing side effects when they are run.</span>
+<a name="line-56"></a><a name="ControlNode"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>ControlNode</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>ControlNode</span> <span class='hs-layout'>{</span> <span class='hs-varid'>unControlNode</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Op</span> <span class='hs-layout'>}</span>
+<a name="line-57"></a>
+<a name="line-58"></a><a name="OpType"></a><span class='hs-comment'>-- | The type of op of a node in the graph.  This corresponds to the proto field</span>
+<a name="line-59"></a><a name="OpType"></a><span class='hs-comment'>-- NodeDef.op.</span>
+<a name="line-60"></a><a name="OpType"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>OpType</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OpType</span> <span class='hs-layout'>{</span> <span class='hs-varid'>unOpType</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span> <span class='hs-layout'>}</span>
+<a name="line-61"></a>    <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Eq</span><span class='hs-layout'>,</span> <span class='hs-conid'>Ord</span><span class='hs-layout'>,</span> <span class='hs-conid'>Show</span><span class='hs-layout'>)</span>
+<a name="line-62"></a>
+<a name="line-63"></a><a name="instance%20IsString%20OpType"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>IsString</span> <span class='hs-conid'>OpType</span> <span class='hs-keyword'>where</span>
+<a name="line-64"></a>    <span class='hs-varid'>fromString</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OpType</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>pack</span>
+<a name="line-65"></a>
+<a name="line-66"></a><a name="Output"></a><span class='hs-comment'>-- | An output of a TensorFlow node.</span>
+<a name="line-67"></a><a name="Output"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>Output</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Output</span> <span class='hs-varop'>!</span><span class='hs-conid'>OutputIx</span> <span class='hs-varop'>!</span><span class='hs-conid'>Op</span>
+<a name="line-68"></a>    <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Eq</span><span class='hs-layout'>,</span> <span class='hs-conid'>Ord</span><span class='hs-layout'>,</span> <span class='hs-conid'>Show</span><span class='hs-layout'>)</span>
+<a name="line-69"></a>
+<a name="line-70"></a><a name="output"></a><span class='hs-definition'>output</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OutputIx</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Op</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Output</span>
+<a name="line-71"></a><span class='hs-definition'>output</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Output</span>
+<a name="line-72"></a>
+<a name="line-73"></a><a name="outputOp"></a><span class='hs-definition'>outputOp</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>Output</span> <span class='hs-conid'>Op</span>
+<a name="line-74"></a><span class='hs-definition'>outputOp</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-layout'>(</span><span class='hs-conid'>Output</span> <span class='hs-keyword'>_</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-layout'>(</span><span class='hs-conid'>Output</span> <span class='hs-varid'>i</span> <span class='hs-keyword'>_</span><span class='hs-layout'>)</span> <span class='hs-varid'>o</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Output</span> <span class='hs-varid'>i</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span>
+<a name="line-75"></a>
+<a name="line-76"></a><a name="outputIndex"></a><span class='hs-definition'>outputIndex</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>Output</span> <span class='hs-conid'>OutputIx</span>
+<a name="line-77"></a><span class='hs-definition'>outputIndex</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-layout'>(</span><span class='hs-conid'>Output</span> <span class='hs-varid'>i</span> <span class='hs-keyword'>_</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>i</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-layout'>(</span><span class='hs-conid'>Output</span> <span class='hs-keyword'>_</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span> <span class='hs-varid'>i</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Output</span> <span class='hs-varid'>i</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span>
+<a name="line-78"></a>
+<a name="line-79"></a><a name="OutputIx"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>OutputIx</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OutputIx</span> <span class='hs-layout'>{</span> <span class='hs-varid'>unOutputIx</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int</span> <span class='hs-layout'>}</span>
+<a name="line-80"></a>    <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Eq</span><span class='hs-layout'>,</span> <span class='hs-conid'>Ord</span><span class='hs-layout'>,</span> <span class='hs-conid'>Num</span><span class='hs-layout'>,</span> <span class='hs-conid'>Enum</span><span class='hs-layout'>,</span> <span class='hs-conid'>Show</span><span class='hs-layout'>)</span>
+<a name="line-81"></a>
+<a name="line-82"></a><a name="Device"></a><span class='hs-comment'>-- | A device that a node can be assigned to.</span>
+<a name="line-83"></a><a name="Device"></a><span class='hs-comment'>-- There's a naming convention where the device names</span>
+<a name="line-84"></a><a name="Device"></a><span class='hs-comment'>-- are constructed from job and replica names.</span>
+<a name="line-85"></a><a name="Device"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>Device</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Device</span> <span class='hs-layout'>{</span><span class='hs-varid'>deviceName</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span><span class='hs-layout'>}</span>
+<a name="line-86"></a>    <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Eq</span><span class='hs-layout'>,</span> <span class='hs-conid'>Ord</span><span class='hs-layout'>,</span> <span class='hs-conid'>IsString</span><span class='hs-layout'>)</span>
+<a name="line-87"></a>
+<a name="line-88"></a><a name="instance%20Show%20Device"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Show</span> <span class='hs-conid'>Device</span> <span class='hs-keyword'>where</span>
+<a name="line-89"></a>    <span class='hs-varid'>show</span> <span class='hs-layout'>(</span><span class='hs-conid'>Device</span> <span class='hs-varid'>d</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>show</span> <span class='hs-varid'>d</span>
+<a name="line-90"></a>
+<a name="line-91"></a><a name="Op"></a><span class='hs-comment'>-- | The representation of a node in a TensorFlow graph.</span>
+<a name="line-92"></a><a name="Op"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>Op</span>
+<a name="line-93"></a>    <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Rendered</span> <span class='hs-varop'>!</span><span class='hs-conid'>NodeDef</span>  <span class='hs-comment'>-- ^ Properties are fixed, including the</span>
+<a name="line-94"></a>                         <span class='hs-comment'>-- device, name, and scope.</span>
+<a name="line-95"></a>    <span class='hs-keyglyph'>|</span> <span class='hs-conid'>Unrendered</span> <span class='hs-varop'>!</span><span class='hs-conid'>OpDef</span>  <span class='hs-comment'>-- ^ Properties are not fixed, and may change depending</span>
+<a name="line-96"></a>                         <span class='hs-comment'>-- on which context this op is rendered in.</span>
+<a name="line-97"></a>    <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Eq</span><span class='hs-layout'>,</span> <span class='hs-conid'>Ord</span><span class='hs-layout'>)</span>
+<a name="line-98"></a>
+<a name="line-99"></a><a name="instance%20Show%20Op"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Show</span> <span class='hs-conid'>Op</span> <span class='hs-keyword'>where</span>
+<a name="line-100"></a>    <span class='hs-varid'>show</span> <span class='hs-layout'>(</span><span class='hs-conid'>Rendered</span> <span class='hs-varid'>n</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Rendered "</span> <span class='hs-varop'>++</span> <span class='hs-varid'>showMessage</span> <span class='hs-varid'>n</span>
+<a name="line-101"></a>    <span class='hs-varid'>show</span> <span class='hs-layout'>(</span><span class='hs-conid'>Unrendered</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Unrendered "</span> <span class='hs-varop'>++</span> <span class='hs-varid'>show</span> <span class='hs-layout'>(</span><span class='hs-varid'>o</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>opName</span><span class='hs-layout'>)</span>
+<a name="line-102"></a>
+<a name="line-103"></a><a name="opUnrendered"></a><span class='hs-comment'>-- | Traverse on the 'Unrendered' of an 'Op'.</span>
+<a name="line-104"></a><span class='hs-comment'>--</span>
+<a name="line-105"></a><span class='hs-comment'>-- Same implementation as _Left.</span>
+<a name="line-106"></a><span class='hs-definition'>opUnrendered</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Traversal'</span> <span class='hs-conid'>Op</span> <span class='hs-conid'>OpDef</span>
+<a name="line-107"></a><span class='hs-definition'>opUnrendered</span> <span class='hs-varid'>f</span> <span class='hs-layout'>(</span><span class='hs-conid'>Unrendered</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Unrendered</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>f</span> <span class='hs-varid'>a</span>
+<a name="line-108"></a><span class='hs-definition'>opUnrendered</span> <span class='hs-keyword'>_</span> <span class='hs-layout'>(</span><span class='hs-conid'>Rendered</span> <span class='hs-varid'>b</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>pure</span> <span class='hs-layout'>(</span><span class='hs-conid'>Rendered</span> <span class='hs-varid'>b</span><span class='hs-layout'>)</span>
+<a name="line-109"></a>
+<a name="line-110"></a><a name="OpDef"></a><span class='hs-comment'>-- | Op definition. This corresponds somewhat to the 'NodeDef' proto.</span>
+<a name="line-111"></a><a name="OpDef"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OpDef</span>
+<a name="line-112"></a>    <span class='hs-layout'>{</span> <span class='hs-sel'>_opName</span> <span class='hs-keyglyph'>::</span> <span class='hs-varop'>!</span><span class='hs-conid'>PendingNodeName</span>
+<a name="line-113"></a>    <span class='hs-layout'>,</span> <span class='hs-sel'>_opType</span> <span class='hs-keyglyph'>::</span> <span class='hs-varop'>!</span><span class='hs-conid'>OpType</span>
+<a name="line-114"></a>    <span class='hs-layout'>,</span> <span class='hs-sel'>_opAttrs</span> <span class='hs-keyglyph'>::</span> <span class='hs-varop'>!</span><span class='hs-layout'>(</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>Text</span> <span class='hs-conid'>AttrValue</span><span class='hs-layout'>)</span>
+<a name="line-115"></a>    <span class='hs-layout'>,</span> <span class='hs-sel'>_opInputs</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Output</span><span class='hs-keyglyph'>]</span>
+<a name="line-116"></a>    <span class='hs-layout'>,</span> <span class='hs-sel'>_opControlInputs</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>NodeName</span><span class='hs-keyglyph'>]</span>
+<a name="line-117"></a>    <span class='hs-layout'>}</span>  <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Eq</span><span class='hs-layout'>,</span> <span class='hs-conid'>Ord</span><span class='hs-layout'>)</span>
+<a name="line-118"></a>
+<a name="line-119"></a><a name="PendingNodeName"></a><span class='hs-comment'>-- | The name specified for an unrendered Op.  If an Op has an</span>
+<a name="line-120"></a><a name="PendingNodeName"></a><span class='hs-comment'>-- ImplicitName, it will be assigned based on the opType plus a</span>
+<a name="line-121"></a><a name="PendingNodeName"></a><span class='hs-comment'>-- unique identifier.  Does not contain the "scope" prefix.</span>
+<a name="line-122"></a><a name="PendingNodeName"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>PendingNodeName</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>ExplicitName</span> <span class='hs-varop'>!</span><span class='hs-conid'>Text</span> <span class='hs-keyglyph'>|</span> <span class='hs-conid'>ImplicitName</span>
+<a name="line-123"></a>    <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Eq</span><span class='hs-layout'>,</span> <span class='hs-conid'>Ord</span><span class='hs-layout'>,</span> <span class='hs-conid'>Show</span><span class='hs-layout'>)</span>
+<a name="line-124"></a>
+<a name="line-125"></a><a name="NodeName"></a><span class='hs-comment'>-- | The name of a node in the graph.  This corresponds to the proto field</span>
+<a name="line-126"></a><a name="NodeName"></a><span class='hs-comment'>-- NodeDef.name.  Includes the scope prefix (if any) and a unique identifier</span>
+<a name="line-127"></a><a name="NodeName"></a><span class='hs-comment'>-- (if the node was implicitly named).</span>
+<a name="line-128"></a><a name="NodeName"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>NodeName</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NodeName</span> <span class='hs-layout'>{</span> <span class='hs-varid'>unNodeName</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span> <span class='hs-layout'>}</span>
+<a name="line-129"></a>    <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Eq</span><span class='hs-layout'>,</span> <span class='hs-conid'>Ord</span><span class='hs-layout'>,</span> <span class='hs-conid'>Show</span><span class='hs-layout'>)</span>
+<a name="line-130"></a>
+<a name="line-131"></a><a name="opName"></a><span class='hs-definition'>opName</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>OpDef</span> <span class='hs-conid'>PendingNodeName</span>
+<a name="line-132"></a><span class='hs-definition'>opName</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_opName</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>o</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>o</span> <span class='hs-layout'>{</span><span class='hs-sel'>_opName</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-133"></a>
+<a name="line-134"></a><a name="opType"></a><span class='hs-definition'>opType</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>OpDef</span> <span class='hs-conid'>OpType</span>
+<a name="line-135"></a><span class='hs-definition'>opType</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_opType</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>o</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>o</span> <span class='hs-layout'>{</span> <span class='hs-sel'>_opType</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-136"></a>
+<a name="line-137"></a><a name="opAttr"></a><span class='hs-definition'>opAttr</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Attribute</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>OpDef</span> <span class='hs-varid'>a</span>
+<a name="line-138"></a><span class='hs-definition'>opAttr</span> <span class='hs-varid'>n</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_opAttrs</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>o</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>o</span> <span class='hs-layout'>{</span><span class='hs-sel'>_opAttrs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-139"></a>              <span class='hs-varop'>.</span> <span class='hs-varid'>lens</span> <span class='hs-layout'>(</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>findWithDefault</span> <span class='hs-varid'>def</span> <span class='hs-varid'>n</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>flip</span> <span class='hs-layout'>(</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>insert</span> <span class='hs-varid'>n</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-140"></a>              <span class='hs-varop'>.</span> <span class='hs-varid'>attrLens</span>
+<a name="line-141"></a>
+<a name="line-142"></a><a name="opInputs"></a><span class='hs-definition'>opInputs</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Output</span><span class='hs-keyglyph'>]</span>
+<a name="line-143"></a><span class='hs-definition'>opInputs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_opInputs</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>o</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>o</span> <span class='hs-layout'>{</span><span class='hs-sel'>_opInputs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-144"></a>
+<a name="line-145"></a><a name="opControlInputs"></a><span class='hs-definition'>opControlInputs</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>NodeName</span><span class='hs-keyglyph'>]</span>
+<a name="line-146"></a><span class='hs-definition'>opControlInputs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_opControlInputs</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>o</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>o</span> <span class='hs-layout'>{</span><span class='hs-sel'>_opControlInputs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-147"></a>
+<a name="line-148"></a><a name="instance%20IsString%20Output"></a><span class='hs-comment'>-- TODO(gnezdo): IsString instance is weird and we should move that</span>
+<a name="line-149"></a><a name="instance%20IsString%20Output"></a><span class='hs-comment'>-- code into a Build function</span>
+<a name="line-150"></a><a name="instance%20IsString%20Output"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>IsString</span> <span class='hs-conid'>Output</span> <span class='hs-keyword'>where</span>
+<a name="line-151"></a>    <span class='hs-varid'>fromString</span> <span class='hs-varid'>s</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>case</span> <span class='hs-varid'>break</span> <span class='hs-layout'>(</span><span class='hs-varop'>==</span><span class='hs-chr'>':'</span><span class='hs-layout'>)</span> <span class='hs-varid'>s</span> <span class='hs-keyword'>of</span>
+<a name="line-152"></a>        <span class='hs-layout'>(</span><span class='hs-varid'>n</span><span class='hs-layout'>,</span> <span class='hs-chr'>':'</span><span class='hs-conop'>:</span><span class='hs-varid'>ixStr</span><span class='hs-layout'>)</span>
+<a name="line-153"></a>            <span class='hs-keyglyph'>|</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-varid'>ix</span><span class='hs-layout'>,</span> <span class='hs-str'>""</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>read</span> <span class='hs-varid'>ixStr</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Output</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromInteger</span> <span class='hs-varid'>ix</span><span class='hs-layout'>)</span> <span class='hs-varop'>$</span> <span class='hs-varid'>assigned</span> <span class='hs-varid'>n</span>
+<a name="line-154"></a>        <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Output</span> <span class='hs-num'>0</span> <span class='hs-varop'>$</span> <span class='hs-varid'>assigned</span> <span class='hs-varid'>s</span>
+<a name="line-155"></a>        <span class='hs-keyword'>where</span> <span class='hs-varid'>assigned</span> <span class='hs-varid'>n</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Rendered</span> <span class='hs-varop'>$</span> <span class='hs-varid'>def</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>name</span> <span class='hs-varop'>.~</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>pack</span> <span class='hs-varid'>n</span>
+<a name="line-156"></a>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Session.html b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Session.html
new file mode 100644
index 0000000..be4612a
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Session.html
@@ -0,0 +1,213 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/Session.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE GeneralizedNewtypeDeriving #-}</span>
+<a name="line-16"></a><span class='hs-comment'>{-# LANGUAGE Rank2Types #-}</span>
+<a name="line-17"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables #-}</span>
+<a name="line-18"></a><span class='hs-comment'>{-# LANGUAGE TupleSections #-}</span>
+<a name="line-19"></a>
+<a name="line-20"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Session</span> <span class='hs-layout'>(</span>
+<a name="line-21"></a>    <span class='hs-conid'>Session</span><span class='hs-layout'>,</span>
+<a name="line-22"></a>    <span class='hs-comment'>-- * Opaque value created via 'sessionConfig' and 'sessionTarget'.</span>
+<a name="line-23"></a>    <span class='hs-conid'>SessionOption</span><span class='hs-layout'>,</span>
+<a name="line-24"></a>    <span class='hs-varid'>sessionConfig</span><span class='hs-layout'>,</span>
+<a name="line-25"></a>    <span class='hs-varid'>sessionTarget</span><span class='hs-layout'>,</span>
+<a name="line-26"></a>    <span class='hs-varid'>runSession</span><span class='hs-layout'>,</span>
+<a name="line-27"></a>    <span class='hs-varid'>runSessionWithOptions</span><span class='hs-layout'>,</span>
+<a name="line-28"></a>    <span class='hs-varid'>build</span><span class='hs-layout'>,</span>
+<a name="line-29"></a>    <span class='hs-varid'>buildAnd</span><span class='hs-layout'>,</span>
+<a name="line-30"></a>    <span class='hs-varid'>buildWithSummary</span><span class='hs-layout'>,</span>
+<a name="line-31"></a>    <span class='hs-varid'>extend</span><span class='hs-layout'>,</span>
+<a name="line-32"></a>    <span class='hs-varid'>addGraphDef</span><span class='hs-layout'>,</span>
+<a name="line-33"></a>    <span class='hs-varid'>run</span><span class='hs-layout'>,</span>
+<a name="line-34"></a>    <span class='hs-varid'>runWithFeeds</span><span class='hs-layout'>,</span>
+<a name="line-35"></a>    <span class='hs-varid'>run_</span><span class='hs-layout'>,</span>
+<a name="line-36"></a>    <span class='hs-varid'>runWithFeeds_</span><span class='hs-layout'>,</span>
+<a name="line-37"></a>    <span class='hs-varid'>asyncProdNodes</span><span class='hs-layout'>,</span>
+<a name="line-38"></a>    <span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-39"></a>
+<a name="line-40"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Monad</span> <span class='hs-layout'>(</span><span class='hs-varid'>forever</span><span class='hs-layout'>,</span> <span class='hs-varid'>unless</span><span class='hs-layout'>,</span> <span class='hs-varid'>void</span><span class='hs-layout'>)</span>
+<a name="line-41"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Monad</span><span class='hs-varop'>.</span><span class='hs-conid'>IO</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span> <span class='hs-layout'>(</span><span class='hs-conid'>MonadIO</span><span class='hs-layout'>,</span> <span class='hs-varid'>liftIO</span><span class='hs-layout'>)</span>
+<a name="line-42"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Monad</span><span class='hs-varop'>.</span><span class='hs-conid'>Trans</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span> <span class='hs-layout'>(</span><span class='hs-varid'>lift</span><span class='hs-layout'>)</span>
+<a name="line-43"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Monad</span><span class='hs-varop'>.</span><span class='hs-conid'>Trans</span><span class='hs-varop'>.</span><span class='hs-conid'>Reader</span> <span class='hs-layout'>(</span><span class='hs-conid'>ReaderT</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-varid'>ask</span><span class='hs-layout'>,</span> <span class='hs-varid'>asks</span><span class='hs-layout'>)</span>
+<a name="line-44"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-layout'>(</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>
+<a name="line-45"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Functor</span><span class='hs-varop'>.</span><span class='hs-conid'>Identity</span> <span class='hs-layout'>(</span><span class='hs-varid'>runIdentity</span><span class='hs-layout'>)</span>
+<a name="line-46"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Strict</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Map</span>
+<a name="line-47"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Set</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Set</span>
+<a name="line-48"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Set</span> <span class='hs-layout'>(</span><span class='hs-conid'>Set</span><span class='hs-layout'>)</span>
+<a name="line-49"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Encoding</span> <span class='hs-layout'>(</span><span class='hs-varid'>encodeUtf8</span><span class='hs-layout'>)</span>
+<a name="line-50"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span> <span class='hs-layout'>(</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-51"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varop'>&amp;</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>.~</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-52"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Graph</span> <span class='hs-layout'>(</span><span class='hs-varid'>node</span><span class='hs-layout'>)</span>
+<a name="line-53"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Protobuf</span><span class='hs-varop'>.</span><span class='hs-conid'>Config</span> <span class='hs-layout'>(</span><span class='hs-conid'>ConfigProto</span><span class='hs-layout'>)</span>
+<a name="line-54"></a>
+<a name="line-55"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Build</span>
+<a name="line-56"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Internal</span><span class='hs-varop'>.</span><span class='hs-conid'>FFI</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>FFI</span>
+<a name="line-57"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Internal</span><span class='hs-varop'>.</span><span class='hs-conid'>Raw</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Raw</span>
+<a name="line-58"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Nodes</span>
+<a name="line-59"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Output</span> <span class='hs-layout'>(</span><span class='hs-conid'>NodeName</span><span class='hs-layout'>,</span> <span class='hs-varid'>unNodeName</span><span class='hs-layout'>)</span>
+<a name="line-60"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span>
+<a name="line-61"></a>
+<a name="line-62"></a><a name="SessionState"></a><span class='hs-comment'>-- Common state threaded through the session.</span>
+<a name="line-63"></a><a name="SessionState"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>SessionState</span>
+<a name="line-64"></a>    <span class='hs-keyglyph'>=</span> <span class='hs-conid'>SessionState</span> <span class='hs-layout'>{</span>
+<a name="line-65"></a>          <span class='hs-varid'>rawSession</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-conid'>Session</span>
+<a name="line-66"></a>        <span class='hs-layout'>,</span> <span class='hs-varid'>asyncCollector</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span>
+<a name="line-67"></a>          <span class='hs-comment'>-- ^ Starts the given action concurrently.</span>
+<a name="line-68"></a>        <span class='hs-layout'>}</span>
+<a name="line-69"></a>
+<a name="line-70"></a><a name="Session"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>Session</span> <span class='hs-varid'>a</span>
+<a name="line-71"></a>    <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Session</span> <span class='hs-layout'>(</span><span class='hs-conid'>ReaderT</span> <span class='hs-conid'>SessionState</span> <span class='hs-layout'>(</span><span class='hs-conid'>BuildT</span> <span class='hs-conid'>IO</span><span class='hs-layout'>)</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-72"></a>    <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Functor</span><span class='hs-layout'>,</span> <span class='hs-conid'>Applicative</span><span class='hs-layout'>,</span> <span class='hs-conid'>Monad</span><span class='hs-layout'>,</span> <span class='hs-conid'>MonadIO</span><span class='hs-layout'>)</span>
+<a name="line-73"></a>
+<a name="line-74"></a><a name="runSession"></a><span class='hs-comment'>-- | Run 'Session' actions in a new TensorFlow session.</span>
+<a name="line-75"></a><span class='hs-definition'>runSession</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Session</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span>
+<a name="line-76"></a><span class='hs-definition'>runSession</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>runSessionWithOptions</span> <span class='hs-conid'>[]</span>
+<a name="line-77"></a>
+<a name="line-78"></a><a name="SessionOption"></a><span class='hs-comment'>-- | Setting of an option for the session (see 'runSessionWithOptions').</span>
+<a name="line-79"></a><a name="SessionOption"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>SessionOption</span> <span class='hs-keyglyph'>=</span>
+<a name="line-80"></a>    <span class='hs-conid'>SessionOption</span> <span class='hs-layout'>{</span> <span class='hs-varid'>unSesssionOption</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Raw</span><span class='hs-varop'>.</span><span class='hs-conid'>SessionOptions</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>()</span> <span class='hs-layout'>}</span>
+<a name="line-81"></a>
+<a name="line-82"></a><a name="sessionTarget"></a><span class='hs-comment'>-- | Target can be: "local", ip:port, host:port.</span>
+<a name="line-83"></a><span class='hs-comment'>-- The set of supported factories depends on the linked in libraries.</span>
+<a name="line-84"></a><span class='hs-comment'>-- REQUIRES "//learning/brain/public:tensorflow_remote" dependency for the binary.</span>
+<a name="line-85"></a><span class='hs-definition'>sessionTarget</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>ByteString</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>SessionOption</span>
+<a name="line-86"></a><span class='hs-definition'>sessionTarget</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>SessionOption</span> <span class='hs-varop'>.</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-varid'>setSessionTarget</span>
+<a name="line-87"></a>
+<a name="line-88"></a><a name="sessionConfig"></a><span class='hs-comment'>-- | Uses the specified config for the created session.</span>
+<a name="line-89"></a><span class='hs-definition'>sessionConfig</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>SessionOption</span>
+<a name="line-90"></a><span class='hs-definition'>sessionConfig</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>SessionOption</span> <span class='hs-varop'>.</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-varid'>setSessionConfig</span>
+<a name="line-91"></a>
+<a name="line-92"></a><a name="runSessionWithOptions"></a><span class='hs-comment'>-- | Run 'Session' actions in a new TensorFlow session created with</span>
+<a name="line-93"></a><span class='hs-comment'>-- the given option setter actions ('sessionTarget', 'sessionConfig').</span>
+<a name="line-94"></a><span class='hs-definition'>runSessionWithOptions</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>SessionOption</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Session</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span>
+<a name="line-95"></a><span class='hs-definition'>runSessionWithOptions</span> <span class='hs-varid'>options</span> <span class='hs-layout'>(</span><span class='hs-conid'>Session</span> <span class='hs-varid'>m</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span>
+<a name="line-96"></a>    <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-varid'>withSession</span> <span class='hs-varid'>applyOptions</span> <span class='hs-varop'>$</span>
+<a name="line-97"></a>        <span class='hs-keyglyph'>\</span><span class='hs-keyword'>as</span> <span class='hs-varid'>rs</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>evalBuildT</span> <span class='hs-layout'>(</span><span class='hs-varid'>runReaderT</span> <span class='hs-varid'>m</span> <span class='hs-layout'>(</span><span class='hs-conid'>SessionState</span> <span class='hs-varid'>rs</span> <span class='hs-keyword'>as</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-98"></a>  <span class='hs-keyword'>where</span> <span class='hs-varid'>applyOptions</span> <span class='hs-varid'>opt</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>mapM_</span> <span class='hs-layout'>(</span><span class='hs-varop'>`unSesssionOption`</span> <span class='hs-varid'>opt</span><span class='hs-layout'>)</span> <span class='hs-varid'>options</span>
+<a name="line-99"></a>
+<a name="line-100"></a><a name="build"></a><span class='hs-comment'>-- | Lift a 'Build' action into a 'Session', including any explicit op</span>
+<a name="line-101"></a><span class='hs-comment'>-- renderings.</span>
+<a name="line-102"></a><span class='hs-definition'>build</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Build</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Session</span> <span class='hs-varid'>a</span>
+<a name="line-103"></a><span class='hs-definition'>build</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Session</span> <span class='hs-varop'>.</span> <span class='hs-varid'>lift</span> <span class='hs-varop'>.</span> <span class='hs-varid'>hoistBuildT</span> <span class='hs-layout'>(</span><span class='hs-varid'>return</span> <span class='hs-varop'>.</span> <span class='hs-varid'>runIdentity</span><span class='hs-layout'>)</span>
+<a name="line-104"></a>
+<a name="line-105"></a><a name="buildWithSummary"></a><span class='hs-comment'>-- | Lift a 'Build' action into a 'Session', including any explicit op</span>
+<a name="line-106"></a><span class='hs-comment'>-- renderings. Returns the merged summary ops which can be used for</span>
+<a name="line-107"></a><span class='hs-comment'>-- logging, see 'TensorFlow.Logging.build' for a convenient wrapper.</span>
+<a name="line-108"></a><span class='hs-definition'>buildWithSummary</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Build</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Session</span> <span class='hs-layout'>(</span><span class='hs-varid'>a</span><span class='hs-layout'>,</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>SummaryTensor</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-109"></a><span class='hs-definition'>buildWithSummary</span> <span class='hs-varid'>b</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Session</span> <span class='hs-varop'>$</span> <span class='hs-varid'>lift</span> <span class='hs-varop'>$</span> <span class='hs-conid'>(,)</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>v</span> <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>collectAllSummaries</span>
+<a name="line-110"></a>  <span class='hs-keyword'>where</span> <span class='hs-varid'>v</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>BuildT</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span>
+<a name="line-111"></a>        <span class='hs-varid'>v</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>hoistBuildT</span> <span class='hs-layout'>(</span><span class='hs-varid'>return</span> <span class='hs-varop'>.</span> <span class='hs-varid'>runIdentity</span><span class='hs-layout'>)</span> <span class='hs-varid'>b</span>
+<a name="line-112"></a>
+<a name="line-113"></a><a name="extend"></a><span class='hs-comment'>-- | Add all pending rendered nodes to the TensorFlow graph and runs</span>
+<a name="line-114"></a><span class='hs-comment'>-- any pending initializers.</span>
+<a name="line-115"></a><span class='hs-comment'>--</span>
+<a name="line-116"></a><span class='hs-comment'>-- Note that run, runWithFeeds, etc. will all call this function implicitly.</span>
+<a name="line-117"></a><span class='hs-definition'>extend</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Session</span> <span class='hs-conid'>()</span>
+<a name="line-118"></a><span class='hs-definition'>extend</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-119"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>withSessionWhen</span> <span class='hs-varid'>vs</span> <span class='hs-varid'>action</span> <span class='hs-keyglyph'>=</span>
+<a name="line-120"></a>            <span class='hs-varid'>unless</span> <span class='hs-layout'>(</span><span class='hs-varid'>null</span> <span class='hs-varid'>vs</span><span class='hs-layout'>)</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Session</span> <span class='hs-layout'>(</span><span class='hs-varid'>asks</span> <span class='hs-varid'>rawSession</span><span class='hs-layout'>)</span> <span class='hs-varop'>&gt;&gt;=</span> <span class='hs-varid'>action</span>
+<a name="line-121"></a>    <span class='hs-varid'>nodesToExtend</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>build</span> <span class='hs-varid'>flushNodeBuffer</span>
+<a name="line-122"></a>    <span class='hs-varid'>withSessionWhen</span> <span class='hs-varid'>nodesToExtend</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-varid'>session</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-123"></a>        <span class='hs-varid'>liftIO</span> <span class='hs-varop'>$</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-varid'>extendGraph</span> <span class='hs-varid'>session</span>
+<a name="line-124"></a>               <span class='hs-varop'>$</span> <span class='hs-varid'>def</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>node</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>nodesToExtend</span>
+<a name="line-125"></a>    <span class='hs-comment'>-- Now that all the nodes are created, run the initializers.</span>
+<a name="line-126"></a>    <span class='hs-varid'>initializers</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>build</span> <span class='hs-varid'>flushInitializers</span>
+<a name="line-127"></a>    <span class='hs-varid'>withSessionWhen</span> <span class='hs-varid'>initializers</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-varid'>session</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-128"></a>        <span class='hs-varid'>void</span> <span class='hs-varop'>$</span> <span class='hs-varid'>liftIO</span> <span class='hs-varop'>$</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-varid'>run</span> <span class='hs-varid'>session</span> <span class='hs-conid'>[]</span> <span class='hs-conid'>[]</span> <span class='hs-layout'>(</span><span class='hs-varid'>toNodeNames</span> <span class='hs-varid'>initializers</span><span class='hs-layout'>)</span>
+<a name="line-129"></a>
+<a name="line-130"></a><a name="buildAnd"></a><span class='hs-comment'>-- | Helper combinator for doing something with the result of a 'Build' action.</span>
+<a name="line-131"></a><span class='hs-comment'>-- Example usage:</span>
+<a name="line-132"></a><span class='hs-comment'>--</span>
+<a name="line-133"></a><span class='hs-comment'>-- &gt; buildAnd run :: Fetchable t a =&gt; Build t -&gt; Session a</span>
+<a name="line-134"></a><span class='hs-definition'>buildAnd</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Session</span> <span class='hs-varid'>b</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Session</span> <span class='hs-varid'>b</span>
+<a name="line-135"></a><span class='hs-definition'>buildAnd</span> <span class='hs-varid'>f</span> <span class='hs-varid'>m</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>build</span> <span class='hs-varid'>m</span> <span class='hs-varop'>&gt;&gt;=</span> <span class='hs-varid'>f</span>
+<a name="line-136"></a>
+<a name="line-137"></a><a name="run"></a><span class='hs-comment'>-- | Run a subgraph 't', rendering any dependent nodes that aren't already</span>
+<a name="line-138"></a><span class='hs-comment'>-- rendered, and fetch the corresponding values for 'a'.</span>
+<a name="line-139"></a><span class='hs-definition'>run</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Fetchable</span> <span class='hs-varid'>t</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Session</span> <span class='hs-varid'>a</span>
+<a name="line-140"></a><span class='hs-definition'>run</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>runWithFeeds</span> <span class='hs-conid'>[]</span>
+<a name="line-141"></a>
+<a name="line-142"></a><a name="runWithFeeds"></a><span class='hs-comment'>-- | Run a subgraph 't', rendering any dependent nodes that aren't already</span>
+<a name="line-143"></a><span class='hs-comment'>-- rendered, feed the given input values, and fetch the corresponding result</span>
+<a name="line-144"></a><span class='hs-comment'>-- values for 'a'.</span>
+<a name="line-145"></a><span class='hs-definition'>runWithFeeds</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Fetchable</span> <span class='hs-varid'>t</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Feed</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Session</span> <span class='hs-varid'>a</span>
+<a name="line-146"></a><span class='hs-definition'>runWithFeeds</span> <span class='hs-varid'>feeds</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-147"></a>    <span class='hs-varid'>ns</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>build</span> <span class='hs-varop'>$</span> <span class='hs-varid'>getNodes</span> <span class='hs-varid'>t</span>
+<a name="line-148"></a>    <span class='hs-comment'>-- Note that this call to "fetch" shouldn't affect the following "extend"</span>
+<a name="line-149"></a>    <span class='hs-comment'>-- call, since all nodes in t and its inputs/deps will be rendered by the</span>
+<a name="line-150"></a>    <span class='hs-comment'>-- above call to getNodes.</span>
+<a name="line-151"></a>    <span class='hs-varid'>fetch</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>build</span> <span class='hs-varop'>$</span> <span class='hs-varid'>getFetch</span> <span class='hs-varid'>t</span>
+<a name="line-152"></a>    <span class='hs-varid'>runFetchWithFeeds</span> <span class='hs-varid'>feeds</span> <span class='hs-varid'>ns</span> <span class='hs-varid'>fetch</span>
+<a name="line-153"></a>
+<a name="line-154"></a><a name="runFetchWithFeeds"></a><span class='hs-definition'>runFetchWithFeeds</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Feed</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Set</span> <span class='hs-conid'>NodeName</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Fetch</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Session</span> <span class='hs-varid'>a</span>
+<a name="line-155"></a><span class='hs-definition'>runFetchWithFeeds</span> <span class='hs-varid'>feeds</span> <span class='hs-varid'>target</span> <span class='hs-layout'>(</span><span class='hs-conid'>Fetch</span> <span class='hs-varid'>fetch</span> <span class='hs-varid'>restore</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-156"></a>    <span class='hs-varid'>extend</span>
+<a name="line-157"></a>    <span class='hs-varid'>feeds'</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>build</span> <span class='hs-varop'>$</span> <span class='hs-varid'>fixFeeds</span> <span class='hs-varid'>feeds</span>
+<a name="line-158"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>fetchNames</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>encodeUtf8</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>toList</span> <span class='hs-varid'>fetch</span>
+<a name="line-159"></a>        <span class='hs-varid'>targetNames</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>toNodeNames</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>toList</span> <span class='hs-varid'>target</span>
+<a name="line-160"></a>    <span class='hs-varid'>session</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>Session</span> <span class='hs-layout'>(</span><span class='hs-varid'>asks</span> <span class='hs-varid'>rawSession</span><span class='hs-layout'>)</span>
+<a name="line-161"></a>    <span class='hs-varid'>runResult</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>liftIO</span> <span class='hs-varop'>$</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-varid'>run</span> <span class='hs-varid'>session</span>
+<a name="line-162"></a>                                  <span class='hs-varid'>feeds'</span>
+<a name="line-163"></a>                                  <span class='hs-varid'>fetchNames</span>
+<a name="line-164"></a>                                  <span class='hs-varid'>targetNames</span>
+<a name="line-165"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>resultTensorsMap</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span> <span class='hs-varop'>$</span> <span class='hs-varid'>zip</span> <span class='hs-layout'>(</span><span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>toList</span> <span class='hs-varid'>fetch</span><span class='hs-layout'>)</span> <span class='hs-varid'>runResult</span>
+<a name="line-166"></a>    <span class='hs-varid'>return</span> <span class='hs-varop'>$</span> <span class='hs-varid'>restore</span> <span class='hs-varid'>resultTensorsMap</span>
+<a name="line-167"></a>
+<a name="line-168"></a><a name="toNodeNames"></a><span class='hs-definition'>toNodeNames</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>NodeName</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>ByteString</span><span class='hs-keyglyph'>]</span>
+<a name="line-169"></a><span class='hs-definition'>toNodeNames</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>map</span> <span class='hs-layout'>(</span><span class='hs-varid'>encodeUtf8</span> <span class='hs-varop'>.</span> <span class='hs-varid'>unNodeName</span><span class='hs-layout'>)</span>
+<a name="line-170"></a>
+<a name="line-171"></a><a name="run_"></a><span class='hs-comment'>-- | Run a subgraph 't', rendering and extending any dependent nodes that aren't</span>
+<a name="line-172"></a><span class='hs-comment'>-- already rendered.  This behaves like 'run' except that it doesn't do any</span>
+<a name="line-173"></a><span class='hs-comment'>-- fetches.</span>
+<a name="line-174"></a><span class='hs-definition'>run_</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Nodes</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Session</span> <span class='hs-conid'>()</span>
+<a name="line-175"></a><span class='hs-definition'>run_</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>runWithFeeds_</span> <span class='hs-conid'>[]</span>
+<a name="line-176"></a>
+<a name="line-177"></a><a name="runWithFeeds_"></a><span class='hs-comment'>-- | Run a subgraph 't', rendering any dependent nodes that aren't already</span>
+<a name="line-178"></a><span class='hs-comment'>-- rendered, feed the given input values, and fetch the corresponding result</span>
+<a name="line-179"></a><span class='hs-comment'>-- values for 'a'.  This behaves like 'runWithFeeds' except that it doesn't do</span>
+<a name="line-180"></a><span class='hs-comment'>-- any fetches.</span>
+<a name="line-181"></a><span class='hs-definition'>runWithFeeds_</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Nodes</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Feed</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Session</span> <span class='hs-conid'>()</span>
+<a name="line-182"></a><span class='hs-definition'>runWithFeeds_</span> <span class='hs-varid'>feeds</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-183"></a>    <span class='hs-varid'>ns</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>build</span> <span class='hs-varop'>$</span> <span class='hs-varid'>getNodes</span> <span class='hs-varid'>t</span>
+<a name="line-184"></a>    <span class='hs-varid'>runFetchWithFeeds</span> <span class='hs-varid'>feeds</span> <span class='hs-varid'>ns</span> <span class='hs-layout'>(</span><span class='hs-varid'>pure</span> <span class='hs-conid'>()</span><span class='hs-layout'>)</span>
+<a name="line-185"></a>
+<a name="line-186"></a><a name="fixFeeds"></a><span class='hs-definition'>fixFeeds</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Feed</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>,</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorData</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span>
+<a name="line-187"></a><span class='hs-definition'>fixFeeds</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>mapM</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-layout'>(</span><span class='hs-conid'>Feed</span> <span class='hs-varid'>o</span> <span class='hs-varid'>d</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-layout'>,</span><span class='hs-varid'>d</span><span class='hs-layout'>)</span> <span class='hs-varop'>.</span> <span class='hs-varid'>encodeUtf8</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>renderOutput</span> <span class='hs-varid'>o</span>
+<a name="line-188"></a>
+<a name="line-189"></a><a name="asyncProdNodes"></a><span class='hs-comment'>-- | Starts a concurrent thread which evaluates the given Nodes</span>
+<a name="line-190"></a><span class='hs-comment'>-- forever until runSession exits or an exception occurs. Graph</span>
+<a name="line-191"></a><span class='hs-comment'>-- extension happens synchronously, but the resultant run proceeds as</span>
+<a name="line-192"></a><span class='hs-comment'>-- a separate thread.</span>
+<a name="line-193"></a><span class='hs-definition'>asyncProdNodes</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Nodes</span> <span class='hs-varid'>t</span>
+<a name="line-194"></a>                  <span class='hs-keyglyph'>=&gt;</span> <span class='hs-varid'>t</span>  <span class='hs-comment'>-- ^ Node to evaluate concurrently.</span>
+<a name="line-195"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Session</span> <span class='hs-conid'>()</span>
+<a name="line-196"></a><span class='hs-definition'>asyncProdNodes</span> <span class='hs-varid'>nodes</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-197"></a>    <span class='hs-varid'>target</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>build</span> <span class='hs-layout'>(</span><span class='hs-varid'>getNodes</span> <span class='hs-varid'>nodes</span><span class='hs-layout'>)</span>
+<a name="line-198"></a>    <span class='hs-varid'>extend</span>
+<a name="line-199"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>targetNames</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>toNodeNames</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>toList</span> <span class='hs-varid'>target</span>
+<a name="line-200"></a>    <span class='hs-varid'>state</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>Session</span> <span class='hs-varid'>ask</span>
+<a name="line-201"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>loop</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>forever</span> <span class='hs-layout'>(</span><span class='hs-varid'>void</span> <span class='hs-layout'>(</span><span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-varid'>run</span> <span class='hs-layout'>(</span><span class='hs-varid'>rawSession</span> <span class='hs-varid'>state</span><span class='hs-layout'>)</span> <span class='hs-conid'>[]</span> <span class='hs-conid'>[]</span> <span class='hs-varid'>targetNames</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-202"></a>    <span class='hs-varid'>liftIO</span> <span class='hs-layout'>(</span><span class='hs-varid'>asyncCollector</span> <span class='hs-varid'>state</span> <span class='hs-varid'>loop</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Tensor.html b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Tensor.html
new file mode 100644
index 0000000..282d865
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Tensor.html
@@ -0,0 +1,96 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/Tensor.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE FlexibleInstances #-}</span>
+<a name="line-16"></a><span class='hs-comment'>{-# LANGUAGE GADTs #-}</span>
+<a name="line-17"></a><span class='hs-comment'>{-# LANGUAGE OverloadedStrings #-}</span>
+<a name="line-18"></a><span class='hs-comment'>{-# LANGUAGE Rank2Types #-}</span>
+<a name="line-19"></a>
+<a name="line-20"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span> <span class='hs-keyword'>where</span>
+<a name="line-21"></a>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>String</span> <span class='hs-layout'>(</span><span class='hs-conid'>IsString</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Text</span>
+<a name="line-24"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span> <span class='hs-layout'>(</span><span class='hs-conid'>Lens'</span><span class='hs-layout'>,</span> <span class='hs-conid'>Traversal'</span><span class='hs-layout'>)</span>
+<a name="line-25"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span> <span class='hs-layout'>(</span><span class='hs-varid'>lens</span><span class='hs-layout'>)</span>
+<a name="line-26"></a>
+<a name="line-27"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Output</span> <span class='hs-layout'>(</span><span class='hs-conid'>Output</span><span class='hs-layout'>,</span> <span class='hs-varid'>outputOp</span><span class='hs-layout'>,</span> <span class='hs-varid'>opUnrendered</span><span class='hs-layout'>,</span> <span class='hs-varid'>opAttr</span><span class='hs-layout'>)</span>
+<a name="line-28"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorData</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-conid'>Attribute</span><span class='hs-layout'>)</span>
+<a name="line-29"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Internal</span><span class='hs-varop'>.</span><span class='hs-conid'>FFI</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>FFI</span>
+<a name="line-30"></a>
+<a name="line-31"></a><a name="Tensor"></a><span class='hs-comment'>-- | A named output of a TensorFlow operation.</span>
+<a name="line-32"></a><a name="Tensor"></a><span class='hs-comment'>--</span>
+<a name="line-33"></a><a name="Tensor"></a><span class='hs-comment'>-- The type parameter @a@ is the type of the elements in the 'Tensor'.  The</span>
+<a name="line-34"></a><a name="Tensor"></a><span class='hs-comment'>-- parameter @v@ is either 'Value' or 'Ref', depending on whether the graph is</span>
+<a name="line-35"></a><a name="Tensor"></a><span class='hs-comment'>-- treating this op output as an immutable 'Value' or a stateful 'Ref' (e.g., a</span>
+<a name="line-36"></a><a name="Tensor"></a><span class='hs-comment'>-- variable).  Note that a @Tensor Ref@ can be casted into a @Tensor Value@ via</span>
+<a name="line-37"></a><a name="Tensor"></a><span class='hs-comment'>-- 'value'.</span>
+<a name="line-38"></a><a name="Tensor"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Tensor</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorKind</span> <span class='hs-varid'>v</span><span class='hs-layout'>)</span> <span class='hs-conid'>Output</span>
+<a name="line-39"></a>
+<a name="line-40"></a><a name="Value"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>Value</span>
+<a name="line-41"></a><a name="Ref"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>Ref</span>
+<a name="line-42"></a>
+<a name="line-43"></a><a name="TensorKind"></a><span class='hs-comment'>-- | This class provides a runtime switch on whether a 'Tensor' should be</span>
+<a name="line-44"></a><a name="TensorKind"></a><span class='hs-comment'>-- treated as a 'Value' or as a 'Ref'.</span>
+<a name="line-45"></a><a name="TensorKind"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>TensorKind</span> <span class='hs-varid'>v</span> <span class='hs-keyword'>where</span>
+<a name="line-46"></a>  <span class='hs-conid'>ValueKind</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>TensorKind</span> <span class='hs-conid'>Value</span>
+<a name="line-47"></a>  <span class='hs-conid'>RefKind</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>TensorKind</span> <span class='hs-conid'>Ref</span>
+<a name="line-48"></a>
+<a name="line-49"></a><a name="tensorKind"></a><span class='hs-definition'>tensorKind</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorKind</span> <span class='hs-varid'>v</span><span class='hs-layout'>)</span>
+<a name="line-50"></a><span class='hs-definition'>tensorKind</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-keyword'>_</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>v</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-keyword'>_</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span> <span class='hs-varid'>v</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span>
+<a name="line-51"></a>
+<a name="line-52"></a><a name="tensorOutput"></a><span class='hs-definition'>tensorOutput</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-conid'>Output</span>
+<a name="line-53"></a><span class='hs-definition'>tensorOutput</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-keyword'>_</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-keyword'>_</span><span class='hs-layout'>)</span> <span class='hs-varid'>o</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span>
+<a name="line-54"></a>
+<a name="line-55"></a><a name="tensorAttr"></a><span class='hs-comment'>-- TODO: Come up with a better API for handling attributes.</span>
+<a name="line-56"></a><span class='hs-comment'>-- | Lens for the attributes of a tensor.</span>
+<a name="line-57"></a><span class='hs-comment'>--</span>
+<a name="line-58"></a><span class='hs-comment'>-- Only valid if the tensor has not yet been rendered. If the tensor has been</span>
+<a name="line-59"></a><span class='hs-comment'>-- rendered, the traversal will be over nothing (nothing can be read or</span>
+<a name="line-60"></a><span class='hs-comment'>-- written).</span>
+<a name="line-61"></a><span class='hs-definition'>tensorAttr</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Attribute</span> <span class='hs-varid'>attr</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Traversal'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-varid'>attr</span>
+<a name="line-62"></a><span class='hs-definition'>tensorAttr</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tensorOutput</span> <span class='hs-varop'>.</span> <span class='hs-varid'>outputOp</span> <span class='hs-varop'>.</span> <span class='hs-varid'>opUnrendered</span> <span class='hs-varop'>.</span> <span class='hs-varid'>opAttr</span> <span class='hs-varid'>x</span>
+<a name="line-63"></a>
+<a name="line-64"></a><a name="value"></a><span class='hs-comment'>-- | Cast a 'Tensor *' into a 'Tensor Value'. Common usage is to cast a</span>
+<a name="line-65"></a><span class='hs-comment'>-- Ref into Value. This behaves like a no-op.</span>
+<a name="line-66"></a><span class='hs-definition'>value</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span>
+<a name="line-67"></a><span class='hs-definition'>value</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-keyword'>_</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>ValueKind</span> <span class='hs-varid'>o</span>
+<a name="line-68"></a>
+<a name="line-69"></a><a name="Feed"></a><span class='hs-comment'>-- | A pair of a 'Tensor' and some data that should be fed into that 'Tensor'</span>
+<a name="line-70"></a><a name="Feed"></a><span class='hs-comment'>-- when running the graph.</span>
+<a name="line-71"></a><a name="Feed"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>Feed</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Feed</span> <span class='hs-conid'>Output</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorData</span>
+<a name="line-72"></a>
+<a name="line-73"></a><a name="feed"></a><span class='hs-comment'>-- | Create a 'Feed' for feeding the given data into a 'Tensor' when running</span>
+<a name="line-74"></a><span class='hs-comment'>-- the graph.</span>
+<a name="line-75"></a><span class='hs-comment'>--</span>
+<a name="line-76"></a><span class='hs-comment'>-- Note that if a 'Tensor' is rendered, its identity may change; so feeding the</span>
+<a name="line-77"></a><span class='hs-comment'>-- rendered 'Tensor' may be different than feeding the original 'Tensor'.</span>
+<a name="line-78"></a><span class='hs-definition'>feed</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>TensorData</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Feed</span>
+<a name="line-79"></a><span class='hs-definition'>feed</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-keyword'>_</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorData</span> <span class='hs-varid'>td</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Feed</span> <span class='hs-varid'>o</span> <span class='hs-varid'>td</span>
+<a name="line-80"></a>
+<a name="line-81"></a><a name="tensorFromName"></a><span class='hs-comment'>-- | Create a 'Tensor' for a given name.  This can be used to reference nodes</span>
+<a name="line-82"></a><span class='hs-comment'>-- in a 'GraphDef' that was loaded via 'addGraphDef'.</span>
+<a name="line-83"></a><span class='hs-comment'>-- TODO(judahjacobson): add more safety checks here.</span>
+<a name="line-84"></a><span class='hs-definition'>tensorFromName</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>TensorKind</span> <span class='hs-varid'>v</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span>
+<a name="line-85"></a><span class='hs-definition'>tensorFromName</span> <span class='hs-varid'>v</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varop'>.</span> <span class='hs-varid'>fromString</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>unpack</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Types.html b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Types.html
new file mode 100644
index 0000000..0f3bf98
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/src/TensorFlow-Types.html
@@ -0,0 +1,393 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/Types.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE ConstraintKinds #-}</span>
+<a name="line-16"></a><span class='hs-comment'>{-# LANGUAGE DataKinds #-}</span>
+<a name="line-17"></a><span class='hs-comment'>{-# LANGUAGE FlexibleContexts #-}</span>
+<a name="line-18"></a><span class='hs-comment'>{-# LANGUAGE FlexibleInstances #-}</span>
+<a name="line-19"></a><span class='hs-comment'>{-# LANGUAGE OverloadedStrings #-}</span>
+<a name="line-20"></a><span class='hs-comment'>{-# LANGUAGE RankNTypes #-}</span>
+<a name="line-21"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables #-}</span>
+<a name="line-22"></a><span class='hs-comment'>{-# LANGUAGE TypeFamilies #-}</span>
+<a name="line-23"></a><span class='hs-comment'>{-# LANGUAGE TypeOperators #-}</span>
+<a name="line-24"></a><span class='hs-comment'>-- We use UndecidableInstances for type families with recursive definitions</span>
+<a name="line-25"></a><span class='hs-comment'>-- like "\\".  Those instances will terminate since each equation unwraps one</span>
+<a name="line-26"></a><span class='hs-comment'>-- cons cell of a type-level list.</span>
+<a name="line-27"></a><span class='hs-comment'>{-# LANGUAGE UndecidableInstances #-}</span>
+<a name="line-28"></a>
+<a name="line-29"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span>
+<a name="line-30"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>TensorType</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-31"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>TensorData</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-32"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Shape</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-33"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>protoShape</span>
+<a name="line-34"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Attribute</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-35"></a>    <span class='hs-comment'>-- * Type constraints</span>
+<a name="line-36"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span>
+<a name="line-37"></a>    <span class='hs-layout'>,</span> <span class='hs-keyword'>type</span> <span class='hs-layout'>(</span><span class='hs-varop'>/=</span><span class='hs-layout'>)</span>
+<a name="line-38"></a>    <span class='hs-comment'>-- ** Implementation of constraints</span>
+<a name="line-39"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>TypeError</span>
+<a name="line-40"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>ExcludedCase</span>
+<a name="line-41"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>TensorTypes</span>
+<a name="line-42"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>NoneOf</span>
+<a name="line-43"></a>    <span class='hs-layout'>,</span> <span class='hs-keyword'>type</span> <span class='hs-layout'>(</span><span class='hs-varop'>\\</span><span class='hs-layout'>)</span>
+<a name="line-44"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Delete</span>
+<a name="line-45"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>AllTensorTypes</span>
+<a name="line-46"></a>    <span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-47"></a>
+<a name="line-48"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-layout'>(</span><span class='hs-conid'>Complex</span><span class='hs-layout'>)</span>
+<a name="line-49"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-layout'>(</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-50"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span> <span class='hs-layout'>(</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-51"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Monoid</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varop'>&lt;&gt;</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-52"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span> <span class='hs-layout'>(</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Word64</span><span class='hs-layout'>)</span>
+<a name="line-53"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Foreign</span><span class='hs-varop'>.</span><span class='hs-conid'>Storable</span> <span class='hs-layout'>(</span><span class='hs-conid'>Storable</span><span class='hs-layout'>)</span>
+<a name="line-54"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>GHC</span><span class='hs-varop'>.</span><span class='hs-conid'>Exts</span> <span class='hs-layout'>(</span><span class='hs-conid'>Constraint</span><span class='hs-layout'>,</span> <span class='hs-conid'>IsList</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-55"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span> <span class='hs-layout'>(</span><span class='hs-conid'>Lens'</span><span class='hs-layout'>,</span> <span class='hs-varid'>view</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>&amp;</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>.~</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-56"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span> <span class='hs-layout'>(</span><span class='hs-varid'>iso</span><span class='hs-layout'>)</span>
+<a name="line-57"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Attoparsec</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Atto</span>
+<a name="line-58"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-layout'>(</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>
+<a name="line-59"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>B</span>
+<a name="line-60"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>Builder</span> <span class='hs-layout'>(</span><span class='hs-conid'>Builder</span><span class='hs-layout'>)</span>
+<a name="line-61"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>Builder</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Builder</span>
+<a name="line-62"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>Lazy</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>L</span>
+<a name="line-63"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Vector</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>V</span>
+<a name="line-64"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Vector</span><span class='hs-varop'>.</span><span class='hs-conid'>Storable</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>S</span>
+<a name="line-65"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span>
+<a name="line-66"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>AttrValue</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-67"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>AttrValue'ListValue</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-68"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>b</span>
+<a name="line-69"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>f</span>
+<a name="line-70"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>i</span>
+<a name="line-71"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>s</span>
+<a name="line-72"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>list</span>
+<a name="line-73"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>type'</span>
+<a name="line-74"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>shape</span>
+<a name="line-75"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>tensor</span>
+<a name="line-76"></a>    <span class='hs-layout'>)</span>
+<a name="line-77"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Tensor</span>
+<a name="line-78"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>TensorProto</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-79"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>floatVal</span>
+<a name="line-80"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>doubleVal</span>
+<a name="line-81"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>intVal</span>
+<a name="line-82"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>stringVal</span>
+<a name="line-83"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>int64Val</span>
+<a name="line-84"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>stringVal</span>
+<a name="line-85"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>boolVal</span>
+<a name="line-86"></a>    <span class='hs-layout'>)</span>
+<a name="line-87"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span>
+<a name="line-88"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>TensorShapeProto</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-89"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>dim</span>
+<a name="line-90"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>size</span>
+<a name="line-91"></a>    <span class='hs-layout'>)</span>
+<a name="line-92"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span> <span class='hs-layout'>(</span><span class='hs-conid'>DataType</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-93"></a>
+<a name="line-94"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Internal</span><span class='hs-varop'>.</span><span class='hs-conid'>VarInt</span> <span class='hs-layout'>(</span><span class='hs-varid'>getVarInt</span><span class='hs-layout'>,</span> <span class='hs-varid'>putVarInt</span><span class='hs-layout'>)</span>
+<a name="line-95"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Internal</span><span class='hs-varop'>.</span><span class='hs-conid'>FFI</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>FFI</span>
+<a name="line-96"></a>
+<a name="line-97"></a><a name="TensorData"></a><span class='hs-comment'>-- | Data about a tensor that is encoded for the TensorFlow APIs.</span>
+<a name="line-98"></a><a name="TensorData"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>TensorData</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TensorData</span> <span class='hs-layout'>{</span> <span class='hs-varid'>unTensorData</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorData</span> <span class='hs-layout'>}</span>
+<a name="line-99"></a>
+<a name="line-100"></a><a name="TensorData"></a><span class='hs-comment'>-- | The class of scalar types supported by tensorflow.</span>
+<a name="line-101"></a><a name="TensorData"></a><span class='hs-keyword'>class</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span> <span class='hs-keyword'>where</span>
+<a name="line-102"></a>    <span class='hs-varid'>tensorType</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>DataType</span>
+<a name="line-103"></a>    <span class='hs-varid'>tensorRefType</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>DataType</span>
+<a name="line-104"></a>    <span class='hs-varid'>tensorVal</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span>
+<a name="line-105"></a>    <span class='hs-comment'>-- | Decode the bytes of a TensorData into a Vector.</span>
+<a name="line-106"></a>    <span class='hs-varid'>decodeTensorData</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>TensorData</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>V</span><span class='hs-varop'>.</span><span class='hs-conid'>Vector</span> <span class='hs-varid'>a</span>
+<a name="line-107"></a>    <span class='hs-comment'>-- | Encode a Vector into a TensorData.</span>
+<a name="line-108"></a>    <span class='hs-comment'>--</span>
+<a name="line-109"></a>    <span class='hs-comment'>-- The values should be in row major order, e.g.,</span>
+<a name="line-110"></a>    <span class='hs-comment'>--</span>
+<a name="line-111"></a>    <span class='hs-comment'>--   element 0:   index (0, ..., 0)</span>
+<a name="line-112"></a>    <span class='hs-comment'>--   element 1:   index (0, ..., 1)</span>
+<a name="line-113"></a>    <span class='hs-comment'>--   ...</span>
+<a name="line-114"></a>    <span class='hs-varid'>encodeTensorData</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Shape</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>V</span><span class='hs-varop'>.</span><span class='hs-conid'>Vector</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>TensorData</span> <span class='hs-varid'>a</span>
+<a name="line-115"></a>
+<a name="line-116"></a><span class='hs-comment'>-- All types, besides ByteString, are encoded as simple arrays and we can use</span>
+<a name="line-117"></a><span class='hs-comment'>-- Vector.Storable to encode/decode by type casting pointers.</span>
+<a name="line-118"></a>
+<a name="line-119"></a><a name="simpleDecode"></a><span class='hs-comment'>-- TODO(fmayle): Assert that the data type matches the return type.</span>
+<a name="line-120"></a><span class='hs-definition'>simpleDecode</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Storable</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>TensorData</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>V</span><span class='hs-varop'>.</span><span class='hs-conid'>Vector</span> <span class='hs-varid'>a</span>
+<a name="line-121"></a><span class='hs-definition'>simpleDecode</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>S</span><span class='hs-varop'>.</span><span class='hs-varid'>convert</span> <span class='hs-varop'>.</span> <span class='hs-conid'>S</span><span class='hs-varop'>.</span><span class='hs-varid'>unsafeCast</span> <span class='hs-varop'>.</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-varid'>tensorDataBytes</span> <span class='hs-varop'>.</span> <span class='hs-varid'>unTensorData</span>
+<a name="line-122"></a>
+<a name="line-123"></a><a name="simpleEncode"></a><span class='hs-definition'>simpleEncode</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span><span class='hs-layout'>,</span> <span class='hs-conid'>Storable</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-124"></a>             <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Shape</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>V</span><span class='hs-varop'>.</span><span class='hs-conid'>Vector</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>TensorData</span> <span class='hs-varid'>a</span>
+<a name="line-125"></a><span class='hs-definition'>simpleEncode</span> <span class='hs-layout'>(</span><span class='hs-conid'>Shape</span> <span class='hs-varid'>xs</span><span class='hs-layout'>)</span>
+<a name="line-126"></a>    <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TensorData</span> <span class='hs-varop'>.</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorData</span> <span class='hs-varid'>xs</span> <span class='hs-varid'>dt</span> <span class='hs-varop'>.</span> <span class='hs-conid'>S</span><span class='hs-varop'>.</span><span class='hs-varid'>unsafeCast</span> <span class='hs-varop'>.</span> <span class='hs-conid'>S</span><span class='hs-varop'>.</span><span class='hs-varid'>convert</span>
+<a name="line-127"></a>  <span class='hs-keyword'>where</span>
+<a name="line-128"></a>    <span class='hs-varid'>dt</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-129"></a>
+<a name="line-130"></a><a name="instance%20TensorType%20Float"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>TensorType</span> <span class='hs-conid'>Float</span> <span class='hs-keyword'>where</span>
+<a name="line-131"></a>    <span class='hs-varid'>tensorType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_FLOAT</span>
+<a name="line-132"></a>    <span class='hs-varid'>tensorRefType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_FLOAT_REF</span>
+<a name="line-133"></a>    <span class='hs-varid'>tensorVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>floatVal</span>
+<a name="line-134"></a>    <span class='hs-varid'>decodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleDecode</span>
+<a name="line-135"></a>    <span class='hs-varid'>encodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleEncode</span>
+<a name="line-136"></a>
+<a name="line-137"></a><a name="instance%20TensorType%20Double"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>TensorType</span> <span class='hs-conid'>Double</span> <span class='hs-keyword'>where</span>
+<a name="line-138"></a>    <span class='hs-varid'>tensorType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_DOUBLE</span>
+<a name="line-139"></a>    <span class='hs-varid'>tensorRefType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_DOUBLE_REF</span>
+<a name="line-140"></a>    <span class='hs-varid'>tensorVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>doubleVal</span>
+<a name="line-141"></a>    <span class='hs-varid'>decodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleDecode</span>
+<a name="line-142"></a>    <span class='hs-varid'>encodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleEncode</span>
+<a name="line-143"></a>
+<a name="line-144"></a><a name="instance%20TensorType%20Int32"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>TensorType</span> <span class='hs-conid'>Int32</span> <span class='hs-keyword'>where</span>
+<a name="line-145"></a>    <span class='hs-varid'>tensorType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT32</span>
+<a name="line-146"></a>    <span class='hs-varid'>tensorRefType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT32_REF</span>
+<a name="line-147"></a>    <span class='hs-varid'>tensorVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>intVal</span>
+<a name="line-148"></a>    <span class='hs-varid'>decodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleDecode</span>
+<a name="line-149"></a>    <span class='hs-varid'>encodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleEncode</span>
+<a name="line-150"></a>
+<a name="line-151"></a><a name="instance%20TensorType%20Int64"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>TensorType</span> <span class='hs-conid'>Int64</span> <span class='hs-keyword'>where</span>
+<a name="line-152"></a>    <span class='hs-varid'>tensorType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT64</span>
+<a name="line-153"></a>    <span class='hs-varid'>tensorRefType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT64_REF</span>
+<a name="line-154"></a>    <span class='hs-varid'>tensorVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>int64Val</span>
+<a name="line-155"></a>    <span class='hs-varid'>decodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleDecode</span>
+<a name="line-156"></a>    <span class='hs-varid'>encodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleEncode</span>
+<a name="line-157"></a>
+<a name="line-158"></a><a name="integral"></a><span class='hs-definition'>integral</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Integral</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Lens'</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Int32</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span>
+<a name="line-159"></a><span class='hs-definition'>integral</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>iso</span> <span class='hs-layout'>(</span><span class='hs-varid'>fmap</span> <span class='hs-varid'>fromIntegral</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>fmap</span> <span class='hs-varid'>fromIntegral</span><span class='hs-layout'>)</span>
+<a name="line-160"></a>
+<a name="line-161"></a><a name="instance%20TensorType%20Word8"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>TensorType</span> <span class='hs-conid'>Word8</span> <span class='hs-keyword'>where</span>
+<a name="line-162"></a>    <span class='hs-varid'>tensorType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_UINT8</span>
+<a name="line-163"></a>    <span class='hs-varid'>tensorRefType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_UINT8_REF</span>
+<a name="line-164"></a>    <span class='hs-varid'>tensorVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>intVal</span> <span class='hs-varop'>.</span> <span class='hs-varid'>integral</span>
+<a name="line-165"></a>    <span class='hs-varid'>decodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleDecode</span>
+<a name="line-166"></a>    <span class='hs-varid'>encodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleEncode</span>
+<a name="line-167"></a>
+<a name="line-168"></a><a name="instance%20TensorType%20Word16"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>TensorType</span> <span class='hs-conid'>Word16</span> <span class='hs-keyword'>where</span>
+<a name="line-169"></a>    <span class='hs-varid'>tensorType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_UINT16</span>
+<a name="line-170"></a>    <span class='hs-varid'>tensorRefType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_UINT16_REF</span>
+<a name="line-171"></a>    <span class='hs-varid'>tensorVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>intVal</span> <span class='hs-varop'>.</span> <span class='hs-varid'>integral</span>
+<a name="line-172"></a>    <span class='hs-varid'>decodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleDecode</span>
+<a name="line-173"></a>    <span class='hs-varid'>encodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleEncode</span>
+<a name="line-174"></a>
+<a name="line-175"></a><a name="instance%20TensorType%20Int16"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>TensorType</span> <span class='hs-conid'>Int16</span> <span class='hs-keyword'>where</span>
+<a name="line-176"></a>    <span class='hs-varid'>tensorType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT16</span>
+<a name="line-177"></a>    <span class='hs-varid'>tensorRefType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT16_REF</span>
+<a name="line-178"></a>    <span class='hs-varid'>tensorVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>intVal</span> <span class='hs-varop'>.</span> <span class='hs-varid'>integral</span>
+<a name="line-179"></a>    <span class='hs-varid'>decodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleDecode</span>
+<a name="line-180"></a>    <span class='hs-varid'>encodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleEncode</span>
+<a name="line-181"></a>
+<a name="line-182"></a><a name="instance%20TensorType%20Int8"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>TensorType</span> <span class='hs-conid'>Int8</span> <span class='hs-keyword'>where</span>
+<a name="line-183"></a>    <span class='hs-varid'>tensorType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT8</span>
+<a name="line-184"></a>    <span class='hs-varid'>tensorRefType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT8_REF</span>
+<a name="line-185"></a>    <span class='hs-varid'>tensorVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>intVal</span> <span class='hs-varop'>.</span> <span class='hs-varid'>integral</span>
+<a name="line-186"></a>    <span class='hs-varid'>decodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleDecode</span>
+<a name="line-187"></a>    <span class='hs-varid'>encodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleEncode</span>
+<a name="line-188"></a>
+<a name="line-189"></a><a name="instance%20TensorType%20ByteString"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>TensorType</span> <span class='hs-conid'>ByteString</span> <span class='hs-keyword'>where</span>
+<a name="line-190"></a>    <span class='hs-varid'>tensorType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_STRING</span>
+<a name="line-191"></a>    <span class='hs-varid'>tensorRefType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_STRING_REF</span>
+<a name="line-192"></a>    <span class='hs-varid'>tensorVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>stringVal</span>
+<a name="line-193"></a>    <span class='hs-comment'>-- Encoded data layout (described in third_party/tensorflow/c/c_api.h):</span>
+<a name="line-194"></a>    <span class='hs-comment'>--   table offsets for each element :: [Word64]</span>
+<a name="line-195"></a>    <span class='hs-comment'>--   at each element offset:</span>
+<a name="line-196"></a>    <span class='hs-comment'>--     string length :: VarInt64</span>
+<a name="line-197"></a>    <span class='hs-comment'>--     string data   :: [Word8]</span>
+<a name="line-198"></a>    <span class='hs-comment'>-- TODO(fmayle): Benchmark these functions.</span>
+<a name="line-199"></a>    <span class='hs-varid'>decodeTensorData</span> <span class='hs-varid'>tensorData</span> <span class='hs-keyglyph'>=</span>
+<a name="line-200"></a>        <span class='hs-varid'>either</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>err</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>error</span> <span class='hs-varop'>$</span> <span class='hs-str'>"Malformed TF_STRING tensor; "</span> <span class='hs-varop'>++</span> <span class='hs-varid'>err</span><span class='hs-layout'>)</span> <span class='hs-varid'>id</span> <span class='hs-varop'>$</span>
+<a name="line-201"></a>            <span class='hs-keyword'>if</span> <span class='hs-varid'>expected</span> <span class='hs-varop'>/=</span> <span class='hs-varid'>count</span>
+<a name="line-202"></a>                <span class='hs-keyword'>then</span> <span class='hs-conid'>Left</span> <span class='hs-varop'>$</span> <span class='hs-str'>"decodeTensorData for ByteString count mismatch "</span> <span class='hs-varop'>++</span>
+<a name="line-203"></a>                            <span class='hs-varid'>show</span> <span class='hs-layout'>(</span><span class='hs-varid'>expected</span><span class='hs-layout'>,</span> <span class='hs-varid'>count</span><span class='hs-layout'>)</span>
+<a name="line-204"></a>                <span class='hs-keyword'>else</span> <span class='hs-conid'>V</span><span class='hs-varop'>.</span><span class='hs-varid'>mapM</span> <span class='hs-varid'>decodeString</span> <span class='hs-layout'>(</span><span class='hs-conid'>S</span><span class='hs-varop'>.</span><span class='hs-varid'>convert</span> <span class='hs-varid'>offsets</span><span class='hs-layout'>)</span>
+<a name="line-205"></a>      <span class='hs-keyword'>where</span>
+<a name="line-206"></a>        <span class='hs-varid'>expected</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>S</span><span class='hs-varop'>.</span><span class='hs-varid'>length</span> <span class='hs-varid'>offsets</span>
+<a name="line-207"></a>        <span class='hs-varid'>count</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-varop'>$</span> <span class='hs-varid'>product</span> <span class='hs-varop'>$</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-varid'>tensorDataDimensions</span>
+<a name="line-208"></a>                    <span class='hs-varop'>$</span> <span class='hs-varid'>unTensorData</span> <span class='hs-varid'>tensorData</span>
+<a name="line-209"></a>        <span class='hs-varid'>bytes</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-varid'>tensorDataBytes</span> <span class='hs-varop'>$</span> <span class='hs-varid'>unTensorData</span> <span class='hs-varid'>tensorData</span>
+<a name="line-210"></a>        <span class='hs-varid'>offsets</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>S</span><span class='hs-varop'>.</span><span class='hs-varid'>take</span> <span class='hs-varid'>count</span> <span class='hs-varop'>$</span> <span class='hs-conid'>S</span><span class='hs-varop'>.</span><span class='hs-varid'>unsafeCast</span> <span class='hs-varid'>bytes</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>S</span><span class='hs-varop'>.</span><span class='hs-conid'>Vector</span> <span class='hs-conid'>Word64</span>
+<a name="line-211"></a>        <span class='hs-varid'>dataBytes</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-varid'>pack</span> <span class='hs-varop'>$</span> <span class='hs-conid'>S</span><span class='hs-varop'>.</span><span class='hs-varid'>toList</span> <span class='hs-varop'>$</span> <span class='hs-conid'>S</span><span class='hs-varop'>.</span><span class='hs-varid'>drop</span> <span class='hs-layout'>(</span><span class='hs-varid'>count</span> <span class='hs-varop'>*</span> <span class='hs-num'>8</span><span class='hs-layout'>)</span> <span class='hs-varid'>bytes</span>
+<a name="line-212"></a>        <span class='hs-varid'>decodeString</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Word64</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Either</span> <span class='hs-conid'>String</span> <span class='hs-conid'>ByteString</span>
+<a name="line-213"></a>        <span class='hs-varid'>decodeString</span> <span class='hs-varid'>offset</span> <span class='hs-keyglyph'>=</span>
+<a name="line-214"></a>            <span class='hs-keyword'>let</span> <span class='hs-varid'>stringDataStart</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-varid'>drop</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-varid'>offset</span><span class='hs-layout'>)</span> <span class='hs-varid'>dataBytes</span>
+<a name="line-215"></a>            <span class='hs-keyword'>in</span> <span class='hs-conid'>Atto</span><span class='hs-varop'>.</span><span class='hs-varid'>eitherResult</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Atto</span><span class='hs-varop'>.</span><span class='hs-varid'>parse</span> <span class='hs-varid'>stringParser</span> <span class='hs-varid'>stringDataStart</span>
+<a name="line-216"></a>        <span class='hs-varid'>stringParser</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Atto</span><span class='hs-varop'>.</span><span class='hs-conid'>Parser</span> <span class='hs-conid'>ByteString</span>
+<a name="line-217"></a>        <span class='hs-varid'>stringParser</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>getVarInt</span> <span class='hs-varop'>&gt;&gt;=</span> <span class='hs-conid'>Atto</span><span class='hs-varop'>.</span><span class='hs-varid'>take</span> <span class='hs-varop'>.</span> <span class='hs-varid'>fromIntegral</span>
+<a name="line-218"></a>    <span class='hs-varid'>encodeTensorData</span> <span class='hs-layout'>(</span><span class='hs-conid'>Shape</span> <span class='hs-varid'>xs</span><span class='hs-layout'>)</span> <span class='hs-varid'>vec</span> <span class='hs-keyglyph'>=</span>
+<a name="line-219"></a>        <span class='hs-conid'>TensorData</span> <span class='hs-varop'>$</span> <span class='hs-conid'>FFI</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorData</span> <span class='hs-varid'>xs</span> <span class='hs-varid'>dt</span> <span class='hs-varid'>byteVector</span>
+<a name="line-220"></a>      <span class='hs-keyword'>where</span>
+<a name="line-221"></a>        <span class='hs-varid'>dt</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>
+<a name="line-222"></a>        <span class='hs-comment'>-- Add a string to an offset table and data blob.</span>
+<a name="line-223"></a>        <span class='hs-varid'>addString</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>Builder</span><span class='hs-layout'>,</span> <span class='hs-conid'>Builder</span><span class='hs-layout'>,</span> <span class='hs-conid'>Word64</span><span class='hs-layout'>)</span>
+<a name="line-224"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ByteString</span>
+<a name="line-225"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Builder</span><span class='hs-layout'>,</span> <span class='hs-conid'>Builder</span><span class='hs-layout'>,</span> <span class='hs-conid'>Word64</span><span class='hs-layout'>)</span>
+<a name="line-226"></a>        <span class='hs-varid'>addString</span> <span class='hs-layout'>(</span><span class='hs-varid'>table</span><span class='hs-layout'>,</span> <span class='hs-varid'>strings</span><span class='hs-layout'>,</span> <span class='hs-varid'>offset</span><span class='hs-layout'>)</span> <span class='hs-varid'>str</span> <span class='hs-keyglyph'>=</span>
+<a name="line-227"></a>            <span class='hs-layout'>(</span> <span class='hs-varid'>table</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-conid'>Builder</span><span class='hs-varop'>.</span><span class='hs-varid'>word64LE</span> <span class='hs-varid'>offset</span>
+<a name="line-228"></a>            <span class='hs-layout'>,</span> <span class='hs-varid'>strings</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-varid'>lengthBytes</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-conid'>Builder</span><span class='hs-varop'>.</span><span class='hs-varid'>byteString</span> <span class='hs-varid'>str</span>
+<a name="line-229"></a>            <span class='hs-layout'>,</span> <span class='hs-varid'>offset</span> <span class='hs-varop'>+</span> <span class='hs-varid'>lengthBytesLen</span> <span class='hs-varop'>+</span> <span class='hs-varid'>strLen</span>
+<a name="line-230"></a>            <span class='hs-layout'>)</span>
+<a name="line-231"></a>          <span class='hs-keyword'>where</span>
+<a name="line-232"></a>            <span class='hs-varid'>strLen</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-varop'>$</span> <span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-varid'>length</span> <span class='hs-varid'>str</span>
+<a name="line-233"></a>            <span class='hs-varid'>lengthBytes</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>putVarInt</span> <span class='hs-varop'>$</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-varop'>$</span> <span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-varid'>length</span> <span class='hs-varid'>str</span>
+<a name="line-234"></a>            <span class='hs-varid'>lengthBytesLen</span> <span class='hs-keyglyph'>=</span>
+<a name="line-235"></a>                <span class='hs-varid'>fromIntegral</span> <span class='hs-varop'>$</span> <span class='hs-conid'>L</span><span class='hs-varop'>.</span><span class='hs-varid'>length</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Builder</span><span class='hs-varop'>.</span><span class='hs-varid'>toLazyByteString</span> <span class='hs-varid'>lengthBytes</span>
+<a name="line-236"></a>        <span class='hs-comment'>-- Encode all strings.</span>
+<a name="line-237"></a>        <span class='hs-layout'>(</span><span class='hs-varid'>table'</span><span class='hs-layout'>,</span> <span class='hs-varid'>strings'</span><span class='hs-layout'>,</span> <span class='hs-keyword'>_</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>V</span><span class='hs-varop'>.</span><span class='hs-varid'>foldl'</span> <span class='hs-varid'>addString</span> <span class='hs-layout'>(</span><span class='hs-varid'>mempty</span><span class='hs-layout'>,</span> <span class='hs-varid'>mempty</span><span class='hs-layout'>,</span> <span class='hs-num'>0</span><span class='hs-layout'>)</span> <span class='hs-varid'>vec</span>
+<a name="line-238"></a>        <span class='hs-comment'>-- Concat offset table with data.</span>
+<a name="line-239"></a>        <span class='hs-varid'>bytes</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>table'</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-varid'>strings'</span>
+<a name="line-240"></a>        <span class='hs-comment'>-- Convert to Vector Word8.</span>
+<a name="line-241"></a>        <span class='hs-varid'>byteVector</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>S</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span> <span class='hs-varop'>$</span> <span class='hs-conid'>L</span><span class='hs-varop'>.</span><span class='hs-varid'>unpack</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Builder</span><span class='hs-varop'>.</span><span class='hs-varid'>toLazyByteString</span> <span class='hs-varid'>bytes</span>
+<a name="line-242"></a>
+<a name="line-243"></a>
+<a name="line-244"></a><a name="instance%20TensorType%20Bool"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>TensorType</span> <span class='hs-conid'>Bool</span> <span class='hs-keyword'>where</span>
+<a name="line-245"></a>    <span class='hs-varid'>tensorType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_BOOL</span>
+<a name="line-246"></a>    <span class='hs-varid'>tensorRefType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_BOOL_REF</span>
+<a name="line-247"></a>    <span class='hs-varid'>tensorVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>boolVal</span>
+<a name="line-248"></a>    <span class='hs-varid'>decodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleDecode</span>
+<a name="line-249"></a>    <span class='hs-varid'>encodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>simpleEncode</span>
+<a name="line-250"></a>
+<a name="line-251"></a><a name="instance%20TensorType%20(Complex%20Float)"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>TensorType</span> <span class='hs-layout'>(</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-252"></a>    <span class='hs-varid'>tensorType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_COMPLEX64</span>
+<a name="line-253"></a>    <span class='hs-varid'>tensorRefType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_COMPLEX64</span>
+<a name="line-254"></a>    <span class='hs-varid'>tensorVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-str'>"TODO (Complex Float)"</span>
+<a name="line-255"></a>    <span class='hs-varid'>decodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-str'>"TODO (Complex Float)"</span>
+<a name="line-256"></a>    <span class='hs-varid'>encodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-str'>"TODO (Complex Float)"</span>
+<a name="line-257"></a>
+<a name="line-258"></a><a name="instance%20TensorType%20(Complex%20Double)"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>TensorType</span> <span class='hs-layout'>(</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-259"></a>    <span class='hs-varid'>tensorType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_COMPLEX128</span>
+<a name="line-260"></a>    <span class='hs-varid'>tensorRefType</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_COMPLEX128</span>
+<a name="line-261"></a>    <span class='hs-varid'>tensorVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-str'>"TODO (Complex Double)"</span>
+<a name="line-262"></a>    <span class='hs-varid'>decodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-str'>"TODO (Complex Double)"</span>
+<a name="line-263"></a>    <span class='hs-varid'>encodeTensorData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-str'>"TODO (Complex Double)"</span>
+<a name="line-264"></a>
+<a name="line-265"></a><a name="Shape"></a><span class='hs-comment'>-- | Shape (dimensions) of a tensor.</span>
+<a name="line-266"></a><a name="Shape"></a><span class='hs-keyword'>newtype</span> <span class='hs-conid'>Shape</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Shape</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-keyword'>deriving</span> <span class='hs-conid'>Show</span>
+<a name="line-267"></a>
+<a name="line-268"></a><a name="instance%20IsList%20Shape"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>IsList</span> <span class='hs-conid'>Shape</span> <span class='hs-keyword'>where</span>
+<a name="line-269"></a>    <span class='hs-keyword'>type</span> <span class='hs-conid'>Item</span> <span class='hs-conid'>Shape</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Int64</span>
+<a name="line-270"></a>    <span class='hs-varid'>fromList</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Shape</span> <span class='hs-varop'>.</span> <span class='hs-varid'>fromList</span>
+<a name="line-271"></a>    <span class='hs-varid'>toList</span> <span class='hs-layout'>(</span><span class='hs-conid'>Shape</span> <span class='hs-varid'>ss</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>toList</span> <span class='hs-varid'>ss</span>
+<a name="line-272"></a>
+<a name="line-273"></a><a name="protoShape"></a><span class='hs-definition'>protoShape</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>TensorShapeProto</span> <span class='hs-conid'>Shape</span>
+<a name="line-274"></a><span class='hs-definition'>protoShape</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>iso</span> <span class='hs-varid'>protoToShape</span> <span class='hs-varid'>shapeToProto</span>
+<a name="line-275"></a>  <span class='hs-keyword'>where</span>
+<a name="line-276"></a>    <span class='hs-varid'>protoToShape</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Shape</span> <span class='hs-varop'>.</span> <span class='hs-varid'>fmap</span> <span class='hs-layout'>(</span><span class='hs-varid'>view</span> <span class='hs-varid'>size</span><span class='hs-layout'>)</span> <span class='hs-varop'>.</span> <span class='hs-varid'>view</span> <span class='hs-varid'>dim</span>
+<a name="line-277"></a>    <span class='hs-varid'>shapeToProto</span> <span class='hs-layout'>(</span><span class='hs-conid'>Shape</span> <span class='hs-varid'>ds</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>def</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>dim</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>fmap</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>d</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>def</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>size</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>d</span><span class='hs-layout'>)</span> <span class='hs-varid'>ds</span>
+<a name="line-278"></a>
+<a name="line-279"></a>
+<a name="line-280"></a><a name="Attribute"></a><span class='hs-keyword'>class</span> <span class='hs-conid'>Attribute</span> <span class='hs-varid'>a</span> <span class='hs-keyword'>where</span>
+<a name="line-281"></a>    <span class='hs-varid'>attrLens</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>AttrValue</span> <span class='hs-varid'>a</span>
+<a name="line-282"></a>
+<a name="line-283"></a><a name="instance%20Attribute%20Float"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Attribute</span> <span class='hs-conid'>Float</span> <span class='hs-keyword'>where</span>
+<a name="line-284"></a>    <span class='hs-varid'>attrLens</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>f</span>
+<a name="line-285"></a>
+<a name="line-286"></a><a name="instance%20Attribute%20ByteString"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Attribute</span> <span class='hs-conid'>ByteString</span> <span class='hs-keyword'>where</span>
+<a name="line-287"></a>    <span class='hs-varid'>attrLens</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>s</span>
+<a name="line-288"></a>
+<a name="line-289"></a><a name="instance%20Attribute%20Int64"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Attribute</span> <span class='hs-conid'>Int64</span> <span class='hs-keyword'>where</span>
+<a name="line-290"></a>    <span class='hs-varid'>attrLens</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>i</span>
+<a name="line-291"></a>
+<a name="line-292"></a><a name="instance%20Attribute%20DataType"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Attribute</span> <span class='hs-conid'>DataType</span> <span class='hs-keyword'>where</span>
+<a name="line-293"></a>    <span class='hs-varid'>attrLens</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>type'</span>
+<a name="line-294"></a>
+<a name="line-295"></a><a name="instance%20Attribute%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Attribute</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyword'>where</span>
+<a name="line-296"></a>    <span class='hs-varid'>attrLens</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tensor</span>
+<a name="line-297"></a>
+<a name="line-298"></a><a name="instance%20Attribute%20Bool"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Attribute</span> <span class='hs-conid'>Bool</span> <span class='hs-keyword'>where</span>
+<a name="line-299"></a>    <span class='hs-varid'>attrLens</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>b</span>
+<a name="line-300"></a>
+<a name="line-301"></a><a name="instance%20Attribute%20Shape"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Attribute</span> <span class='hs-conid'>Shape</span> <span class='hs-keyword'>where</span>
+<a name="line-302"></a>    <span class='hs-varid'>attrLens</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>shape</span> <span class='hs-varop'>.</span> <span class='hs-varid'>protoShape</span>
+<a name="line-303"></a>
+<a name="line-304"></a><a name="instance%20Attribute%20AttrValue'ListValue"></a><span class='hs-comment'>-- TODO(gnezdo): support generating list(Foo) from [Foo].</span>
+<a name="line-305"></a><a name="instance%20Attribute%20AttrValue'ListValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Attribute</span> <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyword'>where</span>
+<a name="line-306"></a>    <span class='hs-varid'>attrLens</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>list</span>
+<a name="line-307"></a>
+<a name="line-308"></a><a name="instance%20Attribute%20%5bDataType%5d"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Attribute</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>DataType</span><span class='hs-keyglyph'>]</span> <span class='hs-keyword'>where</span>
+<a name="line-309"></a>    <span class='hs-varid'>attrLens</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>list</span> <span class='hs-varop'>.</span> <span class='hs-varid'>type'</span>
+<a name="line-310"></a>
+<a name="line-311"></a><a name="instance%20Attribute%20%5bInt64%5d"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Attribute</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-keyword'>where</span>
+<a name="line-312"></a>    <span class='hs-varid'>attrLens</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>list</span> <span class='hs-varop'>.</span> <span class='hs-varid'>i</span>
+<a name="line-313"></a>
+<a name="line-314"></a><a name="OneOf"></a><span class='hs-comment'>-- | A 'Constraint' specifying the possible choices of a 'TensorType'.</span>
+<a name="line-315"></a><a name="OneOf"></a><span class='hs-comment'>--</span>
+<a name="line-316"></a><a name="OneOf"></a><span class='hs-comment'>-- We implement a 'Constraint' like @OneOf '[Double, Float] a@ by turning the</span>
+<a name="line-317"></a><a name="OneOf"></a><span class='hs-comment'>-- natural representation as a conjunction, i.e.,</span>
+<a name="line-318"></a><a name="OneOf"></a><span class='hs-comment'>--</span>
+<a name="line-319"></a><a name="OneOf"></a><span class='hs-comment'>-- @</span>
+<a name="line-320"></a><a name="OneOf"></a><span class='hs-comment'>--    a == Double || a == Float</span>
+<a name="line-321"></a><a name="OneOf"></a><span class='hs-comment'>-- @</span>
+<a name="line-322"></a><a name="OneOf"></a><span class='hs-comment'>--</span>
+<a name="line-323"></a><a name="OneOf"></a><span class='hs-comment'>-- into a disjunction like</span>
+<a name="line-324"></a><a name="OneOf"></a><span class='hs-comment'>--</span>
+<a name="line-325"></a><a name="OneOf"></a><span class='hs-comment'>-- @</span>
+<a name="line-326"></a><a name="OneOf"></a><span class='hs-comment'>--     a \/= Int32 &amp;&amp; a \/= Int64 &amp;&amp; a \/= ByteString &amp;&amp; ...</span>
+<a name="line-327"></a><a name="OneOf"></a><span class='hs-comment'>-- @</span>
+<a name="line-328"></a><a name="OneOf"></a><span class='hs-comment'>--</span>
+<a name="line-329"></a><a name="OneOf"></a><span class='hs-comment'>-- using an enumeration of all the possible 'TensorType's.</span>
+<a name="line-330"></a><a name="OneOf"></a><span class='hs-keyword'>type</span> <span class='hs-conid'>OneOf</span> <span class='hs-varid'>ts</span> <span class='hs-varid'>a</span>
+<a name="line-331"></a>    <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorTypes</span> <span class='hs-varid'>ts</span><span class='hs-layout'>,</span> <span class='hs-conid'>NoneOf</span> <span class='hs-layout'>(</span><span class='hs-conid'>AllTensorTypes</span> <span class='hs-varop'>\\</span> <span class='hs-varid'>ts</span><span class='hs-layout'>)</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-332"></a>
+<a name="line-333"></a><a name="TensorTypes"></a><span class='hs-comment'>-- | A 'Constraint' checking that the input is a list of 'TensorType's.</span>
+<a name="line-334"></a><a name="TensorTypes"></a><span class='hs-comment'>-- Helps improve error messages when using 'OneOf'.</span>
+<a name="line-335"></a><a name="TensorTypes"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>family</span> <span class='hs-conid'>TensorTypes</span> <span class='hs-varid'>ts</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Constraint</span> <span class='hs-keyword'>where</span>
+<a name="line-336"></a>    <span class='hs-conid'>TensorTypes</span> <span class='hs-chr'>'</span><span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>()</span>
+<a name="line-337"></a>    <span class='hs-conid'>TensorTypes</span> <span class='hs-layout'>(</span><span class='hs-varid'>t</span> <span class='hs-chr'>'</span><span class='hs-conop'>:</span> <span class='hs-varid'>ts</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorTypes</span> <span class='hs-varid'>ts</span><span class='hs-layout'>)</span>
+<a name="line-338"></a>
+<a name="line-339"></a><span class='hs-comment'>-- | A constraint checking that two types are different.</span>
+<a name="line-340"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>family</span> <span class='hs-varid'>a</span> <span class='hs-varop'>/=</span> <span class='hs-varid'>b</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Constraint</span> <span class='hs-keyword'>where</span>
+<a name="line-341"></a>    <span class='hs-varid'>a</span> <span class='hs-varop'>/=</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TypeError</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>~</span> <span class='hs-conid'>ExcludedCase</span>
+<a name="line-342"></a>    <span class='hs-varid'>a</span> <span class='hs-varop'>/=</span> <span class='hs-varid'>b</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>()</span>
+<a name="line-343"></a>
+<a name="line-344"></a><a name="TypeError"></a><span class='hs-comment'>-- | Helper types to produce a reasonable type error message when the Constraint</span>
+<a name="line-345"></a><a name="TypeError"></a><span class='hs-comment'>-- "a /= a" fails.</span>
+<a name="line-346"></a><a name="TypeError"></a><span class='hs-comment'>-- TODO(judahjacobson): Use ghc-8's CustomTypeErrors for this.</span>
+<a name="line-347"></a><a name="TypeError"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>TypeError</span> <span class='hs-varid'>a</span>
+<a name="line-348"></a><a name="ExcludedCase"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>ExcludedCase</span>
+<a name="line-349"></a>
+<a name="line-350"></a><a name="AllTensorTypes"></a><span class='hs-comment'>-- | An enumeration of all valid 'TensorType's.</span>
+<a name="line-351"></a><a name="AllTensorTypes"></a><span class='hs-keyword'>type</span> <span class='hs-conid'>AllTensorTypes</span> <span class='hs-keyglyph'>=</span>
+<a name="line-352"></a>    <span class='hs-comment'>-- NOTE: This list should be kept in sync with</span>
+<a name="line-353"></a>    <span class='hs-comment'>-- TensorFlow.OpGen.dtTypeToHaskell.</span>
+<a name="line-354"></a>    <span class='hs-comment'>-- TODO: Add support for Complex Float/Double.</span>
+<a name="line-355"></a>    <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span> <span class='hs-conid'>Float</span>
+<a name="line-356"></a>     <span class='hs-layout'>,</span> <span class='hs-conid'>Double</span>
+<a name="line-357"></a>     <span class='hs-layout'>,</span> <span class='hs-conid'>Int8</span>
+<a name="line-358"></a>     <span class='hs-layout'>,</span> <span class='hs-conid'>Int16</span>
+<a name="line-359"></a>     <span class='hs-layout'>,</span> <span class='hs-conid'>Int32</span>
+<a name="line-360"></a>     <span class='hs-layout'>,</span> <span class='hs-conid'>Int64</span>
+<a name="line-361"></a>     <span class='hs-layout'>,</span> <span class='hs-conid'>Word8</span>
+<a name="line-362"></a>     <span class='hs-layout'>,</span> <span class='hs-conid'>Word16</span>
+<a name="line-363"></a>     <span class='hs-layout'>,</span> <span class='hs-conid'>ByteString</span>
+<a name="line-364"></a>     <span class='hs-layout'>,</span> <span class='hs-conid'>Bool</span>
+<a name="line-365"></a>     <span class='hs-keyglyph'>]</span>
+<a name="line-366"></a>
+<a name="line-367"></a><a name="Delete"></a><span class='hs-comment'>-- | Removes a type from the given list of types.</span>
+<a name="line-368"></a><a name="Delete"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>family</span> <span class='hs-conid'>Delete</span> <span class='hs-varid'>a</span> <span class='hs-keyword'>as</span> <span class='hs-keyword'>where</span>
+<a name="line-369"></a>    <span class='hs-conid'>Delete</span> <span class='hs-varid'>a</span> <span class='hs-chr'>'</span><span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span> <span class='hs-chr'>'</span><span class='hs-conid'>[]</span>
+<a name="line-370"></a>    <span class='hs-conid'>Delete</span> <span class='hs-varid'>a</span> <span class='hs-layout'>(</span><span class='hs-varid'>a</span> <span class='hs-chr'>'</span><span class='hs-conop'>:</span> <span class='hs-keyword'>as</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Delete</span> <span class='hs-varid'>a</span> <span class='hs-keyword'>as</span>
+<a name="line-371"></a>    <span class='hs-conid'>Delete</span> <span class='hs-varid'>a</span> <span class='hs-layout'>(</span><span class='hs-varid'>b</span> <span class='hs-chr'>'</span><span class='hs-conop'>:</span> <span class='hs-keyword'>as</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>b</span> <span class='hs-chr'>'</span><span class='hs-conop'>:</span> <span class='hs-conid'>Delete</span> <span class='hs-varid'>a</span> <span class='hs-keyword'>as</span>
+<a name="line-372"></a>
+<a name="line-373"></a><span class='hs-comment'>-- | Takes the difference of two lists of types.</span>
+<a name="line-374"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>family</span> <span class='hs-keyword'>as</span> <span class='hs-varop'>\\</span> <span class='hs-varid'>bs</span> <span class='hs-keyword'>where</span>
+<a name="line-375"></a>    <span class='hs-keyword'>as</span> <span class='hs-varop'>\\</span> <span class='hs-chr'>'</span><span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>as</span>
+<a name="line-376"></a>    <span class='hs-keyword'>as</span> <span class='hs-varop'>\\</span> <span class='hs-varid'>b</span> <span class='hs-chr'>'</span><span class='hs-conop'>:</span> <span class='hs-varid'>bs</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Delete</span> <span class='hs-varid'>b</span> <span class='hs-keyword'>as</span> <span class='hs-varop'>\\</span> <span class='hs-varid'>bs</span>
+<a name="line-377"></a>
+<a name="line-378"></a><a name="NoneOf"></a><span class='hs-comment'>-- | A constraint that the type @a@ doesn't appear in the type list @ts@.</span>
+<a name="line-379"></a><a name="NoneOf"></a><span class='hs-comment'>-- Assumes that @a@ and each of the elements of @ts@ are 'TensorType's.</span>
+<a name="line-380"></a><a name="NoneOf"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>family</span> <span class='hs-conid'>NoneOf</span> <span class='hs-varid'>ts</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Constraint</span> <span class='hs-keyword'>where</span>
+<a name="line-381"></a>    <span class='hs-conid'>NoneOf</span> <span class='hs-chr'>'</span><span class='hs-conid'>[]</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>()</span>
+<a name="line-382"></a>    <span class='hs-conid'>NoneOf</span> <span class='hs-layout'>(</span><span class='hs-varid'>t</span> <span class='hs-chr'>'</span><span class='hs-conop'>:</span> <span class='hs-varid'>ts</span><span class='hs-layout'>)</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-varid'>a</span> <span class='hs-varop'>/=</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>NoneOf</span> <span class='hs-varid'>ts</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-0.1.0.0/src/hscolour.css b/docs/haddock/tensorflow-0.1.0.0/src/hscolour.css
new file mode 100644
index 0000000..c15919e
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/src/hscolour.css
@@ -0,0 +1,5 @@
+.hs-keyglyph, .hs-layout {color: red;}
+.hs-keyword {color: blue;}
+.hs-comment, .hs-comment a {color: green;}
+.hs-str, .hs-chr {color: teal;}
+.hs-keyword, .hs-conid, .hs-varid, .hs-conop, .hs-varop, .hs-num, .hs-cpp, .hs-sel, .hs-definition {}
diff --git a/docs/haddock/tensorflow-0.1.0.0/synopsis.png b/docs/haddock/tensorflow-0.1.0.0/synopsis.png
new file mode 100644
index 0000000000000000000000000000000000000000..85fb86ec84907bcc86531dc82871948ff4d471fa
GIT binary patch
literal 11327
zcmV-FEWp!=P)<h;3K|Lk000e1NJLTq004jh003wR1^@s6IVt!g000+IX+uL$Nkc;*
zP;zf(X>4Tx0C)k_S!GyNTeqHT_l8Y(cXyX`gGi?cY`Qxn1VID|MJXwjPC)?)F$h6K
zMMOd+6hs7sqbPzXbr*U(-*=zy-hcPcUC*=TdiNM(jyd-lv&OpsU|J&v2m2!^0SE{T
z54F(O;<U4O;KWV=44?vZKnREv-NJ%3&CHI1|Ml9M1kj;!OzhCt|7!L>E2!K(!rTCW
z%wV;vdzf1QjBf#e&~gh74F>?Z4a=WLg$KhJ^$5nap>PLbJadS>e&h8+?D`9%QNL`g
zEVKbYGXj7k5Q(8)0Fd#*a?VIMFW3*64geVHKzE-&0BG!BtmfuTbO(T`0Jaeg2nagF
z{V*1E{Wm{e|AvV~*MEExiC+KU-~R=!2{)|c6Bg`GjQ;iG|FQ`1kAUCTuZtQk34#8{
z4r4<P-^=bq9b@>(3g7#|{=Z@d+d#}7f!3C=>=26vx*jwA8>@MS>RG@Tt_zt3hie^T
z_?<zX!Qp0Bf6L=zytR+Pm;lRjJ;HSV_$l=z8Xd-r9G16)Mp#(Gm>0%9VUd=)Fos7I
z^ghPh%Jy%YZ|)vCf6EaFPai$Q-!=$ppK!y&wrJs)bNdAuANB!m3n34Tfj{s75g-&U
z1A!Pg3bcXF-=!Gv1VmU93G2duANT;{0JugFTqg*|oPXPC|A$2HS3NJd-hcPV3EW`Y
zh=1Dr-5Mv{<{zIvz#Yba<V3iQL)~HT!km4Q{?+Cmi$j0EdL!8&xgxpyr%xz;lr%~X
zrG-*OsiBkrCyE}$48%~fC`FVeN*$#F>y&^Vcn^E_`qRfl{{bzYkp)4~$~NAx_VB;E
z{?P)PU)DbV{Qi#~0H0@T9czDj06@6MNq8OrpdAz(9qQxd9nPr<&s+~tPQySqaZyfb
zNh!%g_5YjeaLxMN*$s<dYY{|0@-La=c>v_p;d%b#U$Wpz0Geb0U>E+EOsEQ;I<Re%
zqFY30<Zpkd!v`9W019}7G68nL0|epsEDhv=3eW($zzCQED_{?t;PK-PkHJt71>!&=
zNC6q(BFF<p;2Nj^wV)BSfDUjM^n!=r33v`Bz%-Zxi(nP3gDtQFK@b)qgJ>Wohy&t-
zL?CHM5mJM6p`(xmWDmJOUQi$u0mVUQpbRJ*DuT+OI;a`C4fR4p&?xj8nuk`Puh35f
z5<x)FBG?f82nmEDLK9((utJ<bcp*X%F^FWu1w;X&9C01dj_5@^K};a#5UYqyBtViP
z>5*JWF{C0=8)=GkKzbrWk@3iMWInPS*@Wyu4kE{pbI3L14-^JPgW^Pq!Q<2bWsPz}
zg`nb5nW!REEvg;Wj~YYGqt;RTXfiY_S_G|(HbmQ@z0gtU6m&ki8r_B-Ku@3-(OVb{
zh8`n;QNS2r>@mKWSWG773g!l;2Q!LUz-(f%SSG9pRuyZCC1S&|DcC~nb!<2G1$Gg;
zjU&Zz;G}VSI0sxHE(w>9tH<5Py}&KucJP#VKD;vC6z`6Y#%JLx@m=^4{33pbgo;Ff
zM3uyf#Fr$Iq=2M}WPoIbWP_BHl$%tE)ST3Z^fYM!=}po{r1PXd2-E~&f;PdC5J9*=
zs3G(aUK2LR$jJD~G{_vt!pSa>)sa0QdqcKOPD3tEZb<G<o<Lqq-cJ64{1XM9f{#Lz
z!ige=BA=p_VuWIi5>LrbsZB|wjHfK7yiNI%a+8XNN{Y&qDu61Js-9|yYMB~K%}=dM
z?M|IcT|xbTdVvN>!$YG@<3@9arjllWW|0;{D?n>V>r0zK+erJ2cAbuzPL|Gw?j&6?
z-95TFdL%tRy&=6neHMKS{UrTQ1~vvw1`mcbh9-s=4Br`97&RC@7}FVVFitT3Wa4Df
zW%6UX#MHqw%Zy?cW;SPzV!p~ez`Vvn%c8>K#*)s`!ZO8*U=?PyV2x$1V13HE$;Qs6
z&lb#9$o7D3jh&udgWZ=sm;FBb3I`2`8ix-@E=M=VM@~9UO-_H#0?vNUbuLye1Fi_J
zGOlM_JKO@?*4#<lH@IhcNO%-^Jb7|?`gqoPIe1NY<9Hi*r}*%EihSODm-z<yzVi$4
z+w!OKcknL@FbWt7#0WG9%m|VRstbk)mJ5yxVT2Tg{DiIuJr_m@%L#i67YmPwAVlOv
zd_=B@jEbT~l|%zY%SB&`5yUjbBE=fS=Edp7kBTRWcZh$I;FWNYxFj(su`ekn=`UF+
zIW0vkWh9j()g|>+T3Fgmx>$N#hD=6JCPAiC=8LR|tcUDX*;jHjawc-Aa(!}p@(S{y
z@=fw93cLy~3MC3J6=@aC6f+ecDWR3LloFKgD*aHFR}NQhQU0tVrsAhkud;kZ;E2bO
z$|DP^+^R&?GSxXXPBj;`QnfjCE_I@Mx%xW|9u0SmYK<jLAx&S+>zbdmB(*}d+O)oF
zD{G(9?$JT&=<B5GJkzDtwbm`teWS;%=cU)6x2`XzAFJPEfHp8TxMc9skj>D|u+DJZ
zNWtioQNJ<4*wVPj_}x+AqoGH;Ob{kUCOIZE$M}u~9_ug#riP|Drn6=OW+7&G%rWL>
z=Ede8ETk;rECwxUES)XuEw`++tg@`8tp%+ktov*zY#eRsY`)v-*k;?#*-6-)vU_6B
zZ0}>=>40^xaj16KJg$2@@A#sloMVdP<OzWjCr=DHF+2G?-E*dJCOWsepj~WT>Ron;
zro?jMrmLZAiR-$Xw%cX5Rd)^dT=x|ZRgY|sB~Mk)Y|mvcRj(Yc6>oL#eD5_MZJ#2a
zFTMu8*L=VGnflfE9r)Y&-w413xCGn|qz?28><?lOIvF$?EE1d&JQt!Ik{7ZbY8+Y}
z28Epn>kOxb4~I`91S8Hy%txw47DsMJ*+jLTq&gXR@@ceibXxRMj9yGtEGpJ5wl9t=
zE-`NYl;)|jcqraAzAu3%Avt03wEpSZM3O|m#Ni~#r0k?`XKc@OC9@@;PF^^xf3_io
zJS8;cWvWW*wR5O*KIfjL$)pvg?Wen^KhBWM$j{i#bjy5vUg~_o`GX6d7oKIwXI;IB
zxfpnH@{;<c+H9)qnCyidlbrTk&fN6eZ+XsnL;14#*DjM>j<`HmaI~Pakhkz+;ck(4
z(L}LU@r@GJlC+ZVSKP0>xT<lr=^ER$jB7tjeM(=I>6f*a^OxsWU@9UjK2+LN4pu2v
z)m1ZBXH@Ui1lG*eTGaN}Db&@~v({%dAQ~bXR<1i<f8MCm*x4l3RCa^@M#fEWGvemQ
zW|!v47UPz_R>jt)TYR@l+GyI++oAU8_Vo_$j=4_z&e7XOxBI$Oy4voD->JFFb+`B)
z-My^)B=?i=A9TlbZ}tTDto3^JF7!F~O+T=EFy3$8|7^f`;L$_9hYtod2fH7sKDs-k
zJaqf9;^U4d@=w~I$~|oxmK$z+CjYE`L}8@!xzh8l(IcbxU#P$69n%?mIBq!pWa8Mw
z=%n@JtCx;1=U%zLT7K>S`pZ=0)Xwzj8T3s0Eahze8`d}FZ-w68n3JEoH?K4Q^qu9q
z=>@li)%RiVcNddCkbTHs;#jI%mR`QQqPOz=<MEG+tHG-~Ybk$G{Z;Zw@YAi&8lRu8
z+pRBt3H@^LHFJY$qvo6Jx4uoY%{N;CTRY#=x0$wUe<=PK{Av4hX(wtIyIZg)wAZz7
zygzdgaBvVz3?=?L3n1YV;pqv$*BUs7u!nPp=KzrO{LU$US-|NE1lz#a2I7HkB7Bib
zC_HKj-GeE>CgGy+9whdp4g`BLCvp!8U&;uov(!a2t+bEnRv6HXyi9t`-YglcEo`$K
zI8GTZXYLH1F5YE+b^&9-c%dfYc~N>X1MygiCdpZ8N*OKLV7W5+5rusvVP$KTgd_E;
zV`@J%*flk^Jhjj1)aX9cTQC5ItVZ(2W=FkE;*aH-)|+*kk6SET?pjmWaNEk+>D${o
z_#cmV%sNr-bj$gX%QW$m8{|&wA?SI;%go!uC))SCU%7vKz~jI-L0?1Ap^RZ7;i?hG
zB3+__P9{WW#uUa@#oavB8Q+`m==5;nXwvwZiR6j1<0+%5!{;8Q^`_s>XwIxTUvlAM
z)|rdpmprp=bM$iM@_6#8@((Vr7Q8HcP;{fXs3iGH;8nY8TBRaov}JqcixtC_ZB<vQ
z&(`?YTGlDob2eZb{<=Qi*ws{eBju)Fvqg(SE88vPt>w07?YBCLI#1vB=rX<|d6)j~
z?!9;SA9XkN<n~7QIX}?p7apJ-*nBuW*z>4rDD83J6N{$`!z{xG&lW}=KCd6md=WHe
zF)la3F!5t@`sLkMS6?Sg5vR3gcxTbGOK%>(y*_twKH{Cjg64anMViI^4{J-a%g<LH
ze(YYoyLRud-cLiHC*aZaV}tse#HQ(1*!QCCfgc+?yt~$WY5Tng2XOpFKtGg+ut(A%
z-=j*=VHi8C5{?;<#;=l0k&Y4u$@<72Q1np_Pz_U$(Jatz(CstOF!D30Fk7<(v!=0C
zv-fel=R$LH^XTz<^JVZi3XBSF3Nwow5p@<z60edNko+Kxkr9;Dk#m)gS143$Q5sTy
zqw-A^tHz`*q@kp#r)8miT*qBEKrc)`${^M-!6?x<@o1vS>0=3|@n*5+(H4=G;Z`Bm
z0XDw2UUnY#t`5ZG&W<NeoN#h<c62#GbaHieBf7hKczXJJ`FjWZMEFMgo$^l#ND0gg
zx)_`nQW$zAtSr1TqB^oJs{Uj{bVE#iY<*nqsjB$0ge#{D6LXTzpE;Wxe>ObDFO_)C
zCe0{aEki1k_dNXt+=U-mA1_W_8p^(%Q<IyQmz*DR*{Q&|P@#yYn4);6Wa-M-)$VKc
zrTJw^<pCA;mAX|@)oe8+HQ#IB*NxV9H#A-^X-sd5zTtDzzS*!vsa50_Ya2Pdf^Bpx
zbxz-Y-ZgON&fVsFb@$7<i+XZ;v-&a~r1qx_oPC%wnEEJVDC=?lldDf_hFhL>j|@Mb
z9sM+h7-yIepVWIvd=>Y)XzKR#)XeT1j<?!#-t!mU-CCGlggyu?nJkB_<bCW|eYv*z
ziR!cPy2cmFuWlPb-=a6;w@!ag*pB@X{?m8o#IEU{#=iIg)35lmfin;%)QHeRj3K>H
zI8-@&65hs?W6g0$Tn9b?K9MevmJ{6JljSOT6GbGYHWfM5G<6M41g#z&E8Qx6H$yI?
z50eHn6Z1ODBi1suSavH8F-{EUJXaTYHjh8AJ|73)7XPq7g<!OhsW6Z5mdKE3k(jTz
zvII_ILb6OMKw41-$PCI}kh79!lAl#5RdiS4Q~ID>t>OirQ5IDz)!g7S$y<#pnvPn`
zTCcP(>sag3>W=B<=vx}l7>pa{8`&AN7|$LpGx0noeC)GnyV)so9SefRgyl6WA8Q%w
zeVfO&`F8I1(hk7k+3~B6fhW|RD4pIpx4EPekGo2^q1>k2n?25Xx_BviQ+coYJoGK~
zi}SY&kPV~?{2VkK+z^r;>Jw%VE)ao<Sd8qCsyuln+BHTamLqmAZW>-y@)AN%A4?QY
z!X(X~xtpASHaNvFl_z!g+(cSqdP;^<ru=#G^Gg@HvkESTTr$e$&EC&>mD`$^mG5`i
zpn$&+Rk%>pUtCp^dd2Um*){o6wlZ|t=klqF!OHfk>gs};%-W>7nEHr@(CeX%5lwM7
zQg7xp*S7SwzHLLbOLn+*Uc0?`NAB*$d)wWCJsW)~{h|X4gV%@BpPU*_8L1qd8t0!(
zdySmVd!st{bK%K{=9Rj&=Ffv)KX1|hFxkC)82{hg(&3(fkq6-NB>?O?0kGBtAd?QJ
zm0$~|LIBLj0I*U5i1iA9XzK$|?dCuG2lOlFq=GX}9v}f{nuc(O=>uZH1yBw;!3bD_
zU{(i`gLA_m=mOLPjX+-zbO<GcGvX|w5itp8bfQQ*WD2qkUVrG}Of3%84DUniXj}9p
z^ivET&bv}EkFcb0hLww*!SUjPaCh){ydAy{ze{3HQcDVv9w+S}&=VpFugO%&D#;1t
zQRE9078DOC<tZDexT%V%>8W#QsA+O&>1m7Uxak_`<>>nu%o*kx!<ev4CCswS!z|vc
zXx2(L9kvhb=^T<AuQ(IA#JHxo(|J^QKJ!-bo#1ES9~a0Gv=U+xdMjKf5+SN1MlZG|
z-Y;=YGG59>T2DqomQ{`*59GHMHWa@qZ7S~^!Kl)z@vEz7SZjuAWovinywxMoS2FN7
zEH|1t%4A}H?2754xrD_j%Moi{n>gE7_6iP##}7_;J59Lg5Ifz(-D^B~y{dc!eQ)?H
z1`GsQ2d{)Cgfm98MOmHv9&;s5@6?xs(nO0hxa6LcxN|CLdl`M_GqP+i31t7w9nHU9
zkY40hVt!S*RG^%pl2DDR1@+)Ms)_U_Lks^c#r9*J-d)LeEAIFAEIl9{kQ}rbihXiz
zxOZfJbZ?wtQtXx5l+ld&8>=~scSi5kK8P(dtn9DO{nh=s_)Emb(M`^+uiKA)7VrA)
zEB#tO5ODlSVZM$P@WWh#2Fx+Iz|6u~m`%6|24UXdCqxG`1g0=2kOkd@#-Q&AR(P%P
zMdTpvAy(jBM;jT2tUyk{D~~EF3{{U>K(nFk;T(JdLx-`&6l3PF0@xsI7Y>87!d2q7
z@J9GD{0|aKlAELyq`{<P1SLWl8G$UCY?<7Fe1t-eqMK5MvWrTEsvFK%hiUBK+%$p?
zLsvvEML)#g&IlN<F{v}XV@_oeW0_<<%_hn=1;_jd$2w;fmm@bl_bAT=UK2hFz8Cxj
z0?vX$f}2A9!g(UzqH1CcVqe9dOEgR7NySNf$XLi~$;ru!DDWzBDsd|Fs0baAQPog0
zQ+L;h)y&gs)gIT`*5lMSHi$92ZZv<C*~A7u*Ss=kxA3yOWsS12wQaP=IJh3~J;CFY
z>in5#@A}YP&ZEYQ#XH-V)Gsvv6_^~14ao?j4lj=6k7|w9iW!UZJhhvUlPHq(FxfQ)
zq?V>>q`%8dxgeZ1aw#H*HTOZjUjc35y<*QR6jwV-iRB~}tyPXS=-S45n}+?ysv9OZ
zzqJ(K(rR1j$hs}xHG4PtzG(M&@2Lj@{VyISJQ5#z^W@U7{<DMUcV5Jft50HIPQ0$2
zPMURnYc#L4Ah{^9q_ASN>hV|l=i6Vte3RLV-yYuK+dKCw{z!laG%#N$3ABJM%p<0O
zYA^skKqQbP%m$r-WBwLFh0ujLomRwONMWQ8vL5*f<`CmhgJ?Rm2f718hVj63W7)9r
z*mpQXTq~XnpG|@xNg&xFjU_!Gq>|CVvs#J#1w}9=HDxE2J2egUAWZ`85!yYvKKcv>
zJ4PYKJ*G+KW|m8=VQlv7TJY|}%0<qN;Qq<;jdz`Io&T%Amf)T+R)kiRM@&}SK*Cw_
zq*ShSi_DnpraX&+wqk%%q4J>0wyKDli~41a=UN19Bb{{JVSQ=?d&3H&&qviwE*<+|
zre<Dfv2EpG-EAvqm*;>!9^?4cDF}{Txa*#Kx+jZQvyZXwvVVG@WYFu7)G)>HwaCho
zPBE;pGpDX4cqED@Z6)`nTsY^LE}F4-ek7|Lj+#LpTmF}Vfuf?4z^j_2v}GSEI;v7@
ztn0YySFg7=Mcq_r{?^*qM(m*I?Cd&z=li|$-7G!jeOwO;25=992S<m5AICiD8%90T
z7zuw~Iy(4bWgIcVFe&g-@|E0cxhctM!5QXRlG$%>X5MzsmCeV$vtN*Wk9q%cvGzm6
zlGZYQ`Nc~9M~79`)tR-DzwAEIeH!_EZe4SI`^$~5?i-97Prt=)N^Q<<UHq>3ePg@o
zht*Hi&(<B6UGm+oJ@<X`{k8*#!}y1J%19sX9snX}!Sv&sgM$rz0I=Br>|HuI*eO3a
z*sFk(4fq>KkN@xQ6^F(cm~$_2K14li9;XkV|9<@!M&f%8Nam8p00009a7bBm000XU
z000XU0RWnu7ytkil}SWFRCodHT?u#;Rkr@KbUNvfeG_5`YY-wNfPp{+o{ADgGcxep
z5O;8ydCWk3pWowCbe1RjK4lzy;4&jKqk}U-a1=+ud7z@;LLwlFC>S)v1jwFrI_XY2
zop;WyuIf%_F<t2n-Ic!IS6#QNZr{59Ip>~x?x|CCgE~7q5lBOq0>MKUdH^|7ARquk
zTn+*P5D<VNPJtl~KcE2u0x-lWFvQ`9_7{IpP!Ma2Ag8UrfAvN=8&%31Xc{{x^H{Co
z5Bf3s5%3VAbLh&IH$@c|uDl{TIw}*k&TyNplZ9bWQ*bafH>lHMG@8ELxbaVWHf?&T
znHpfF&E_pZ&^rD;1;7qozi0Q$(`V)7{8<+kI>wdbHk%E>!9AN2eO+^{$KB)hHtVU6
z4;0@%KYw`%{kM%aj|)L>`1``u*EM%B_Ep|f_<QCY-I>7iHT~t6&rZsneaT;XVt##n
z3*O&%0=#!k4Gq$@x_XoAC663)d$?Wm=U<vc73yj8s7C4ye?8%O?|sJH9hZdp8`Rda
zV)AbC6#gCsfWR{`**=IUEL^cLOs^a53WvcKmwX4axn1m?g`lRU8J`}Mu9cL0kEW(p
z*S~w@nA^2_^<7tUJ89F|+dbgAd&6H>XTrha?_sgD)BZa!4dhf)W5g$)o+5f!@!6p=
z7>#E6lGpa0z~7?)*juclePn!mT$U>W2F?VqT7?}(LqHHhL#3+DoNXk5_#Pb{(lwSP
zZ<=X|iSbjYeFoatR`H}3=!RdX3qeSTbc>FTPC<i4yKox}dW8l|7rAq|($3Sz1I6E)
z0PuMxfT6vuJ<7S!uFMY$3&G48V-OV;j??9JIDGgtTjt^~;o&;W${mB481A6nNfvWk
zG}X)k!0qh~kGnVT5kF&`L0DN3I5DFSlBQz~XE21}l1oOBfJjYXO~BgEYVC0M-fop2
zDE{6E#$wJW!`duuVRDQ5Vt#39HNO0^LN1EPNWEbGuwmiKId5c0K$-Du1N3`J+mCki
zh@bx#O+t{9<6v}qdxy{j&CM3X#uzYdYMN4g6Ac50s6((}=&dwR{Cx?4=mpk}4k7Lh
zja}8gbI(8i^%;C~;;e!(COSeeKO#br9p##*Gxqi|^SQmA8UnbUF8H0>&5WKoW3<xS
z+uAzF<~aqE$%3S$Xk2`8s#2WoA?J>vT<}n4p!j<e0B)WjD5-NnRG#WdwxQH=@&k}$
zOR8R0dom`Lc=?)q&RcuB7q?Ti)x-Dbg5R}n#wZVUe2wiO-Z*u-^!Q-=uplma_-CN_
z8Dw8c=~1luSn8d}KeqH68(WbvJ_VB|4Oi+hk~;kO6RAfiC;c;{;CF65PXUWd$`m2j
z>ve)Qtntp05&Y$`N~L&mauhNrjZlt#E%Rdnz*4RdA(~WsS0P~4Cker*^h9K3rID79
zAhx!)2_f*-6tD+E@|~5o_HbR*DQEm#fix64W;xPOIEsuwz3>ej`Mg}wlx+M?%^s;7
zt7<_1|D+24j|zb6{d*Duo)R*nQ%A&N`m}UK6}Gim#oV|jr-^I5{&3u6Y!z0&JjK=N
zf~iA{0UNr_&1RH*=FkdaRxmwXu@ih1pW6b!KwO1@&&hNB<m70YDveg=#B4Xn!v>f0
z=VYU~zns|bF>|Ig{pE8Oi&e4q8Sf>;d>$HnJ*g4^2E{@!BWJXj|MK2>t{)#4iCiKM
z_X3_Wd3!22SVWGECF_5t9Wx1ebd<xtk&rM9vu2GY%sRn*f5AMUxid3;-rS5H$2b;{
z-mVQ6!@O~ym}o<<xL6?O&7IIAJ-pEtPt>Ve1IRabo*K&M<b7Y)<TC!a?+aI7*L)G{
zfCa!8K?T~>e+mp(08G`jsI~A7O*rz=A?*I(Ym_y4*Z<vrc-}ABZsXaPWE_55X~M~}
zTIJaLcOUp224n-2zVikie7yn%YnEfm!5#SRe*5FmzkQB}p11+o{Fp|=>BHj<`2EIL
z@XCfeu<hfsczydWJ6bb}o?S)x0g-o5D5qWnbezFoiTRwunk%qdX1+L6i)|mA5jidY
z{0}jolhGWEzbnk=GvPS&`Ekzpd{Qp}jg2@(@Fp)^ip7aWa9kF~U5c$wF9JUgt8d{b
zE$t=?X?=dwwZ3ox^kn_>GtW8G6RGFlFM<@CjE-OtU#5a;0kB%yXw(N%<3n(~<Qsu8
zZ|+#=eP8Q&gdZK`A?MLQF(CmT9+sYjF8fKIN9rETN>sBeG(H{~)Y9EAyo%kT#Rg2j
zpdOnacnjrpoDswQL%S&=xD)LJZ^c?^7~tUKxVSW2U-+UJ`I8c2{Q|sd4FLUcTr-0M
zaqMa26wFKpz7U~s3AlNV<LRJWL`!?_{RP)vp+K58DGni_Hng{J{G%C}6UQspd^w|9
z18~65>^qhrHMbm9<`9gTLcVV_VCkY<jGCK;u}-q#%LEum+wQu00v693hqg8|E+F?e
zr#1I3!kny;lw%R`{p7X)SelGzum&R`J{*=hGlr!kVHo3VG@_P-B6~JTzO@CWjwnQm
zSR&sif;%k~Gk*Hv$KL#GDAqtA4UzOxf>cW$bp+1aV?*4j`n;5NQvl5P$NHC1)DVqF
ze?14Uta}S5dTDmrRR#Fn;tPAZ>c6M&cw`%zt17X5(`x+mXPZPMYENh$xHA{IIn#Q&
z<WuL|wgH?ccYQ_s*J*Sb=&abC^R8TNFm``&#I=ks9J{Fj!m;GPmSYjgeN40tkN)>^
zG}YF_5*3HIuofIEDMeLB1jc8M#;C+D(d52>)gx`#@~i9ZqkAV_+e~x*&R~QFvHtHw
zX=O8P?QIyJ9<S9QfZh~|rm@{l%9`SLom|1Yuu=8~?mQv6f;~cV->Ss9*B|&g;0hMp
z3Alm-uHb+xn7Ts16&!E{`__2XkJh+p1UhOAxPk+&;D9SQ;0g}7f`^<d*f9ujPKWd{
zNwA(ei(>~4p*<oBsnf?}Y_tY_z+HhWM{(ruZlLc|d{fqlC~^zWp3Kn@&XYx@s8428
z+W?HF@nkVpZ73FT0nX?mGjp-|=hGB_f%7#j<?C8Quwu<K-PpY#ZzNXSa*ba=b=wuJ
zprVklr=6G|NiR*v4kFXBPw?d4Dy67aGq*W0e|U1NJv0<}=fG*4{I=9DV5`;u*19?o
z5{B?pJoNC@$d2dNpxSZdpKs&0Z&je7$@!^)U(mv>Mp`Hum_uHM8Ep9TllPO>m-^Cs
zpVwg1bK6i`-w1z*2vDs7WXVaJJHyU=rk@Vk3#W^iKzdl}7D4^3u#E2B8*>%rGlt8u
z5=Bg)^vMF>N2OW-kTeo=C=#;#Uwg6hiz=At%UPznGuZL$9uX3jIcgXzEoL+}ne7De
zePX!NLIZ__1sfvp<d*$}QF`*n@c|qD`l$#Bn=o#aJJp@HvT{duzX6s;924PRO?YW@
z5nghs!8um&p{H&~w#Fia`1v1)e}&ROaMy!>aY5fTR<P*VpVM{qeXU#N9z3DGRUK^<
z`8ISD3xNMj_Il@&dJSMm!=vkNfHgFv_xE4u6KU4aaHMFg_cJ-U=WN?_Kka)L1Nw>(
zUH5HKQ7-^w@TCk-ATqS$+;^2Y-9Yg{p~En8>~LcE&~OCN2SO-y!qgT7qsff0kWR!$
z^D81!lBm$TfXL;}=Y9YJK+SF{!{d*<g%X#sAaChntiD3qVd2a){!IDu!avzV!TF4o
z9xGgkjB%soL!aKOw*koiw_`(h<}=y@MgPFg@A$O|ZKq#6{$@AkGvc*R?Si#UlQDr#
z3~i~w{&Resf;DUNF*}X?fEFV@D0-Vh!)meXqjD6C9^>=}ZDsk}pA}{0WdF3_)n|T5
zFNK7P(SF;zrP#jx9qieE2>F-K@p;gyHGt(@rI_!hEt)McpP}lbFn3v=a0JCAI=-Ld
z^HfmLKw}#PgVO)j-n&3BpR3@}{)WrPilHHGIK3w22T8R6=u<`rMwjnBh~jFy5zt}A
zN81hv!KkMXNNPDnh1mq7H@>uwma1@k3;2!wtQCOj+9tn%uigkWBw{AL|5)BofhX2&
zA+XZ302%fCsUzg9CimQPVv`f;C6O8|{n>ML#6sZcPqU_9<QKRdEq1%Pk+|_c^D$ZZ
zpTp?8<hlYZ8Sj`5DMd7vdH&JsA-;|ya5+f%=&CEBr`QO8Na{5Jql+JdN1k1QvXkYg
zs3w&k5rOovX-K7)MutMt)1dz*eGsClwG%b<Mnp!CZzRsYY@Y!MK;y}yczWw0oN9FR
ze!+NJMke*DMR;s+x~Aozt~D6zfBzb)s_VMoY^H}`58OE$vvRV#ocF~8^#ULQLiXc}
zUQkuE$PY|=DWHMVx2^eKu~LOA6rK$<8jY@}eezhf09Y%}*qLuY{&mxkmF%SY=ox@6
zF~uhVd#X`JdPSa>DPe!$!>g7c<pOq#ONp0~Ns93gcerW+2#u6Nz)abBxO;&=#J<Sv
z@ytFfgdg<)OXiP7*2GlLRC*q2ahR^|-aI}H3;8oRM~>oyleK6R!5=0O9Kit+4(r(6
ziv6QJ8-P(X4Sa3SakRGjFIv?a0G4_jZD3}d!^RD-cH>&cq5?d2jrKkeAp_;!Ur#;&
z9W7Y4e9epUX=T6m-g%gom8l&2YDT>Vpn#D2K2TLOYC9;D1)wkDRn>N#8T3J_^Lk0W
z2GEDo5^3Wxdgdfd9w7&WOIUcVywJ$#^9sz{H)rNATQUdN%*}+3f?}K#TL)6Cfb&`3
z%&Qjw3IaWJ_$1z;4dDsM&%YQ~=42pUgopbkSWmW!9lu+5e2Bl(Hp~!=)psw#l#5d7
z<59t4!9`Er%bRtn7l4p3WRMY9&31sf7Q0{HC$^-K>G(;07G_Pk5PmWfQbk{$>nD;C
z$aX+;iw(co_@<~Qn^p+B=a%_MiWA>XQ&sn1{z<(6(1#*dufHEF>#Fe8m!&8!F2%dw
zHlg}-8UFYJZG<8tdn)d^eHPNC3G-m$^7_440RBMV3*u1l6Q_-M<NrRRMWepLk*_Lo
z-`bbN{R`%2;?^Z|DbJ;SGxcT1^sB{@@yW=TI0_Rc#C0ow;>ckXuK!rmQ$k)#dR$sG
z@^U71!@qOSF|2)@pOpG;Qm+AE#NKTmpy<6aRJ-8I$ex7UR10>zRSMI&Dx4*+aC%oe
z$>ksZdHCl3@33X-u5M#~!F>8s>bP;(@Z1iZ5DQ57E<W`F&{Exn`H?9pD7b1W{*9JP
zOVBg7&1l$Lik;?q)RUhiU1Oo)Yy>(pe>^RmdH=2Rkv1Y;;r0f4a|kUQI?AO7tZbEf
zJ(*E203jiWBR5FKRnt*$=_L9l06hS)bRb+XpPQ(|6)W>G1u?i-W6WoCJgU<Q$tb+@
zD$Jffj@}LJ{)WH{zrK~CgV-P9o6Kh%ssy0+_)+}f)#J#erD6&w@S}9^J6Kv&%Joq)
zVCl;1aO2ciq)r?I|MikfXnOjsHU;7FXSU<j4Ywhj0zZnsHR6e<UbYvt;`zsJK|)+S
z62!-nogR|)RM~kx=eB4z9^a39Htt8sxmx;YG^zT!qj*qZKBH+t(F@!0k$*VC+$MjO
zDHQiTunA>lRkTWYJ9y;~2lKhQP~5|72z2_#^8q&npdI^OKWZnM4)jd~l<t#{4Y+Cf
zJmk?iep_h!z8~!6w6%M3%WgX$8wEf`;a??H_G2r4_wX%<Bi{h4{7*`b`2}z{1;B(K
z1YRBPTX!e&hj*a(pcMU@bm?3ynVpH1H&2p)w`(&X<ooZ-37-u`n^q8?b=dLG_XXa&
z|9vVMfYDf2c#}A8B~Ph8B=-cMr4A=W1b8$?C6oGZsl=!338XY{hCK?{ZdF8R@clrx
z<=e!HAY|AD>xFIKK%PKOm(9u+`!IG4P>PAtq9@Rh0<!HW6ucJiefk6H>JE!{0DuH!
zkK`y|6ZXDM&ju*fYcM2?dkd?0BQd?AvKl9=rI$l^%Bzo%82<d<-V^wnjLl;2>pwp_
z3!t@d`N^j}MPee&>2}gr!FRvB)4o^~UCPYDMfxiI>b@c+MsVI_ZG?n%#SdILF9)yD
z8iBv~&32h6$j=)^`5;_--)1F7aK==Pycf`JwRRcIa&EjD`NGhX@h9M+TM4YCmA;oJ
zrO3=nv3MeD1n(z%`&dZj&7(JU#eehVv~0XE^yJ%^arZ3+;^s6cinJi_LRv*8MlRsh
z{Xp^er2%-zvwii|iPQND<~cxwB;)S&_u$&{D%8_7aQMh%>8YP30yAe!z=De>;j*0J
zN>6b7(K|VAAJyy)=J$-BZpMp7n5{I{+sN@1<}jm{UYm<6az<l#QupT~-a!ux6n{`>
zC)2KLBDKeY!To$ha&qG2BZqfAotPNM^BbQ^H8u4$*;5z(vZ|_v=c1LgH4&aJ8cR)s
zhZ25=_;#ffO9d0sLd30K^&jiDoI6+3R|Htse-FYDw`bL=buUu;*yY6jR@v$9iMtOO
z{Jm)a77X@ba<Dr`c8-aSjUKyf*>%$<P8>f%7edh>l!!{woQDqvAyLn?wOiY*$B%zo
zv32X~pEWczvH$rLZ56cfy6vr`0a$epDA9d}4E`PkfT>4BU?%e$j!CrfB%e1P1~}M{
zuQ8DZRRHLI>|J6XE5CNbPoY`u^Tv~L_DESt0J@K9biv&;RPgs@1TwMtC4bqg&n_U&
z^RqpU@fmCZV8(Krcxd8Db|Y=v9v+%_sqO*ye5%7a4GH|cY5=AL^#T?U?(IAraOf}Z
znfd(s?_l?Sx}{(;kM%5!ES&ry9?r8?uz9NYQ(Ynr1^j&q08@d8z|&jaWMSaE-1`Sx
z2*lKk?$1KN8*2mJGw(g3`l+riN$dE3Q~;P7L<D9yC)d4ncJO{>Cd=wx?7hW&8J3pu
z_e%g|LIn2Oqk!C_wTCQ#s9zKa2tdEcq}@UR0njdQ`-LnZ0R1A9b_)drK)bx{7qWl=
z^ovZ|Eff#{?eex?$N~b;FEVMjP(T2*%iDe-`+v|7m{y$1dn*6{002ovPDHLkV1lnB
B5rhB$

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-0.1.0.0/tensorflow.txt b/docs/haddock/tensorflow-0.1.0.0/tensorflow.txt
new file mode 100644
index 0000000..73409d2
--- /dev/null
+++ b/docs/haddock/tensorflow-0.1.0.0/tensorflow.txt
@@ -0,0 +1,626 @@
+-- Hoogle documentation, generated by Haddock
+-- See Hoogle, http://www.haskell.org/hoogle/
+
+
+-- | TensorFlow bindings.
+--   
+--   Please see README.md
+@package tensorflow
+@version 0.1.0.0
+
+
+-- | Originally taken from internal proto-lens code.
+module TensorFlow.Internal.VarInt
+
+-- | Decode an unsigned varint.
+getVarInt :: Parser Word64
+
+-- | Encode a Word64.
+putVarInt :: Word64 -> Builder
+
+module TensorFlow.Internal.FFI
+data TensorFlowException
+TensorFlowException :: Code -> Text -> TensorFlowException
+data Session
+
+-- | Runs the given action after creating a session with options populated
+--   by the given optionSetter.
+withSession :: (SessionOptions -> IO ()) -> ((IO () -> IO ()) -> Session -> IO a) -> IO a
+extendGraph :: Session -> GraphDef -> IO ()
+run :: Session -> [(ByteString, TensorData)] -> [ByteString] -> [ByteString] -> IO [TensorData]
+
+-- | All of the data needed to represent a tensor.
+data TensorData
+TensorData :: [Int64] -> !DataType -> !(Vector Word8) -> TensorData
+[tensorDataDimensions] :: TensorData -> [Int64]
+[tensorDataType] :: TensorData -> !DataType
+[tensorDataBytes] :: TensorData -> !(Vector Word8)
+setSessionConfig :: ConfigProto -> SessionOptions -> IO ()
+setSessionTarget :: ByteString -> SessionOptions -> IO ()
+
+-- | Returns the serialized OpList of all OpDefs defined in this address
+--   space.
+getAllOpList :: IO ByteString
+
+-- | Serializes the given msg and provides it as (ptr,len) argument to the
+--   given action.
+useProtoAsVoidPtrLen :: (Message msg, Num c) => msg -> (Ptr b -> c -> IO a) -> IO a
+instance GHC.Classes.Eq TensorFlow.Internal.FFI.TensorData
+instance GHC.Show.Show TensorFlow.Internal.FFI.TensorData
+instance GHC.Classes.Eq TensorFlow.Internal.FFI.TensorFlowException
+instance GHC.Show.Show TensorFlow.Internal.FFI.TensorFlowException
+instance GHC.Exception.Exception TensorFlow.Internal.FFI.TensorFlowException
+
+module TensorFlow.Types
+
+-- | The class of scalar types supported by tensorflow.
+class TensorType a
+tensorType :: TensorType a => a -> DataType
+tensorRefType :: TensorType a => a -> DataType
+tensorVal :: TensorType a => Lens' TensorProto [a]
+
+-- | Decode the bytes of a TensorData into a Vector.
+decodeTensorData :: TensorType a => TensorData a -> Vector a
+
+-- | Encode a Vector into a TensorData.
+--   
+--   The values should be in row major order, e.g.,
+--   
+--   element 0: index (0, ..., 0) element 1: index (0, ..., 1) ...
+encodeTensorData :: TensorType a => Shape -> Vector a -> TensorData a
+
+-- | Data about a tensor that is encoded for the TensorFlow APIs.
+newtype TensorData a
+TensorData :: TensorData -> TensorData a
+[unTensorData] :: TensorData a -> TensorData
+
+-- | Shape (dimensions) of a tensor.
+newtype Shape
+Shape :: [Int64] -> Shape
+protoShape :: Lens' TensorShapeProto Shape
+class Attribute a
+attrLens :: Attribute a => Lens' AttrValue a
+
+-- | A <a>Constraint</a> specifying the possible choices of a
+--   <a>TensorType</a>.
+--   
+--   We implement a <a>Constraint</a> like <tt>OneOf '[Double, Float]
+--   a</tt> by turning the natural representation as a conjunction, i.e.,
+--   
+--   <pre>
+--   a == Double || a == Float
+--   </pre>
+--   
+--   into a disjunction like
+--   
+--   <pre>
+--   a /= Int32 &amp;&amp; a /= Int64 &amp;&amp; a /= ByteString &amp;&amp; ...
+--   </pre>
+--   
+--   using an enumeration of all the possible <a>TensorType</a>s.
+type OneOf ts a = (TensorType a, TensorTypes ts, NoneOf (AllTensorTypes \\ ts) a)
+
+-- | A constraint checking that two types are different.
+
+-- | Helper types to produce a reasonable type error message when the
+--   Constraint "a /= a" fails. TODO(judahjacobson): Use ghc-8's
+--   CustomTypeErrors for this.
+data TypeError a
+data ExcludedCase
+
+-- | A <a>Constraint</a> checking that the input is a list of
+--   <a>TensorType</a>s. Helps improve error messages when using
+--   <a>OneOf</a>.
+
+-- | A constraint that the type <tt>a</tt> doesn't appear in the type list
+--   <tt>ts</tt>. Assumes that <tt>a</tt> and each of the elements of
+--   <tt>ts</tt> are <a>TensorType</a>s.
+
+-- | Takes the difference of two lists of types.
+
+-- | Removes a type from the given list of types.
+
+-- | An enumeration of all valid <a>TensorType</a>s.
+type AllTensorTypes = '[Float, Double, Int8, Int16, Int32, Int64, Word8, Word16, ByteString, Bool]
+instance GHC.Show.Show TensorFlow.Types.Shape
+instance TensorFlow.Types.TensorType GHC.Types.Float
+instance TensorFlow.Types.TensorType GHC.Types.Double
+instance TensorFlow.Types.TensorType GHC.Int.Int32
+instance TensorFlow.Types.TensorType GHC.Int.Int64
+instance TensorFlow.Types.TensorType GHC.Word.Word8
+instance TensorFlow.Types.TensorType GHC.Word.Word16
+instance TensorFlow.Types.TensorType GHC.Int.Int16
+instance TensorFlow.Types.TensorType GHC.Int.Int8
+instance TensorFlow.Types.TensorType Data.ByteString.Internal.ByteString
+instance TensorFlow.Types.TensorType GHC.Types.Bool
+instance TensorFlow.Types.TensorType (Data.Complex.Complex GHC.Types.Float)
+instance TensorFlow.Types.TensorType (Data.Complex.Complex GHC.Types.Double)
+instance GHC.Exts.IsList TensorFlow.Types.Shape
+instance TensorFlow.Types.Attribute GHC.Types.Float
+instance TensorFlow.Types.Attribute Data.ByteString.Internal.ByteString
+instance TensorFlow.Types.Attribute GHC.Int.Int64
+instance TensorFlow.Types.Attribute Proto.Tensorflow.Core.Framework.Types.DataType
+instance TensorFlow.Types.Attribute Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance TensorFlow.Types.Attribute GHC.Types.Bool
+instance TensorFlow.Types.Attribute TensorFlow.Types.Shape
+instance TensorFlow.Types.Attribute Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue
+instance TensorFlow.Types.Attribute [Proto.Tensorflow.Core.Framework.Types.DataType]
+instance TensorFlow.Types.Attribute [GHC.Int.Int64]
+
+module TensorFlow.Output
+
+-- | A type of graph node which has no outputs. These nodes are valuable
+--   for causing side effects when they are run.
+newtype ControlNode
+ControlNode :: Op -> ControlNode
+[unControlNode] :: ControlNode -> Op
+
+-- | A device that a node can be assigned to. There's a naming convention
+--   where the device names are constructed from job and replica names.
+newtype Device
+Device :: Text -> Device
+[deviceName] :: Device -> Text
+
+-- | The name of a node in the graph. This corresponds to the proto field
+--   NodeDef.name. Includes the scope prefix (if any) and a unique
+--   identifier (if the node was implicitly named).
+newtype NodeName
+NodeName :: Text -> NodeName
+[unNodeName] :: NodeName -> Text
+
+-- | The representation of a node in a TensorFlow graph.
+data Op
+
+-- | Properties are fixed, including the device, name, and scope.
+Rendered :: !NodeDef -> Op
+
+-- | Properties are not fixed, and may change depending on which context
+--   this op is rendered in.
+Unrendered :: !OpDef -> Op
+
+-- | Traverse on the <a>Unrendered</a> of an <a>Op</a>.
+--   
+--   Same implementation as _Left.
+opUnrendered :: Traversal' Op OpDef
+
+-- | Op definition. This corresponds somewhat to the <a>NodeDef</a> proto.
+data OpDef
+OpDef :: !PendingNodeName -> !OpType -> !(Map Text AttrValue) -> [Output] -> [NodeName] -> OpDef
+[_opName] :: OpDef -> !PendingNodeName
+[_opType] :: OpDef -> !OpType
+[_opAttrs] :: OpDef -> !(Map Text AttrValue)
+[_opInputs] :: OpDef -> [Output]
+[_opControlInputs] :: OpDef -> [NodeName]
+opName :: Lens' OpDef PendingNodeName
+opType :: Lens' OpDef OpType
+opAttr :: Attribute a => Text -> Lens' OpDef a
+opInputs :: Lens' OpDef [Output]
+opControlInputs :: Lens' OpDef [NodeName]
+
+-- | The type of op of a node in the graph. This corresponds to the proto
+--   field NodeDef.op.
+newtype OpType
+OpType :: Text -> OpType
+[unOpType] :: OpType -> Text
+newtype OutputIx
+OutputIx :: Int -> OutputIx
+[unOutputIx] :: OutputIx -> Int
+
+-- | An output of a TensorFlow node.
+data Output
+Output :: !OutputIx -> !Op -> Output
+output :: OutputIx -> Op -> Output
+outputIndex :: Lens' Output OutputIx
+outputOp :: Lens' Output Op
+
+-- | The name specified for an unrendered Op. If an Op has an ImplicitName,
+--   it will be assigned based on the opType plus a unique identifier. Does
+--   not contain the "scope" prefix.
+data PendingNodeName
+ExplicitName :: !Text -> PendingNodeName
+ImplicitName :: PendingNodeName
+instance GHC.Classes.Ord TensorFlow.Output.Op
+instance GHC.Classes.Eq TensorFlow.Output.Op
+instance GHC.Show.Show TensorFlow.Output.Output
+instance GHC.Classes.Ord TensorFlow.Output.Output
+instance GHC.Classes.Eq TensorFlow.Output.Output
+instance GHC.Classes.Ord TensorFlow.Output.OpDef
+instance GHC.Classes.Eq TensorFlow.Output.OpDef
+instance GHC.Show.Show TensorFlow.Output.NodeName
+instance GHC.Classes.Ord TensorFlow.Output.NodeName
+instance GHC.Classes.Eq TensorFlow.Output.NodeName
+instance GHC.Show.Show TensorFlow.Output.PendingNodeName
+instance GHC.Classes.Ord TensorFlow.Output.PendingNodeName
+instance GHC.Classes.Eq TensorFlow.Output.PendingNodeName
+instance Data.String.IsString TensorFlow.Output.Device
+instance GHC.Classes.Ord TensorFlow.Output.Device
+instance GHC.Classes.Eq TensorFlow.Output.Device
+instance GHC.Show.Show TensorFlow.Output.OutputIx
+instance GHC.Enum.Enum TensorFlow.Output.OutputIx
+instance GHC.Num.Num TensorFlow.Output.OutputIx
+instance GHC.Classes.Ord TensorFlow.Output.OutputIx
+instance GHC.Classes.Eq TensorFlow.Output.OutputIx
+instance GHC.Show.Show TensorFlow.Output.OpType
+instance GHC.Classes.Ord TensorFlow.Output.OpType
+instance GHC.Classes.Eq TensorFlow.Output.OpType
+instance Data.String.IsString TensorFlow.Output.OpType
+instance GHC.Show.Show TensorFlow.Output.Device
+instance GHC.Show.Show TensorFlow.Output.Op
+instance Data.String.IsString TensorFlow.Output.Output
+
+module TensorFlow.Tensor
+
+-- | A named output of a TensorFlow operation.
+--   
+--   The type parameter <tt>a</tt> is the type of the elements in the
+--   <a>Tensor</a>. The parameter <tt>v</tt> is either <a>Value</a> or
+--   <a>Ref</a>, depending on whether the graph is treating this op output
+--   as an immutable <a>Value</a> or a stateful <a>Ref</a> (e.g., a
+--   variable). Note that a <tt>Tensor Ref</tt> can be casted into a
+--   <tt>Tensor Value</tt> via <a>value</a>.
+data Tensor v a
+Tensor :: (TensorKind v) -> Output -> Tensor v a
+data Value
+data Ref
+
+-- | This class provides a runtime switch on whether a <a>Tensor</a> should
+--   be treated as a <a>Value</a> or as a <a>Ref</a>.
+data TensorKind v
+ValueKind :: TensorKind Value
+RefKind :: TensorKind Ref
+tensorKind :: Lens' (Tensor v a) (TensorKind v)
+tensorOutput :: Lens' (Tensor v a) Output
+
+-- | Lens for the attributes of a tensor.
+--   
+--   Only valid if the tensor has not yet been rendered. If the tensor has
+--   been rendered, the traversal will be over nothing (nothing can be read
+--   or written).
+tensorAttr :: Attribute attr => Text -> Traversal' (Tensor v a) attr
+
+-- | Cast a 'Tensor *' into a 'Tensor Value'. Common usage is to cast a Ref
+--   into Value. This behaves like a no-op.
+value :: Tensor v a -> Tensor Value a
+
+-- | A pair of a <a>Tensor</a> and some data that should be fed into that
+--   <a>Tensor</a> when running the graph.
+data Feed
+Feed :: Output -> TensorData -> Feed
+
+-- | Create a <a>Feed</a> for feeding the given data into a <a>Tensor</a>
+--   when running the graph.
+--   
+--   Note that if a <a>Tensor</a> is rendered, its identity may change; so
+--   feeding the rendered <a>Tensor</a> may be different than feeding the
+--   original <a>Tensor</a>.
+feed :: Tensor v a -> TensorData a -> Feed
+
+-- | Create a <a>Tensor</a> for a given name. This can be used to reference
+--   nodes in a <tt>GraphDef</tt> that was loaded via <tt>addGraphDef</tt>.
+--   TODO(judahjacobson): add more safety checks here.
+tensorFromName :: TensorKind v -> Text -> Tensor v a
+
+module TensorFlow.Build
+
+-- | A type of graph node which has no outputs. These nodes are valuable
+--   for causing side effects when they are run.
+newtype ControlNode
+ControlNode :: Op -> ControlNode
+[unControlNode] :: ControlNode -> Op
+data Unique
+explicitName :: Text -> PendingNodeName
+implicitName :: PendingNodeName
+opDef :: OpType -> OpDef
+opDefWithName :: PendingNodeName -> OpType -> OpDef
+opName :: Lens' OpDef PendingNodeName
+opType :: Lens' OpDef OpType
+opAttr :: Attribute a => Text -> Lens' OpDef a
+opInputs :: Lens' OpDef [Output]
+opControlInputs :: Lens' OpDef [NodeName]
+data GraphState
+
+-- | Render a <a>Tensor</a>, fixing its name, scope, device and control
+--   inputs from the <a>Build</a> context. Also renders any dependencies of
+--   the <a>Tensor</a> that weren't already rendered.
+--   
+--   This operation is idempotent; <tt>render &gt;=&gt; render ===
+--   render</tt>. However, rendering a (previously un-rendered)
+--   <a>Tensor</a> in two different contexts may result in two different
+--   <a>Tensor</a>s.
+render :: Tensor v a -> Build (Tensor v a)
+
+-- | Render a <a>Tensor</a> and get its node's name.
+renderNodeName :: Tensor v a -> Build NodeName
+renderedNodeDefs :: Lens' GraphState (Map NodeName NodeDef)
+
+-- | An action for building nodes in a TensorFlow graph. Used to manage
+--   build state internally as part of the <tt>Session</tt> monad.
+data BuildT m a
+
+-- | An action for building nodes in a TensorFlow graph.
+type Build = BuildT Identity
+
+-- | Registers the given node to be executed before the next <a>run</a>.
+addInitializer :: ControlNode -> Build ()
+
+-- | This is Control.Monad.Morph.hoist sans the dependency.
+hoistBuildT :: (forall a. m a -> n a) -> BuildT m b -> BuildT n b
+evalBuildT :: Monad m => BuildT m a -> m a
+runBuildT :: BuildT m a -> m (a, GraphState)
+
+-- | Produce a GraphDef proto representation of the nodes that are rendered
+--   in the given <a>Build</a> action.
+asGraphDef :: Build a -> GraphDef
+addGraphDef :: GraphDef -> Build ()
+
+-- | Get all the initializers that have accumulated so far, and clear that
+--   buffer.
+flushInitializers :: Monad m => BuildT m [NodeName]
+
+-- | Get all the NodeDefs that have accumulated so far, and clear that
+--   buffer.
+flushNodeBuffer :: Monad m => BuildT m [NodeDef]
+
+-- | Render the given op if it hasn't been rendered already, and return its
+--   name.
+getOrAddOp :: Op -> Build NodeName
+
+-- | Add a new node for a given <a>OpDef</a>. This is used for making
+--   "stateful" ops which are not safe to dedup (e.g, "variable" and
+--   "assign").
+addNewOp :: OpDef -> Build NodeDef
+
+-- | Render an <a>Output</a> and return a string representation for the
+--   TensorFlow foreign APIs.
+renderOutput :: Output -> Build Text
+
+-- | Places all nodes rendered in the given <a>Build</a> action on the same
+--   device as the given Tensor (see also <a>withDevice</a>). Make sure
+--   that the action has side effects of rendering the desired tensors. A
+--   pure return would not have the desired effect.
+colocateWith :: Tensor v b -> Build a -> Build a
+
+-- | Modify some part of the state, run an action, and restore the state
+--   after that action is done.
+withStateLens :: MonadState s m => Lens' s a -> (a -> a) -> m b -> m b
+
+-- | Set a device for all nodes rendered in the given <a>Build</a> action
+--   (unless further overridden by another use of withDevice).
+withDevice :: Maybe Device -> Build a -> Build a
+
+-- | Prepend a scope to all nodes rendered in the given <a>Build</a>
+--   action.
+withNameScope :: Text -> Build a -> Build a
+
+-- | Add control inputs to all nodes rendered in the given <a>Build</a>
+--   action.
+withNodeDependencies :: Set NodeName -> Build a -> Build a
+
+-- | Records the given summary action in Build for retrieval with
+--   <a>collectAllSummaries</a>. The summary op is required to produce a
+--   Summary protocol buffer in string form. For safety, use the
+--   pre-composed functions: Logging.scalarSummary and
+--   Logging.histogramSummary.
+addSummary :: SummaryTensor -> Build ()
+
+-- | Synonym for the tensors that return serialized Summary proto.
+type SummaryTensor = Tensor Value ByteString
+
+-- | Retrieves the summary ops collected thus far. Typically this only
+--   happens once, but if <a>buildWithSummary</a> is used repeatedly, the
+--   values accumulate.
+collectAllSummaries :: Monad m => BuildT m [SummaryTensor]
+instance GHC.Base.Monad m => Control.Monad.State.Class.MonadState TensorFlow.Build.GraphState (TensorFlow.Build.BuildT m)
+instance Control.Monad.Trans.Class.MonadTrans TensorFlow.Build.BuildT
+instance Control.Monad.IO.Class.MonadIO m => Control.Monad.IO.Class.MonadIO (TensorFlow.Build.BuildT m)
+instance GHC.Base.Monad m => GHC.Base.Monad (TensorFlow.Build.BuildT m)
+instance GHC.Base.Monad m => GHC.Base.Applicative (TensorFlow.Build.BuildT m)
+instance GHC.Base.Functor m => GHC.Base.Functor (TensorFlow.Build.BuildT m)
+instance GHC.Classes.Ord TensorFlow.Build.PendingNode
+instance GHC.Classes.Eq TensorFlow.Build.PendingNode
+instance Data.String.IsString TensorFlow.Build.Scope
+instance GHC.Classes.Ord TensorFlow.Build.Scope
+instance GHC.Classes.Eq TensorFlow.Build.Scope
+instance GHC.Enum.Enum TensorFlow.Build.Unique
+instance GHC.Classes.Ord TensorFlow.Build.Unique
+instance GHC.Classes.Eq TensorFlow.Build.Unique
+instance GHC.Show.Show TensorFlow.Build.Scope
+
+module TensorFlow.BuildOp
+
+-- | Class of types that can be used as op outputs.
+class OpResult a
+
+-- | Class of types that can be used as op functions.
+class BuildOp f
+
+-- | Starts an operation that returns a structured set of tensors
+--   (singletons or tuples).
+buildOp :: BuildOp f => OpDef -> f
+
+-- | Starts an operation that returns a list of tensors.
+buildListOp :: BuildOp f => [Int64] -> OpDef -> f
+
+-- | Returns true if all the integers in each tuple are identical. Throws
+--   an error with a descriptive message if not.
+eqLengthGuard :: [(String, [(String, Int)])] -> Bool
+instance GHC.Show.Show TensorFlow.BuildOp.ResultState
+instance (TensorFlow.BuildOp.OpResult a1, TensorFlow.BuildOp.OpResult a2) => TensorFlow.BuildOp.OpResult (a1, a2)
+instance (TensorFlow.BuildOp.OpResult a1, TensorFlow.BuildOp.OpResult a2, TensorFlow.BuildOp.OpResult a3) => TensorFlow.BuildOp.OpResult (a1, a2, a3)
+instance (TensorFlow.BuildOp.OpResult a1, TensorFlow.BuildOp.OpResult a2, TensorFlow.BuildOp.OpResult a3, TensorFlow.BuildOp.OpResult a4) => TensorFlow.BuildOp.OpResult (a1, a2, a3, a4)
+instance (TensorFlow.BuildOp.OpResult a1, TensorFlow.BuildOp.OpResult a2, TensorFlow.BuildOp.OpResult a3, TensorFlow.BuildOp.OpResult a4, TensorFlow.BuildOp.OpResult a5) => TensorFlow.BuildOp.OpResult (a1, a2, a3, a4, a5)
+instance (TensorFlow.BuildOp.OpResult a1, TensorFlow.BuildOp.OpResult a2, TensorFlow.BuildOp.OpResult a3, TensorFlow.BuildOp.OpResult a4, TensorFlow.BuildOp.OpResult a5, TensorFlow.BuildOp.OpResult a6) => TensorFlow.BuildOp.OpResult (a1, a2, a3, a4, a5, a6)
+instance TensorFlow.BuildOp.OpResult (TensorFlow.Tensor.Tensor TensorFlow.Tensor.Value a)
+instance TensorFlow.BuildOp.OpResult (TensorFlow.Tensor.Tensor TensorFlow.Tensor.Ref a)
+instance TensorFlow.BuildOp.OpResult TensorFlow.Output.ControlNode
+instance TensorFlow.BuildOp.OpResult a => TensorFlow.BuildOp.OpResult [a]
+instance TensorFlow.BuildOp.BuildOp TensorFlow.Output.ControlNode
+instance TensorFlow.BuildOp.BuildOp (TensorFlow.Tensor.Tensor TensorFlow.Tensor.Value a)
+instance TensorFlow.BuildOp.BuildOp (TensorFlow.Tensor.Tensor TensorFlow.Tensor.Ref a)
+instance TensorFlow.BuildOp.BuildOp [TensorFlow.Tensor.Tensor TensorFlow.Tensor.Value a]
+instance (TensorFlow.BuildOp.OpResult t1, TensorFlow.BuildOp.OpResult t2) => TensorFlow.BuildOp.BuildOp (t1, t2)
+instance (TensorFlow.BuildOp.OpResult t1, TensorFlow.BuildOp.OpResult t2, TensorFlow.BuildOp.OpResult t3) => TensorFlow.BuildOp.BuildOp (t1, t2, t3)
+instance (TensorFlow.BuildOp.OpResult t1, TensorFlow.BuildOp.OpResult t2, TensorFlow.BuildOp.OpResult t3, TensorFlow.BuildOp.OpResult t4) => TensorFlow.BuildOp.BuildOp (t1, t2, t3, t4)
+instance (TensorFlow.BuildOp.OpResult t1, TensorFlow.BuildOp.OpResult t2, TensorFlow.BuildOp.OpResult t3, TensorFlow.BuildOp.OpResult t4, TensorFlow.BuildOp.OpResult t5) => TensorFlow.BuildOp.BuildOp (t1, t2, t3, t4, t5)
+instance (TensorFlow.BuildOp.OpResult t1, TensorFlow.BuildOp.OpResult t2, TensorFlow.BuildOp.OpResult t3, TensorFlow.BuildOp.OpResult t4, TensorFlow.BuildOp.OpResult t5, TensorFlow.BuildOp.OpResult t6) => TensorFlow.BuildOp.BuildOp (t1, t2, t3, t4, t5, t6)
+instance TensorFlow.BuildOp.OpResult a => TensorFlow.BuildOp.BuildOp (TensorFlow.Build.Build a)
+instance TensorFlow.BuildOp.BuildOp f => TensorFlow.BuildOp.BuildOp (TensorFlow.Tensor.Tensor v a -> f)
+instance TensorFlow.BuildOp.BuildOp f => TensorFlow.BuildOp.BuildOp ([TensorFlow.Tensor.Tensor v a] -> f)
+
+module TensorFlow.Nodes
+
+-- | Types that contain ops which can be run.
+class Nodes t
+getNodes :: Nodes t => t -> Build (Set NodeName)
+
+-- | Types that tensor representations (e.g. <a>Tensor</a>,
+--   <a>ControlNode</a>) can be fetched into.
+--   
+--   Includes collections of tensors (e.g. tuples).
+class Nodes t => Fetchable t a
+getFetch :: Fetchable t a => t -> Build (Fetch a)
+
+-- | Fetch action. Keeps track of what needs to be fetched and how to
+--   decode the fetched data.
+data Fetch a
+Fetch :: Set Text -> (Map Text TensorData -> a) -> Fetch a
+
+-- | Nodes to fetch
+[fetches] :: Fetch a -> Set Text
+
+-- | Function to create an <tt>a</tt> from the fetched data.
+[fetchRestore] :: Fetch a -> Map Text TensorData -> a
+nodesUnion :: (Monoid b, Traversable t, Applicative f) => t (f b) -> f b
+fetchTensorList :: TensorType a => Tensor v a -> Build (Fetch (Shape, [a]))
+fetchTensorVector :: TensorType a => Tensor v a -> Build (Fetch (Shape, Vector a))
+newtype Scalar a
+Scalar :: a -> Scalar a
+[unScalar] :: Scalar a -> a
+instance Data.String.IsString a => Data.String.IsString (TensorFlow.Nodes.Scalar a)
+instance GHC.Real.RealFrac a => GHC.Real.RealFrac (TensorFlow.Nodes.Scalar a)
+instance GHC.Float.RealFloat a => GHC.Float.RealFloat (TensorFlow.Nodes.Scalar a)
+instance GHC.Real.Real a => GHC.Real.Real (TensorFlow.Nodes.Scalar a)
+instance GHC.Float.Floating a => GHC.Float.Floating (TensorFlow.Nodes.Scalar a)
+instance GHC.Real.Fractional a => GHC.Real.Fractional (TensorFlow.Nodes.Scalar a)
+instance GHC.Num.Num a => GHC.Num.Num (TensorFlow.Nodes.Scalar a)
+instance GHC.Classes.Ord a => GHC.Classes.Ord (TensorFlow.Nodes.Scalar a)
+instance GHC.Classes.Eq a => GHC.Classes.Eq (TensorFlow.Nodes.Scalar a)
+instance GHC.Show.Show a => GHC.Show.Show (TensorFlow.Nodes.Scalar a)
+instance GHC.Base.Functor TensorFlow.Nodes.Fetch
+instance GHC.Base.Applicative TensorFlow.Nodes.Fetch
+instance (TensorFlow.Nodes.Nodes t1, TensorFlow.Nodes.Nodes t2) => TensorFlow.Nodes.Nodes (t1, t2)
+instance (TensorFlow.Nodes.Nodes t1, TensorFlow.Nodes.Nodes t2, TensorFlow.Nodes.Nodes t3) => TensorFlow.Nodes.Nodes (t1, t2, t3)
+instance (TensorFlow.Nodes.Fetchable t1 a1, TensorFlow.Nodes.Fetchable t2 a2) => TensorFlow.Nodes.Fetchable (t1, t2) (a1, a2)
+instance (TensorFlow.Nodes.Fetchable t1 a1, TensorFlow.Nodes.Fetchable t2 a2, TensorFlow.Nodes.Fetchable t3 a3) => TensorFlow.Nodes.Fetchable (t1, t2, t3) (a1, a2, a3)
+instance TensorFlow.Nodes.Nodes t => TensorFlow.Nodes.Nodes [t]
+instance TensorFlow.Nodes.Fetchable t a => TensorFlow.Nodes.Fetchable [t] [a]
+instance TensorFlow.Nodes.Nodes TensorFlow.Output.ControlNode
+instance (a ~ ()) => TensorFlow.Nodes.Fetchable TensorFlow.Output.ControlNode a
+instance TensorFlow.Nodes.Nodes (TensorFlow.Tensor.Tensor v a)
+instance (TensorFlow.Types.TensorType a, a ~ a') => TensorFlow.Nodes.Fetchable (TensorFlow.Tensor.Tensor v a) (Data.Vector.Vector a')
+instance (TensorFlow.Types.TensorType a, a ~ a') => TensorFlow.Nodes.Fetchable (TensorFlow.Tensor.Tensor v a) (TensorFlow.Nodes.Scalar a')
+
+module TensorFlow.ControlFlow
+
+-- | Modify a <a>Build</a> action, such that all new ops rendered in it
+--   will depend on the nodes in the first argument.
+withControlDependencies :: Nodes t => t -> Build a -> Build a
+
+-- | Create an op that groups multiple operations.
+--   
+--   When this op finishes, all ops in the input <tt>n</tt> have finished.
+--   This op has no output.
+group :: Nodes t => t -> Build ControlNode
+
+-- | Returns a <a>Tensor</a> with the same shape and contents as the input.
+identity :: TensorType a => Tensor v a -> Tensor v a
+
+-- | Does nothing. Only useful as a placeholder for control edges.
+noOp :: ControlNode
+
+-- | Returns a <a>Tensor</a> with a given name and the same shape and
+--   contents as the input.
+--   
+--   TODO(judahjacobson): This breaks when used with uninitialize
+--   <tt>Tensor Ref</tt>s, since <tt>RefIdentity</tt> doesn't have
+--   SetAllowsUninitializedInput(). Look into whether we can change that
+--   op.
+named :: TensorType a => Text -> Tensor v a -> Tensor v a
+
+module TensorFlow.Session
+data Session a
+
+-- | Setting of an option for the session (see
+--   <a>runSessionWithOptions</a>).
+data SessionOption
+
+-- | Uses the specified config for the created session.
+sessionConfig :: ConfigProto -> SessionOption
+
+-- | Target can be: "local", ip:port, host:port. The set of supported
+--   factories depends on the linked in libraries. REQUIRES
+--   "/<i>learning</i>brain/public:tensorflow_remote" dependency for the
+--   binary.
+sessionTarget :: ByteString -> SessionOption
+
+-- | Run <a>Session</a> actions in a new TensorFlow session.
+runSession :: Session a -> IO a
+
+-- | Run <a>Session</a> actions in a new TensorFlow session created with
+--   the given option setter actions (<a>sessionTarget</a>,
+--   <a>sessionConfig</a>).
+runSessionWithOptions :: [SessionOption] -> Session a -> IO a
+
+-- | Lift a <a>Build</a> action into a <a>Session</a>, including any
+--   explicit op renderings.
+build :: Build a -> Session a
+
+-- | Helper combinator for doing something with the result of a
+--   <a>Build</a> action. Example usage:
+--   
+--   <pre>
+--   buildAnd run :: Fetchable t a =&gt; Build t -&gt; Session a
+--   </pre>
+buildAnd :: (a -> Session b) -> Build a -> Session b
+
+-- | Lift a <a>Build</a> action into a <a>Session</a>, including any
+--   explicit op renderings. Returns the merged summary ops which can be
+--   used for logging, see <a>build</a> for a convenient wrapper.
+buildWithSummary :: Build a -> Session (a, [SummaryTensor])
+
+-- | Add all pending rendered nodes to the TensorFlow graph and runs any
+--   pending initializers.
+--   
+--   Note that run, runWithFeeds, etc. will all call this function
+--   implicitly.
+extend :: Session ()
+addGraphDef :: GraphDef -> Build ()
+
+-- | Run a subgraph <tt>t</tt>, rendering any dependent nodes that aren't
+--   already rendered, and fetch the corresponding values for <tt>a</tt>.
+run :: Fetchable t a => t -> Session a
+
+-- | Run a subgraph <tt>t</tt>, rendering any dependent nodes that aren't
+--   already rendered, feed the given input values, and fetch the
+--   corresponding result values for <tt>a</tt>.
+runWithFeeds :: Fetchable t a => [Feed] -> t -> Session a
+
+-- | Run a subgraph <tt>t</tt>, rendering and extending any dependent nodes
+--   that aren't already rendered. This behaves like <a>run</a> except that
+--   it doesn't do any fetches.
+run_ :: Nodes t => t -> Session ()
+
+-- | Run a subgraph <tt>t</tt>, rendering any dependent nodes that aren't
+--   already rendered, feed the given input values, and fetch the
+--   corresponding result values for <tt>a</tt>. This behaves like
+--   <a>runWithFeeds</a> except that it doesn't do any fetches.
+runWithFeeds_ :: Nodes t => [Feed] -> t -> Session ()
+
+-- | Starts a concurrent thread which evaluates the given Nodes forever
+--   until runSession exits or an exception occurs. Graph extension happens
+--   synchronously, but the resultant run proceeds as a separate thread.
+asyncProdNodes :: Nodes t => t -> Session ()
+instance Control.Monad.IO.Class.MonadIO TensorFlow.Session.Session
+instance GHC.Base.Monad TensorFlow.Session.Session
+instance GHC.Base.Applicative TensorFlow.Session.Session
+instance GHC.Base.Functor TensorFlow.Session.Session
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html
new file mode 100644
index 0000000..1b18c8d
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html
@@ -0,0 +1,2059 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.GenOps.Core</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-GenOps-Core.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-GenOps-Core.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.GenOps.Core</p></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><a href="#v:_HostRecv">_HostRecv</a> :: <span class="keyword">forall</span> tensor_type. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tensor_type =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tensor_type</li><li class="src short"><a href="#v:_Recv">_Recv</a> :: <span class="keyword">forall</span> tensor_type. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tensor_type =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tensor_type</li><li class="src short"><a href="#v:_Send">_Send</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:_Arg">_Arg</a> :: <span class="keyword">forall</span> t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseApplyRMSProp">sparseApplyRMSProp</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 v7 v8 v9 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v8 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v9 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:applyAdam">applyAdam</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 v7 v8 v9 v10 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v8 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v9 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v10 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseApplyMomentum">sparseApplyMomentum</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:applyMomentum">applyMomentum</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:applyFtrl">applyFtrl</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 v7 v8 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v8 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseApplyAdagradDA">sparseApplyAdagradDA</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 v7 v8 v9 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v8 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v9 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseApplyAdagrad">sparseApplyAdagrad</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:applyProximalAdagrad">applyProximalAdagrad</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:applyAdagrad">applyAdagrad</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:applyAdadelta">applyAdadelta</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 v7 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseApplyProximalGradientDescent">sparseApplyProximalGradientDescent</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:applyProximalGradientDescent">applyProximalGradientDescent</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:encodeBase64">encodeBase64</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:stringSplit">stringSplit</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</li><li class="src short"><a href="#v:stringJoin">stringJoin</a> :: [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:asString">asString</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:stringToHashBucketStrong">stringToHashBucketStrong</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></li><li class="src short"><a href="#v:scatterMul">scatterMul</a> :: <span class="keyword">forall</span> v1 v2 v3 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:reduceJoin">reduceJoin</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:scatterSub">scatterSub</a> :: <span class="keyword">forall</span> v1 v2 v3 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:scatterAdd">scatterAdd</a> :: <span class="keyword">forall</span> v1 v2 v3 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:scatterUpdate">scatterUpdate</a> :: <span class="keyword">forall</span> v1 v2 v3 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:assignSub">assignSub</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:assignAdd">assignAdd</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseSegmentMeanGrad">sparseSegmentMeanGrad</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseSoftmax">sparseSoftmax</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:matrixSolve">matrixSolve</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:selfAdjointEigV2">selfAdjointEigV2</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</li><li class="src short"><a href="#v:selfAdjointEig">selfAdjointEig</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:applyGradientDescent">applyGradientDescent</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:stackPush">stackPush</a> :: <span class="keyword">forall</span> v1 v2 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:cholesky">cholesky</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:dynamicStitch">dynamicStitch</a> :: <span class="keyword">forall</span> v1 v2 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>] -&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:readerNumWorkUnitsCompleted">readerNumWorkUnitsCompleted</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></li><li class="src short"><a href="#v:readerRead">readerRead</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>)</li><li class="src short"><a href="#v:fFT2D">fFT2D</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:fixedLengthRecordReader">fixedLengthRecordReader</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:placeholder">placeholder</a> :: <span class="keyword">forall</span> dtype. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</li><li class="src short"><a href="#v:scalarSummary">scalarSummary</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:softmax">softmax</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:shardedFilename">shardedFilename</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:_HostSend">_HostSend</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:sigmoidGrad">sigmoidGrad</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:nonMaxSuppression">nonMaxSuppression</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></li><li class="src short"><a href="#v:identityReader">identityReader</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:extractGlimpse">extractGlimpse</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></li><li class="src short"><a href="#v:conv3DBackpropInput">conv3DBackpropInput</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:matrixSolveLs">matrixSolveLs</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:rGBToHSV">rGBToHSV</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:decodeGif">decodeGif</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a></li><li class="src short"><a href="#v:adjustContrast">adjustContrast</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></li><li class="src short"><a href="#v:depthToSpace">depthToSpace</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:batchMatrixSolve">batchMatrixSolve</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:erfc">erfc</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:resizeBilinearGrad">resizeBilinearGrad</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:fact">fact</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:deleteSessionTensor">deleteSessionTensor</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:logicalOr">logicalOr</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:getSessionTensor">getSessionTensor</a> :: <span class="keyword">forall</span> v1 dtype. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</li><li class="src short"><a href="#v:batchMatrixInverse">batchMatrixInverse</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:shardedFilespec">shardedFilespec</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:decodeBase64">decodeBase64</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:getSessionHandle">getSessionHandle</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:initializeTable">initializeTable</a> :: <span class="keyword">forall</span> v1 v2 v3 tkey tval. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tkey, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tval) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tkey -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tval -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:tan">tan</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:tanh">tanh</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:applyAdagradDA">applyAdagradDA</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 v7 v8 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v8 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:stringToHashBucket">stringToHashBucket</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></li><li class="src short"><a href="#v:eluGrad">eluGrad</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:fractionalAvgPoolGrad">fractionalAvgPoolGrad</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:matrixTriangularSolve">matrixTriangularSolve</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:editDistance">editDistance</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></li><li class="src short"><a href="#v:barrierIncompleteSize">barrierIncompleteSize</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></li><li class="src short"><a href="#v:threadUnsafeUnigramCandidateSampler">threadUnsafeUnigramCandidateSampler</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:barrierReadySize">barrierReadySize</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></li><li class="src short"><a href="#v:barrierClose">barrierClose</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:textLineReader">textLineReader</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:fFT3D">fFT3D</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:refExit">refExit</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:exp">exp</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:restoreSlice">restoreSlice</a> :: <span class="keyword">forall</span> v1 v2 v3 dt. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dt =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dt</li><li class="src short"><a href="#v:conj">conj</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:resizeNearestNeighborGrad">resizeNearestNeighborGrad</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:tensorArrayClose">tensorArrayClose</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:atan">atan</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:tensorArraySize">tensorArraySize</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></li><li class="src short"><a href="#v:tensorArrayConcat">tensorArrayConcat</a> :: <span class="keyword">forall</span> v1 v2 dtype. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</li><li class="src short"><a href="#v:lRN">lRN</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:stringToHashBucketFast">stringToHashBucketFast</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></li><li class="src short"><a href="#v:tensorArrayPack">tensorArrayPack</a> :: <span class="keyword">forall</span> v1 v2 dtype. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</li><li class="src short"><a href="#v:concatOffset">concatOffset</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>] -&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>]</li><li class="src short"><a href="#v:refEnter">refEnter</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:softsign">softsign</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:tensorArrayWrite">tensorArrayWrite</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></li><li class="src short"><a href="#v:diag">diag</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:matrixDiagPart">matrixDiagPart</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:queueSize">queueSize</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></li><li class="src short"><a href="#v:decodePng">decodePng</a> :: <span class="keyword">forall</span> v1 dtype. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>]` dtype) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</li><li class="src short"><a href="#v:ceil">ceil</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:priorityQueue">priorityQueue</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:placeholderWithDefault">placeholderWithDefault</a> :: <span class="keyword">forall</span> v1 dtype. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 dtype -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</li><li class="src short"><a href="#v:cropAndResizeGradImage">cropAndResizeGradImage</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:readerReset">readerReset</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:extractImagePatches">extractImagePatches</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:batchMatrixSetDiag">batchMatrixSetDiag</a> :: <span class="keyword">forall</span> v1 v2 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:stackClose">stackClose</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:quantizeAndDequantize">quantizeAndDequantize</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:isNan">isNan</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:where-39-">where'</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></li><li class="src short"><a href="#v:listDiff">listDiff</a> :: <span class="keyword">forall</span> v1 v2 t out_idx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_idx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` out_idx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_idx)</li><li class="src short"><a href="#v:stridedSlice">stridedSlice</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 index t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> index, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` index, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 index -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 index -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 index -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:randomShuffleQueue">randomShuffleQueue</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:tileGrad">tileGrad</a> :: <span class="keyword">forall</span> v1 v2 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:stridedSliceAssign">stridedSliceAssign</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 index t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> index, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` index, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 index -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 index -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 index -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:reshape">reshape</a> :: <span class="keyword">forall</span> v1 v2 t tshape. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tshape, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tshape) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tshape -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:fIFOQueue">fIFOQueue</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:learnedUnigramCandidateSampler">learnedUnigramCandidateSampler</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:fractionalAvgPool">fractionalAvgPool</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</li><li class="src short"><a href="#v:randomCrop">randomCrop</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:_HostCast">_HostCast</a> :: <span class="keyword">forall</span> v1 dstT srcT. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dstT, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> srcT) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 srcT -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dstT</li><li class="src short"><a href="#v:queueClose">queueClose</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:slice">slice</a> :: <span class="keyword">forall</span> v1 v2 v3 index t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> index, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` index, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 index -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 index -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:stridedSliceGrad">stridedSliceGrad</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 index t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> index, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` index, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 index -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 index -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 index -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 index -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseTensorDenseAdd">sparseTensorDenseAdd</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:size">size</a> :: <span class="keyword">forall</span> v1 t out_type. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_type, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` out_type) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_type</li><li class="src short"><a href="#v:barrier">barrier</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:lgamma">lgamma</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:decodeJpeg">decodeJpeg</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a></li><li class="src short"><a href="#v:shapeN">shapeN</a> :: <span class="keyword">forall</span> v1 t out_type. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_type, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` out_type) =&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t] -&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_type]</li><li class="src short"><a href="#v:uniformCandidateSampler">uniformCandidateSampler</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:unique">unique</a> :: <span class="keyword">forall</span> v1 t out_idx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_idx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` out_idx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_idx)</li><li class="src short"><a href="#v:drawBoundingBoxes">drawBoundingBoxes</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:tensorArraySplit">tensorArraySplit</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></li><li class="src short"><a href="#v:split">split</a> :: <span class="keyword">forall</span> v1 v2 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t]</li><li class="src short"><a href="#v:segmentMax">segmentMax</a> :: <span class="keyword">forall</span> v1 v2 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:abort">abort</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:sparseReorder">sparseReorder</a> :: <span class="keyword">forall</span> v1 v2 v3 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</li><li class="src short"><a href="#v:rsqrtGrad">rsqrtGrad</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:reverseSequence">reverseSequence</a> :: <span class="keyword">forall</span> v1 v2 t tlen. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tlen, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tlen) =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tlen -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:readerNumRecordsProduced">readerNumRecordsProduced</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></li><li class="src short"><a href="#v:deserializeManySparse">deserializeManySparse</a> :: <span class="keyword">forall</span> v1 dtype. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</li><li class="src short"><a href="#v:immutableConst">immutableConst</a> :: <span class="keyword">forall</span> dtype. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</li><li class="src short"><a href="#v:minimum">minimum</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:initializeTableFromTextFile">initializeTableFromTextFile</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:diagPart">diagPart</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:log">log</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:tensorArrayScatter">tensorArrayScatter</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></li><li class="src short"><a href="#v:rank">rank</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></li><li class="src short"><a href="#v:identity">identity</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:adjustContrastv2">adjustContrastv2</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></li><li class="src short"><a href="#v:sparseApplyProximalAdagrad">sparseApplyProximalAdagrad</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 v7 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:gather">gather</a> :: <span class="keyword">forall</span> v1 v2 tindices tparams. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tparams) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tparams -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tparams</li><li class="src short"><a href="#v:isVariableInitialized">isVariableInitialized</a> :: <span class="keyword">forall</span> v1 dtype. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 dtype -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:concat">concat</a> :: <span class="keyword">forall</span> v1 v2 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:randomUniformInt">randomUniformInt</a> :: <span class="keyword">forall</span> v1 v2 v3 t tout. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tout, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tout) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tout -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tout -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tout</li><li class="src short"><a href="#v:stopGradient">stopGradient</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:avgPool">avgPool</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:wholeFileReader">wholeFileReader</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:switch">switch</a> :: <span class="keyword">forall</span> v1 v2 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</li><li class="src short"><a href="#v:randomStandardNormal">randomStandardNormal</a> :: <span class="keyword">forall</span> v1 t dtype. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` dtype) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</li><li class="src short"><a href="#v:sigmoid">sigmoid</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sampleDistortedBoundingBox">sampleDistortedBoundingBox</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:greater">greater</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:refNextIteration">refNextIteration</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:spaceToDepth">spaceToDepth</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:controlTrigger">controlTrigger</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:scatterDiv">scatterDiv</a> :: <span class="keyword">forall</span> v1 v2 v3 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:copy">copy</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:cropAndResizeGradBoxes">cropAndResizeGradBoxes</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></li><li class="src short"><a href="#v:sparseSegmentMean">sparseSegmentMean</a> :: <span class="keyword">forall</span> v1 v2 v3 t tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:assign">assign</a> :: <span class="keyword">forall</span> v1 v2 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:restore">restore</a> :: <span class="keyword">forall</span> v1 v2 dt. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dt =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dt</li><li class="src short"><a href="#v:maxPoolGradWithArgmax">maxPoolGradWithArgmax</a> :: <span class="keyword">forall</span> v1 v2 v3 t targmax. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> targmax, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` targmax) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 targmax -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:checkNumerics">checkNumerics</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:zerosLike">zerosLike</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:readFile">readFile</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:transpose">transpose</a> :: <span class="keyword">forall</span> v1 v2 t tperm. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tperm, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tperm) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tperm -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:parseTensor">parseTensor</a> :: <span class="keyword">forall</span> v1 out_type. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_type =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_type</li><li class="src short"><a href="#v:acos">acos</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:bitcast">bitcast</a> :: <span class="keyword">forall</span> v1 t type'. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> type', <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` type') =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> type'</li><li class="src short"><a href="#v:lookupTableImport">lookupTableImport</a> :: <span class="keyword">forall</span> v1 v2 v3 tin tout. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tin, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tout) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tin -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tout -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:biasAddGrad">biasAddGrad</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:batchSelfAdjointEig">batchSelfAdjointEig</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:prod">prod</a> :: <span class="keyword">forall</span> v1 v2 t tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:resizeBilinear">resizeBilinear</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></li><li class="src short"><a href="#v:tensorArrayUnpack">tensorArrayUnpack</a> :: <span class="keyword">forall</span> v1 v2 v3 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></li><li class="src short"><a href="#v:batchMatrixDeterminant">batchMatrixDeterminant</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sum">sum</a> :: <span class="keyword">forall</span> v1 v2 t tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:iFFT2D">iFFT2D</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:fill">fill</a> :: <span class="keyword">forall</span> v1 v2 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:fixedUnigramCandidateSampler">fixedUnigramCandidateSampler</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:dilation2D">dilation2D</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:polygamma">polygamma</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:refIdentity">refIdentity</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:encodePng">encodePng</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:lookupTableInsert">lookupTableInsert</a> :: <span class="keyword">forall</span> v1 v2 v3 tin tout. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tin, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tout) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tin -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tout -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:batchIFFT2D">batchIFFT2D</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:uniqueWithCounts">uniqueWithCounts</a> :: <span class="keyword">forall</span> v1 t out_idx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_idx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` out_idx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_idx, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_idx)</li><li class="src short"><a href="#v:gatherNd">gatherNd</a> :: <span class="keyword">forall</span> v1 v2 tindices tparams. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tparams) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tparams -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tparams</li><li class="src short"><a href="#v:tensorArrayRead">tensorArrayRead</a> :: <span class="keyword">forall</span> v1 v2 v3 dtype. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</li><li class="src short"><a href="#v:readerReadUpTo">readerReadUpTo</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>)</li><li class="src short"><a href="#v:betainc">betainc</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:batchMatrixBandPart">batchMatrixBandPart</a> :: <span class="keyword">forall</span> v1 v2 v3 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:depthwiseConv2dNativeBackpropInput">depthwiseConv2dNativeBackpropInput</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:refSelect">refSelect</a> :: <span class="keyword">forall</span> v1 v2 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:exit">exit</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:lookupTableFind">lookupTableFind</a> :: <span class="keyword">forall</span> v1 v2 v3 tin tout. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tin, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tout) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tin -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tout -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tout</li><li class="src short"><a href="#v:squeeze">squeeze</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:mean">mean</a> :: <span class="keyword">forall</span> v1 v2 t tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:spaceToBatchND">spaceToBatchND</a> :: <span class="keyword">forall</span> v1 v2 v3 t tblock_shape tpaddings. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tblock_shape, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tblock_shape, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tpaddings, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tpaddings) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tblock_shape -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tpaddings -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:spaceToBatch">spaceToBatch</a> :: <span class="keyword">forall</span> v1 v2 t tpaddings. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tpaddings, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tpaddings) =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tpaddings -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:cTCGreedyDecoder">cTCGreedyDecoder</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:batchToSpaceND">batchToSpaceND</a> :: <span class="keyword">forall</span> v1 v2 v3 t tblock_shape tcrops. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tblock_shape, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tblock_shape, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tcrops, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tcrops) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tblock_shape -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tcrops -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:pack">pack</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:oneHot">oneHot</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t tI. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tI, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>]` tI) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tI -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:broadcastGradientArgs">broadcastGradientArgs</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</li><li class="src short"><a href="#v:matrixSetDiag">matrixSetDiag</a> :: <span class="keyword">forall</span> v1 v2 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:applyRMSProp">applyRMSProp</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 v7 v8 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v8 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:const">const</a> :: <span class="keyword">forall</span> dtype. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</li><li class="src short"><a href="#v:enter">enter</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:debugIdentity">debugIdentity</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:debugNanCount">debugNanCount</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></li><li class="src short"><a href="#v:batchNormWithGlobalNormalization">batchNormWithGlobalNormalization</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:batchMatrixDiag">batchMatrixDiag</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:unpack">unpack</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t]</li><li class="src short"><a href="#v:sparseSplit">sparseSplit</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; ([<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>], [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t], [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>])</li><li class="src short"><a href="#v:mirrorPad">mirrorPad</a> :: <span class="keyword">forall</span> v1 v2 t tpaddings. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tpaddings, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tpaddings) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tpaddings -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:batchMatrixDiagPart">batchMatrixDiagPart</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:fractionalMaxPoolGrad">fractionalMaxPoolGrad</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:matchingFiles">matchingFiles</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:tile">tile</a> :: <span class="keyword">forall</span> v1 v2 t tmultiples. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tmultiples, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tmultiples) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tmultiples -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseSparseMinimum">sparseSparseMinimum</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</li><li class="src short"><a href="#v:allCandidateSampler">allCandidateSampler</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:refSwitch">refSwitch</a> :: <span class="keyword">forall</span> v1 v2 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</li><li class="src short"><a href="#v:mergeSummary">mergeSummary</a> :: [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:logicalNot">logicalNot</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:lRNGrad">lRNGrad</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:stringToNumber">stringToNumber</a> :: <span class="keyword">forall</span> v1 out_type. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_type, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` out_type) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_type</li><li class="src short"><a href="#v:sparseMatMul">sparseMatMul</a> :: <span class="keyword">forall</span> v1 v2 ta tb. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> ta, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` ta, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tb, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` tb) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 ta -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tb -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></li><li class="src short"><a href="#v:merge">merge</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t] -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>)</li><li class="src short"><a href="#v:choleskyGrad">choleskyGrad</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:batchCholeskyGrad">batchCholeskyGrad</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:tensorArrayGather">tensorArrayGather</a> :: <span class="keyword">forall</span> v1 v2 v3 dtype. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</li><li class="src short"><a href="#v:resizeNearestNeighbor">resizeNearestNeighbor</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:negTrain">negTrain</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:tensorArrayGrad">tensorArrayGrad</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:audioSummary">audioSummary</a> :: <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:noOp">noOp</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:nextIteration">nextIteration</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:softplusGrad">softplusGrad</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:svd">svd</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</li><li class="src short"><a href="#v:hSVToRGB">hSVToRGB</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:parameterizedTruncatedNormal">parameterizedTruncatedNormal</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 t dtype. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` dtype) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 dtype -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 dtype -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 dtype -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 dtype -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</li><li class="src short"><a href="#v:square">square</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:elu">elu</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:lookupTableExport">lookupTableExport</a> :: <span class="keyword">forall</span> v1 tkeys tvalues. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tkeys, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tvalues) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tkeys, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tvalues)</li><li class="src short"><a href="#v:lookupTableSize">lookupTableSize</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></li><li class="src short"><a href="#v:avgPoolGrad">avgPoolGrad</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:computeAccidentalHits">computeAccidentalHits</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:cTCLoss">cTCLoss</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:avgPool3D">avgPool3D</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:inv">inv</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:stackPop">stackPop</a> :: <span class="keyword">forall</span> v1 elem_type. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> elem_type =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> elem_type</li><li class="src short"><a href="#v:paddingFIFOQueue">paddingFIFOQueue</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:batchSelfAdjointEigV2">batchSelfAdjointEigV2</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</li><li class="src short"><a href="#v:batchMatrixTriangularSolve">batchMatrixTriangularSolve</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:batchMatrixSolveLs">batchMatrixSolveLs</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:batchSvd">batchSvd</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</li><li class="src short"><a href="#v:tensorSummary">tensorSummary</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:sparseSoftmaxCrossEntropyWithLogits">sparseSoftmaxCrossEntropyWithLogits</a> :: <span class="keyword">forall</span> v1 v2 t tlabels. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tlabels, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tlabels) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tlabels -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</li><li class="src short"><a href="#v:maxPoolWithArgmax">maxPoolWithArgmax</a> :: <span class="keyword">forall</span> v1 t targmax. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> targmax, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` targmax) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> targmax)</li><li class="src short"><a href="#v:fFT">fFT</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:histogramSummary">histogramSummary</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:pad">pad</a> :: <span class="keyword">forall</span> v1 v2 t tpaddings. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tpaddings, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tpaddings) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tpaddings -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:batchIFFT3D">batchIFFT3D</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:imageSummary">imageSummary</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:segmentSum">segmentSum</a> :: <span class="keyword">forall</span> v1 v2 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:encodeJpeg">encodeJpeg</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:batchNormWithGlobalNormalizationGrad">batchNormWithGlobalNormalizationGrad</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</li><li class="src short"><a href="#v:biasAddV1">biasAddV1</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:invertPermutation">invertPermutation</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:mirrorPadGrad">mirrorPadGrad</a> :: <span class="keyword">forall</span> v1 v2 t tpaddings. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tpaddings, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tpaddings) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tpaddings -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:reverse">reverse</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:conv2D">conv2D</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:conv2DBackpropInput">conv2DBackpropInput</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:readerSerializeState">readerSerializeState</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:temporaryVariable">temporaryVariable</a> :: <span class="keyword">forall</span> dtype. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</li><li class="src short"><a href="#v:cropAndResize">cropAndResize</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></li><li class="src short"><a href="#v:maxPoolGrad">maxPoolGrad</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:fusedResizeAndPadConv2D">fusedResizeAndPadConv2D</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:randomUniform">randomUniform</a> :: <span class="keyword">forall</span> v1 t dtype. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` dtype) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</li><li class="src short"><a href="#v:depthwiseConv2dNative">depthwiseConv2dNative</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseApplyAdadelta">sparseApplyAdadelta</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 v7 v8 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v8 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:depthwiseConv2dNativeBackpropFilter">depthwiseConv2dNativeBackpropFilter</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:conv3D">conv3D</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:greaterEqual">greaterEqual</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:sparseDenseCwiseAdd">sparseDenseCwiseAdd</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:conv3DBackpropFilter">conv3DBackpropFilter</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:conv3DBackpropInputV2">conv3DBackpropInputV2</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:mod">mod</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:refMerge">refMerge</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t] -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>)</li><li class="src short"><a href="#v:conv3DBackpropFilterV2">conv3DBackpropFilterV2</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:serializeManySparse">serializeManySparse</a> :: <span class="keyword">forall</span> v1 v2 v3 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:avgPool3DGrad">avgPool3DGrad</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:maxPool3DGrad">maxPool3DGrad</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseReduceSum">sparseReduceSum</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:relu">relu</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:l2Loss">l2Loss</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:readerRestoreState">readerRestoreState</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:shape">shape</a> :: <span class="keyword">forall</span> v1 t out_type. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_type, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` out_type) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_type</li><li class="src short"><a href="#v:softmaxCrossEntropyWithLogits">softmaxCrossEntropyWithLogits</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</li><li class="src short"><a href="#v:maxPool">maxPool</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:dilation2DBackpropInput">dilation2DBackpropInput</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:equal">equal</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a>, <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:dilation2DBackpropFilter">dilation2DBackpropFilter</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:reluGrad">reluGrad</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:relu6">relu6</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:resizeBicubic">resizeBicubic</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></li><li class="src short"><a href="#v:relu6Grad">relu6Grad</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseTensorDenseMatMul">sparseTensorDenseMatMul</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:softplus">softplus</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:batchMatMul">batchMatMul</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:softsignGrad">softsignGrad</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:lessEqual">lessEqual</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:logSoftmax">logSoftmax</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:inTopK">inTopK</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t) =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:matrixDiag">matrixDiag</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:maxPool3D">maxPool3D</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:topK">topK</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>)</li><li class="src short"><a href="#v:topKV2">topKV2</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>)</li><li class="src short"><a href="#v:fractionalMaxPool">fractionalMaxPool</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</li><li class="src short"><a href="#v:matrixBandPart">matrixBandPart</a> :: <span class="keyword">forall</span> v1 v2 v3 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:decodeRaw">decodeRaw</a> :: <span class="keyword">forall</span> v1 out_type. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_type, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` out_type) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_type</li><li class="src short"><a href="#v:decodeJSONExample">decodeJSONExample</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:truncatedNormal">truncatedNormal</a> :: <span class="keyword">forall</span> v1 t dtype. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` dtype) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</li><li class="src short"><a href="#v:randomShuffle">randomShuffle</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:multinomial">multinomial</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></li><li class="src short"><a href="#v:randomGamma">randomGamma</a> :: <span class="keyword">forall</span> v1 v2 s t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> s, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` s, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 s -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:addN">addN</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:max">max</a> :: <span class="keyword">forall</span> v1 v2 t tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:_Retval">_Retval</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:destroyTemporaryVariable">destroyTemporaryVariable</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:cast">cast</a> :: <span class="keyword">forall</span> v1 dstT srcT. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dstT, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> srcT) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 srcT -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dstT</li><li class="src short"><a href="#v:countUpTo">countUpTo</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t) =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:abs">abs</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:neg">neg</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseSparseMaximum">sparseSparseMaximum</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</li><li class="src short"><a href="#v:invGrad">invGrad</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sqrt">sqrt</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:matrixInverse">matrixInverse</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sqrtGrad">sqrtGrad</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:expandDims">expandDims</a> :: <span class="keyword">forall</span> v1 v2 t tdim. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tdim, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tdim) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tdim -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:all">all</a> :: <span class="keyword">forall</span> v1 v2 tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:cTCBeamSearchDecoder">cTCBeamSearchDecoder</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; ([<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>], [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>], [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>], <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:rsqrt">rsqrt</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:tanhGrad">tanhGrad</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sin">sin</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:matrixDeterminant">matrixDeterminant</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:cos">cos</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:batchToSpace">batchToSpace</a> :: <span class="keyword">forall</span> v1 v2 t tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseToDense">sparseToDense</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:asin">asin</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:argMin">argMin</a> :: <span class="keyword">forall</span> v1 v2 t tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></li><li class="src short"><a href="#v:isInf">isInf</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:sign">sign</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:add">add</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseApplyFtrl">sparseApplyFtrl</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 v7 v8 v9 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v8 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v9 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sub">sub</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:batchFFT3D">batchFFT3D</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:sparseReduceSumSparse">sparseReduceSumSparse</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</li><li class="src short"><a href="#v:biasAdd">biasAdd</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:mul">mul</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:div">div</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:loopCond">loopCond</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:squaredDifference">squaredDifference</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:maximum">maximum</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:logUniformCandidateSampler">logUniformCandidateSampler</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:less">less</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:pow">pow</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:igammac">igammac</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:igamma">igamma</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:zeta">zeta</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:imag">imag</a> :: <span class="keyword">forall</span> v1 t tout. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tout, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` tout) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tout</li><li class="src short"><a href="#v:complex">complex</a> :: <span class="keyword">forall</span> v1 v2 t tout. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tout, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` tout) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tout</li><li class="src short"><a href="#v:notEqual">notEqual</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a>, <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:complexAbs">complexAbs</a> :: <span class="keyword">forall</span> v1 t tout. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tout, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` tout) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tout</li><li class="src short"><a href="#v:logicalAnd">logicalAnd</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:batchFFT">batchFFT</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:select">select</a> :: <span class="keyword">forall</span> v1 v2 v3 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:matMul">matMul</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:digamma">digamma</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:conv2DBackpropFilter">conv2DBackpropFilter</a> :: <span class="keyword">forall</span> v1 v2 v3 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:min">min</a> :: <span class="keyword">forall</span> v1 v2 t tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:isFinite">isFinite</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:argMax">argMax</a> :: <span class="keyword">forall</span> v1 v2 t tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></li><li class="src short"><a href="#v:segmentMean">segmentMean</a> :: <span class="keyword">forall</span> v1 v2 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:cumprod">cumprod</a> :: <span class="keyword">forall</span> v1 v2 t tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:segmentMin">segmentMin</a> :: <span class="keyword">forall</span> v1 v2 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:unsortedSegmentSum">unsortedSegmentSum</a> :: <span class="keyword">forall</span> v1 v2 v3 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:tFRecordReader">tFRecordReader</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:sparseSegmentSum">sparseSegmentSum</a> :: <span class="keyword">forall</span> v1 v2 v3 t tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseSegmentSqrtN">sparseSegmentSqrtN</a> :: <span class="keyword">forall</span> v1 v2 v3 t tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:copyHost">copyHost</a> :: <span class="keyword">forall</span> v1 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:variable">variable</a> :: <span class="keyword">forall</span> dtype. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</li><li class="src short"><a href="#v:sparseSegmentSqrtNGrad">sparseSegmentSqrtNGrad</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:range">range</a> :: <span class="keyword">forall</span> v1 v2 v3 tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tidx</li><li class="src short"><a href="#v:any">any</a> :: <span class="keyword">forall</span> v1 v2 tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></li><li class="src short"><a href="#v:linSpace">linSpace</a> :: <span class="keyword">forall</span> v1 v2 v3 t tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:resizeArea">resizeArea</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></li><li class="src short"><a href="#v:real">real</a> :: <span class="keyword">forall</span> v1 t tout. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tout, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` tout) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tout</li><li class="src short"><a href="#v:iFFT">iFFT</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:iFFT3D">iFFT3D</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:cross">cross</a> :: <span class="keyword">forall</span> v1 v2 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:cumsum">cumsum</a> :: <span class="keyword">forall</span> v1 v2 t tidx. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:batchIFFT">batchIFFT</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:erf">erf</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:barrierInsertMany">barrierInsertMany</a> :: <span class="keyword">forall</span> v1 v2 v3 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:floor">floor</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:batchFFT2D">batchFFT2D</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li class="src short"><a href="#v:sparseAddGrad">sparseAddGrad</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</li><li class="src short"><a href="#v:sparseAdd">sparseAdd</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 v5 v6 v7 t treal. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> treal, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` treal) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 treal -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</li><li class="src short"><a href="#v:batchCholesky">batchCholesky</a> :: <span class="keyword">forall</span> v1 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:dynamicPartition">dynamicPartition</a> :: <span class="keyword">forall</span> v1 v2 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t]</li><li class="src short"><a href="#v:serializeSparse">serializeSparse</a> :: <span class="keyword">forall</span> v1 v2 v3 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:sparseConcat">sparseConcat</a> :: <span class="keyword">forall</span> v1 v2 v3 t. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>] -&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t] -&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>] -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</li><li class="src short"><a href="#v:segmentProd">segmentProd</a> :: <span class="keyword">forall</span> v1 v2 t tindices. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseReshape">sparseReshape</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</li><li class="src short"><a href="#v:sparseDenseCwiseMul">sparseDenseCwiseMul</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sparseDenseCwiseDiv">sparseDenseCwiseDiv</a> :: <span class="keyword">forall</span> v1 v2 v3 v4 t. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><a name="v:_HostRecv" class="def">_HostRecv</a> <a href="src/TensorFlow-GenOps-Core.html#_HostRecv" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tensor_type</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>send_device_incarnation</strong>: The current incarnation of send_device.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tensor_type</td><td class="doc"><p><strong>tensor</strong>: The tensor to receive.</p></td></tr></table></div><div class="doc"><p>Receives the named tensor from send_device on recv_device.</p><p>_HostRecv requires its input on host memory whereas _Recv requires its
+ input on device memory.</p></div></div><div class="top"><p class="src"><a name="v:_Recv" class="def">_Recv</a> <a href="src/TensorFlow-GenOps-Core.html#_Recv" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tensor_type</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>send_device_incarnation</strong>: The current incarnation of send_device.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tensor_type</td><td class="doc"><p><strong>tensor</strong>: The tensor to receive.</p></td></tr></table></div><div class="doc"><p>Receives the named tensor from send_device on recv_device.</p></div></div><div class="top"><p class="src"><a name="v:_Send" class="def">_Send</a> <a href="src/TensorFlow-GenOps-Core.html#_Send" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>send_device_incarnation</strong>: The current incarnation of send_device.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>tensor</strong>: The tensor to send.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Sends the named tensor from send_device to recv_device.</p></div></div><div class="top"><p class="src"><a name="v:_Arg" class="def">_Arg</a> <a href="src/TensorFlow-GenOps-Core.html#_Arg" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>index</strong>: This argument is the index-th argument of the function.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The argument.</p></td></tr></table></div><div class="doc"><p>A graph node which represents an argument to a function.</p></div></div><div class="top"><p class="src"><a name="v:sparseApplyRMSProp" class="def">sparseApplyRMSProp</a> <a href="src/TensorFlow-GenOps-Core.html#sparseApplyRMSProp" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>ms</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>mom</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>lr</strong>: Scaling factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>rho</strong>: Decay rate. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t</td><td class="doc"><p><strong>momentum</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t</td><td class="doc"><p><strong>epsilon</strong>: Ridge term. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v8 t</td><td class="doc"><p><strong>grad</strong>: The gradient.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v9 tindices</td><td class="doc"><p><strong>indices</strong>: A vector of indices into the first dimension of var, ms and mom.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Update '*var' according to the RMSProp algorithm.</p><p>Note that in dense implement of this algorithm, ms and mom will
+ update even if the grad is zero, but in this sparse implement, ms
+ and mom will not update in iterations the grad is zero.</p><p>mean_square = decay * mean_square + (1-decay) * gradient ** 2
+ Delta = learning_rate * gradient / sqrt(mean_square + epsilon)</p><p>ms &lt;- rho * ms_{t-1} + (1-rho) * grad * grad
+ mom &lt;- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
+ var &lt;- var - mom</p></div></div><div class="top"><p class="src"><a name="v:applyAdam" class="def">applyAdam</a> <a href="src/TensorFlow-GenOps-Core.html#applyAdam" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>m</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>v</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>beta1_power</strong>: Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>beta2_power</strong>: Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t</td><td class="doc"><p><strong>lr</strong>: Scaling factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t</td><td class="doc"><p><strong>beta1</strong>: Momentum factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v8 t</td><td class="doc"><p><strong>beta2</strong>: Momentum factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v9 t</td><td class="doc"><p><strong>epsilon</strong>: Ridge term. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v10 t</td><td class="doc"><p><strong>grad</strong>: The gradient.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Update '*var' according to the Adam algorithm.</p><p>lr_t &lt;- learning_rate * sqrt(1 - beta2^t) / (1 - beta1^t)
+ m_t &lt;- beta1 * m_{t-1} + (1 - beta1) * g_t
+ v_t &lt;- beta2 * v_{t-1} + (1 - beta2) * g_t * g_t
+ variable &lt;- variable - lr_t * m_t / (sqrt(v_t) + epsilon)</p></div></div><div class="top"><p class="src"><a name="v:sparseApplyMomentum" class="def">sparseApplyMomentum</a> <a href="src/TensorFlow-GenOps-Core.html#sparseApplyMomentum" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>accum</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>lr</strong>: Learning rate. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>grad</strong>: The gradient.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 tindices</td><td class="doc"><p><strong>indices</strong>: A vector of indices into the first dimension of var and accum.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t</td><td class="doc"><p><strong>momentum</strong>: Momentum. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Update relevant entries in '*var' and '*accum' according to the momentum scheme.</p><p>Set use_nesterov = True if you want to use Nesterov momentum.</p><p>That is for rows we have grad for, we update var and accum as follows:</p><p>accum = accum * momentum + grad
+ var -= lr * accum</p></div></div><div class="top"><p class="src"><a name="v:applyMomentum" class="def">applyMomentum</a> <a href="src/TensorFlow-GenOps-Core.html#applyMomentum" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>accum</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>lr</strong>: Scaling factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>grad</strong>: The gradient.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>momentum</strong>: Momentum. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Update '*var' according to the momentum scheme. Set use_nesterov = True if you</p><p>want to use Nesterov momentum.</p><p>accum = accum * momentum + grad
+ var -= lr * accum</p></div></div><div class="top"><p class="src"><a name="v:applyFtrl" class="def">applyFtrl</a> <a href="src/TensorFlow-GenOps-Core.html#applyFtrl" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>accum</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>linear</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>grad</strong>: The gradient.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>lr</strong>: Scaling factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t</td><td class="doc"><p><strong>l1</strong>: L1 regulariation. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t</td><td class="doc"><p><strong>l2</strong>: L2 regulariation. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v8 t</td><td class="doc"><p><strong>lr_power</strong>: Scaling factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Update '*var' according to the Ftrl-proximal scheme.</p><p>accum_new = accum + grad * grad
+ linear += grad + (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
+ quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
+ var = (sign(linear) * l1 - linear) / quadratic if |linear| &gt; l1 else 0.0
+ accum = accum_new</p></div></div><div class="top"><p class="src"><a name="v:sparseApplyAdagradDA" class="def">sparseApplyAdagradDA</a> <a href="src/TensorFlow-GenOps-Core.html#sparseApplyAdagradDA" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>gradient_accumulator</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>gradient_squared_accumulator</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>grad</strong>: The gradient.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 tindices</td><td class="doc"><p><strong>indices</strong>: A vector of indices into the first dimension of var and accum.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t</td><td class="doc"><p><strong>lr</strong>: Learning rate. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t</td><td class="doc"><p><strong>l1</strong>: L1 regularization. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v8 t</td><td class="doc"><p><strong>l2</strong>: L2 regularization. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v9 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>global_step</strong>: Training step number. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Update entries in '*var' and '*accum' according to the proximal adagrad scheme.</p></div></div><div class="top"><p class="src"><a name="v:sparseApplyAdagrad" class="def">sparseApplyAdagrad</a> <a href="src/TensorFlow-GenOps-Core.html#sparseApplyAdagrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>accum</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>lr</strong>: Learning rate. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>grad</strong>: The gradient.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 tindices</td><td class="doc"><p><strong>indices</strong>: A vector of indices into the first dimension of var and accum.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Update relevant entries in '*var' and '*accum' according to the adagrad scheme.</p><p>That is for rows we have grad for, we update var and accum as follows:
+ accum += grad * grad
+ var -= lr * grad * (1 / sqrt(accum))</p></div></div><div class="top"><p class="src"><a name="v:applyProximalAdagrad" class="def">applyProximalAdagrad</a> <a href="src/TensorFlow-GenOps-Core.html#applyProximalAdagrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>accum</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>lr</strong>: Scaling factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>l1</strong>: L1 regularization. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>l2</strong>: L2 regularization. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t</td><td class="doc"><p><strong>grad</strong>: The gradient.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Update '*var' and '*accum' according to FOBOS with Adagrad learning rate.</p><p>accum += grad * grad
+ prox_v = var - lr * grad * (1 / sqrt(accum))
+ var = sign(prox_v)/(1+lr*l2) * max{|prox_v|-lr*l1,0}</p></div></div><div class="top"><p class="src"><a name="v:applyAdagrad" class="def">applyAdagrad</a> <a href="src/TensorFlow-GenOps-Core.html#applyAdagrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>accum</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>lr</strong>: Scaling factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>grad</strong>: The gradient.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Update '*var' according to the adagrad scheme.</p><p>accum += grad * grad
+ var -= lr * grad * (1 / sqrt(accum))</p></div></div><div class="top"><p class="src"><a name="v:applyAdadelta" class="def">applyAdadelta</a> <a href="src/TensorFlow-GenOps-Core.html#applyAdadelta" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>accum</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>accum_update</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>lr</strong>: Scaling factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>rho</strong>: Decay factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t</td><td class="doc"><p><strong>epsilon</strong>: Constant factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t</td><td class="doc"><p><strong>grad</strong>: The gradient.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Update '*var' according to the adadelta scheme.</p><p>accum = rho() * accum + (1 - rho()) * grad.square();
+ update = (update_accum + epsilon).sqrt() * (accum + epsilon()).rsqrt() * grad;
+ update_accum = rho() * update_accum + (1 - rho()) * update.square();
+ var -= update;</p></div></div><div class="top"><p class="src"><a name="v:sparseApplyProximalGradientDescent" class="def">sparseApplyProximalGradientDescent</a> <a href="src/TensorFlow-GenOps-Core.html#sparseApplyProximalGradientDescent" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>alpha</strong>: Scaling factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>l1</strong>: L1 regularization. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>l2</strong>: L2 regularization. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>grad</strong>: The gradient.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 tindices</td><td class="doc"><p><strong>indices</strong>: A vector of indices into the first dimension of var and accum.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Sparse update '*var' as FOBOS algorithm with fixed learning rate.</p><p>That is for rows we have grad for, we update var as follows:
+ prox_v = var - alpha * grad
+ var = sign(prox_v)/(1+alpha*l2) * max{|prox_v|-alpha*l1,0}</p></div></div><div class="top"><p class="src"><a name="v:applyProximalGradientDescent" class="def">applyProximalGradientDescent</a> <a href="src/TensorFlow-GenOps-Core.html#applyProximalGradientDescent" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>alpha</strong>: Scaling factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>l1</strong>: L1 regularization. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>l2</strong>: L2 regularization. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>delta</strong>: The change.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Update '*var' as FOBOS algorithm with fixed learning rate.</p><p>prox_v = var - alpha * delta
+ var = sign(prox_v)/(1+alpha*l2) * max{|prox_v|-alpha*l1,0}</p></div></div><div class="top"><p class="src"><a name="v:encodeBase64" class="def">encodeBase64</a> <a href="src/TensorFlow-GenOps-Core.html#encodeBase64" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>input</strong>: Strings to be encoded.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>output</strong>: Input strings encoded in base64.</p></td></tr></table></div><div class="doc"><p>Encode strings into web-safe base64 format.</p><p>Refer to the following article for more information on base64 format:
+ en.wikipedia.org<em>wiki</em>Base64. Base64 strings may have padding with '=' at the
+ end so that the encoded has length multiple of 4. See Padding section of the
+ link above.</p><p>Web-safe means that the encoder uses - and _ instead of + and /.</p></div></div><div class="top"><p class="src"><a name="v:stringSplit" class="def">stringSplit</a> <a href="src/TensorFlow-GenOps-Core.html#stringSplit" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>input</strong>: 1-D. Strings to split.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>delimiter</strong>: 0-D. Delimiter character, or empty string.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</td><td class="doc"><p>(<strong>indices</strong>, <strong>values</strong>, <strong>shape</strong>)</p><ul><li><strong>indices</strong>: A dense matrix of int64 representing the indices of the sparse tensor.</li><li><strong>values</strong>: A vector of strings corresponding to the splited values.</li><li><strong>shape</strong>: a length-2 vector of int64 representing the shape of the sparse
+ tensor, where the first value is N and the second value is the maximum number
+ of tokens in a single input entry.</li></ul></td></tr></table></div><div class="doc"><p>Split elements of <code>input</code> based on <code>delimiter</code> into a <code>SparseTensor</code>.</p><p>Let N be the size of source (typically N will be the batch size). Split each
+ element of <code>input</code> based on <code>delimiter</code> and return a <code>SparseTensor</code>
+ containing the splitted tokens. Empty tokens are ignored.</p><p><code>delimiter</code> can be empty or a single character. If <code>delimiter</code> is an empty
+  string, each element of <code>input</code> is split into individual 1 character strings.</p><p>For example:
+   N = 2, input[0] is 'hello world' and input[1] is 'a b c', then the output
+   will be</p><p>indices = [0, 0;
+              0, 1;
+              1, 0;
+              1, 1;
+              1, 2]
+   shape = [2, 3]
+   values = [<code>hello</code>, <code>world</code>, <code>a</code>, <code>b</code>, <code>c</code>]</p></div></div><div class="top"><p class="src"><a name="v:stringJoin" class="def">stringJoin</a> <a href="src/TensorFlow-GenOps-Core.html#stringJoin" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>]</td><td class="doc"><p><strong>inputs</strong>: A list of string tensors.  The tensors must all have the same shape,
+ or be scalars.  Scalars may be mixed in; these will be broadcast to the shape
+ of non-scalar inputs.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Joins the strings in the given list of string tensors into one tensor;</p><p>with the given separator (default is an empty separator).</p></div></div><div class="top"><p class="src"><a name="v:asString" class="def">asString</a> <a href="src/TensorFlow-GenOps-Core.html#asString" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Converts each entry in the given tensor to strings.  Supports many numeric</p><p>types and boolean.</p></div></div><div class="top"><p class="src"><a name="v:stringToHashBucketStrong" class="def">stringToHashBucketStrong</a> <a href="src/TensorFlow-GenOps-Core.html#stringToHashBucketStrong" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_buckets</strong>: The number of buckets.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>input</strong>: The strings to assign a hash bucket.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>output</strong>: A Tensor of the same shape as the input <code>string_tensor</code>.</p></td></tr></table></div><div class="doc"><p>Converts each string in the input Tensor to its hash mod by a number of buckets.</p><p>The hash function is deterministic on the content of the string within the
+ process. The hash function is a keyed hash function, where attribute <code>key</code>
+ defines the key of the hash function. <code>key</code> is an array of 2 elements.</p><p>A strong hash is important when inputs may be malicious, e.g. URLs with
+ additional components. Adversaries could try to make their inputs hash to the
+ same bucket for a denial-of-service attack or to skew the results. A strong
+ hash prevents this by making it dificult, if not infeasible, to compute inputs
+ that hash to the same bucket. This comes at a cost of roughly 4x higher compute
+ time than tf.string_to_hash_bucket_fast.</p></div></div><div class="top"><p class="src"><a name="v:scatterMul" class="def">scatterMul</a> <a href="src/TensorFlow-GenOps-Core.html#scatterMul" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>ref</strong>: Should be from a <code>Variable</code> node.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices</td><td class="doc"><p><strong>indices</strong>: A tensor of indices into the first dimension of <code>ref</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>updates</strong>: A tensor of updated values to multiply to <code>ref</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output_ref</strong>: = Same as <code>ref</code>.  Returned as a convenience for operations that want
+ to use the updated values after the update is done.</p></td></tr></table></div><div class="doc"><p>Multiplies sparse updates into a variable reference.</p><p>This operation computes</p><p># Scalar indices
+     ref[indices, ...] *= updates[...]</p><p># Vector indices (for each i)
+     ref[indices[i], ...] *= updates[i, ...]</p><p># High rank indices (for each i, ..., j)
+     ref[indices[i, ..., j], ...] *= updates[i, ..., j, ...]</p><p>This operation outputs <code>ref</code> after the update is done.
+ This makes it easier to chain operations that need to use the reset value.</p><p>Duplicate entries are handled correctly: if multiple <code>indices</code> reference
+ the same location, their contributions multiply.</p><p>Requires `updates.shape = indices.shape + ref.shape[1:]`.</p></div></div><div class="top"><p class="src"><a name="v:reduceJoin" class="def">reduceJoin</a> <a href="src/TensorFlow-GenOps-Core.html#reduceJoin" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>inputs</strong>: The input to be joined.  All reduced indices must have non-zero size.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>reduction_indices</strong>: The dimensions to reduce over.  Dimensions are reduced in the
+ order specified.  Omitting <code>reduction_indices</code> is equivalent to passing
+ `[n-1, n-2, ..., 0]`.  Negative indices from `-n` to `-1` are supported.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>output</strong>: Has shape equal to that of the input with reduced dimensions removed or
+ set to `1` depending on <code>keep_dims</code>.</p></td></tr></table></div><div class="doc"><p>Joins a string Tensor across the given dimensions.</p><p>Computes the string join across dimensions in the given string Tensor of shape
+ `[d_0, d_1, ..., d_n-1]`.  Returns a new Tensor created by joining the input
+ strings with the given separator (default: empty string).  Negative indices are
+ counted backwards from the end, with `-1` being equivalent to `n - 1`.  Passing
+ an empty <code>reduction_indices</code> joins all strings in linear index order and outputs
+ a scalar string.</p><p>For example:</p><p>```
+ # tensor <code>a</code> is [[&quot;a&quot;, &quot;b&quot;], [&quot;c&quot;, &quot;d&quot;]]
+ tf.reduce_join(a, 0) ==&gt; [&quot;ac&quot;, &quot;bd&quot;]
+ tf.reduce_join(a, 1) ==&gt; [&quot;ab&quot;, &quot;cd&quot;]
+ tf.reduce_join(a, -2) = tf.reduce_join(a, 0) ==&gt; [&quot;ac&quot;, &quot;bd&quot;]
+ tf.reduce_join(a, -1) = tf.reduce_join(a, 1) ==&gt; [&quot;ab&quot;, &quot;cd&quot;]
+ tf.reduce_join(a, 0, keep_dims=True) ==&gt; [[&quot;ac&quot;, &quot;bd&quot;]]
+ tf.reduce_join(a, 1, keep_dims=True) ==&gt; [[&quot;ab&quot;], [&quot;cd&quot;]]
+ tf.reduce_join(a, 0, separator=&quot;.&quot;) ==&gt; [&quot;a.c&quot;, &quot;b.d&quot;]
+ tf.reduce_join(a, [0, 1]) ==&gt; [&quot;acbd&quot;]
+ tf.reduce_join(a, [1, 0]) ==&gt; [&quot;abcd&quot;]
+ tf.reduce_join(a, []) ==&gt; [&quot;abcd&quot;]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:scatterSub" class="def">scatterSub</a> <a href="src/TensorFlow-GenOps-Core.html#scatterSub" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>ref</strong>: Should be from a <code>Variable</code> node.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices</td><td class="doc"><p><strong>indices</strong>: A tensor of indices into the first dimension of <code>ref</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>updates</strong>: A tensor of updated values to subtract from <code>ref</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output_ref</strong>: = Same as <code>ref</code>.  Returned as a convenience for operations that want
+ to use the updated values after the update is done.</p></td></tr></table></div><div class="doc"><p>Subtracts sparse updates to a variable reference.</p><p># Scalar indices
+     ref[indices, ...] -= updates[...]</p><p># Vector indices (for each i)
+     ref[indices[i], ...] -= updates[i, ...]</p><p># High rank indices (for each i, ..., j)
+     ref[indices[i, ..., j], ...] -= updates[i, ..., j, ...]</p><p>This operation outputs <code>ref</code> after the update is done.
+ This makes it easier to chain operations that need to use the reset value.</p><p>Duplicate entries are handled correctly: if multiple <code>indices</code> reference
+ the same location, their (negated) contributions add.</p><p>Requires `updates.shape = indices.shape + ref.shape[1:]`.</p><p><a href="div">style=&quot;width:70%; margin:auto; margin-bottom:10px; margin-top:20px;&quot;</a>
+ <a href="img">style=&quot;width:100%&quot; src=&quot;../../images/ScatterSub.png&quot; alt</a>
+ <a href="/div">/div</a></p></div></div><div class="top"><p class="src"><a name="v:scatterAdd" class="def">scatterAdd</a> <a href="src/TensorFlow-GenOps-Core.html#scatterAdd" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>ref</strong>: Should be from a <code>Variable</code> node.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices</td><td class="doc"><p><strong>indices</strong>: A tensor of indices into the first dimension of <code>ref</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>updates</strong>: A tensor of updated values to add to <code>ref</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output_ref</strong>: = Same as <code>ref</code>.  Returned as a convenience for operations that want
+ to use the updated values after the update is done.</p></td></tr></table></div><div class="doc"><p>Adds sparse updates to a variable reference.</p><p>This operation computes</p><p># Scalar indices
+     ref[indices, ...] += updates[...]</p><p># Vector indices (for each i)
+     ref[indices[i], ...] += updates[i, ...]</p><p># High rank indices (for each i, ..., j)
+     ref[indices[i, ..., j], ...] += updates[i, ..., j, ...]</p><p>This operation outputs <code>ref</code> after the update is done.
+ This makes it easier to chain operations that need to use the reset value.</p><p>Duplicate entries are handled correctly: if multiple <code>indices</code> reference
+ the same location, their contributions add.</p><p>Requires `updates.shape = indices.shape + ref.shape[1:]`.</p><p><a href="div">style=&quot;width:70%; margin:auto; margin-bottom:10px; margin-top:20px;&quot;</a>
+ <a href="img">style=&quot;width:100%&quot; src=&quot;../../images/ScatterAdd.png&quot; alt</a>
+ <a href="/div">/div</a></p></div></div><div class="top"><p class="src"><a name="v:scatterUpdate" class="def">scatterUpdate</a> <a href="src/TensorFlow-GenOps-Core.html#scatterUpdate" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>ref</strong>: Should be from a <code>Variable</code> node.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices</td><td class="doc"><p><strong>indices</strong>: A tensor of indices into the first dimension of <code>ref</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>updates</strong>: A tensor of updated values to store in <code>ref</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output_ref</strong>: = Same as <code>ref</code>.  Returned as a convenience for operations that want
+ to use the updated values after the update is done.</p></td></tr></table></div><div class="doc"><p>Applies sparse updates to a variable reference.</p><p>This operation computes</p><p># Scalar indices
+     ref[indices, ...] = updates[...]</p><p># Vector indices (for each i)
+     ref[indices[i], ...] = updates[i, ...]</p><p># High rank indices (for each i, ..., j)
+     ref[indices[i, ..., j], ...] = updates[i, ..., j, ...]</p><p>This operation outputs <code>ref</code> after the update is done.
+ This makes it easier to chain operations that need to use the reset value.</p><p>If values in <code>ref</code> is to be updated more than once, because there are
+ duplicate entires in <code>indices</code>, the order at which the updates happen
+ for each value is undefined.</p><p>Requires `updates.shape = indices.shape + ref.shape[1:]`.</p><p><a href="div">style=&quot;width:70%; margin:auto; margin-bottom:10px; margin-top:20px;&quot;</a>
+ <a href="img">style=&quot;width:100%&quot; src=&quot;../../images/ScatterUpdate.png&quot; alt</a>
+ <a href="/div">/div</a></p></div></div><div class="top"><p class="src"><a name="v:assignSub" class="def">assignSub</a> <a href="src/TensorFlow-GenOps-Core.html#assignSub" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>ref</strong>: Should be from a <code>Variable</code> node.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>value</strong>: The value to be subtracted to the variable.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output_ref</strong>: = Same as &quot;ref&quot;.  Returned as a convenience for operations that want
+ to use the new value after the variable has been updated.</p></td></tr></table></div><div class="doc"><p>Update <code>ref</code> by subtracting <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code> from it.</p><p>This operation outputs &quot;ref&quot; after the update is done.
+ This makes it easier to chain operations that need to use the reset value.</p></div></div><div class="top"><p class="src"><a name="v:assignAdd" class="def">assignAdd</a> <a href="src/TensorFlow-GenOps-Core.html#assignAdd" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>ref</strong>: Should be from a <code>Variable</code> node.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>value</strong>: The value to be added to the variable.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output_ref</strong>: = Same as &quot;ref&quot;.  Returned as a convenience for operations that want
+ to use the new value after the variable has been updated.</p></td></tr></table></div><div class="doc"><p>Update <code>ref</code> by adding <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code> to it.</p><p>This operation outputs &quot;ref&quot; after the update is done.
+ This makes it easier to chain operations that need to use the reset value.</p></div></div><div class="top"><p class="src"><a name="v:sparseSegmentMeanGrad" class="def">sparseSegmentMeanGrad</a> <a href="src/TensorFlow-GenOps-Core.html#sparseSegmentMeanGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>grad</strong>: gradient propagated to the SparseSegmentMean op.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>indices</strong>: indices passed to the corresponding SparseSegmentMean op.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>segment_ids</strong>: segment_ids passed to the corresponding SparseSegmentMean op.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>output_dim0</strong>: dimension 0 of &quot;data&quot; passed to SparseSegmentMean op.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Computes gradients for SparseSegmentMean.</p><p>Returns tensor &quot;output&quot; with same shape as grad, except for dimension 0 whose
+ value is output_dim0.</p></div></div><div class="top"><p class="src"><a name="v:sparseSoftmax" class="def">sparseSoftmax</a> <a href="src/TensorFlow-GenOps-Core.html#sparseSoftmax" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>sp_indices</strong>: 2-D.  `NNZ x R` matrix with the indices of non-empty values in a
+ SparseTensor, in canonical ordering.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>sp_values</strong>: 1-D.  <code>NNZ</code> non-empty values corresponding to <code>sp_indices</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>sp_shape</strong>: 1-D.  Shape of the input SparseTensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 1-D.  The <code>NNZ</code> values for the result <code>SparseTensor</code>.</p></td></tr></table></div><div class="doc"><p>Applies softmax to a batched N-D <code>SparseTensor</code>.</p><p>The inputs represent an N-D SparseTensor  with logical shape `[..., B, C]`
+ (where `N &gt;= 2`), and with indices sorted in the canonical lexicographic order.</p><p>This op is equivalent to applying the normal `tf.nn.softmax()` to each innermost
+ logical submatrix with shape `[B, C]`, but with the catch that *the implicitly
+ zero elements do not participate*.  Specifically, the algorithm is equivalent
+ to the following:</p><ol><li>Applies `tf.nn.softmax()` to a densified view of each innermost submatrix
+       with shape `[B, C]`, along the size-C dimension;</li><li>Masks out the original implicitly-zero locations;</li><li>Renormalizes the remaining elements.</li></ol><p>Hence, the <code>SparseTensor</code> result has exactly the same non-zero indices and
+ shape.</p></div></div><div class="top"><p class="src"><a name="v:matrixSolve" class="def">matrixSolve</a> <a href="src/TensorFlow-GenOps-Core.html#matrixSolve" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>matrix</strong>: Shape is `[..., M, M]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>rhs</strong>: Shape is `[..., M, K]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Shape is `[..., M, K]`.</p></td></tr></table></div><div class="doc"><p>Solves systems of linear equations.</p><p><code>Matrix</code> is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+ form square matrices. <code>Rhs</code> is a tensor of shape `[..., M, K]`. The <code>output</code> is
+ a tensor shape `[..., M, K]`.  If <code>adjoint</code> is <code><a href="../base-4.8.2.0/Data-Bool.html#v:False">False</a></code> then each output matrix
+ satisfies `matrix[..., :, :] * output[..., :, :] = rhs[..., :, :]`.
+ If <code>adjoint</code> is <code><a href="../base-4.8.2.0/Data-Bool.html#v:True">True</a></code> then each output matrix satisfies
+ `adjoint(matrix[..., :, :]) * output[..., :, :] = rhs[..., :, :]`.</p></div></div><div class="top"><p class="src"><a name="v:selfAdjointEigV2" class="def">selfAdjointEigV2</a> <a href="src/TensorFlow-GenOps-Core.html#selfAdjointEigV2" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> input of shape `[N, N]`.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><td class="doc"><p>(<strong>e</strong>, <strong>v</strong>)</p><ul><li><strong>e</strong>: Eigenvalues. Shape is `[N]`.</li><li><strong>v</strong>: Eigenvectors. Shape is `[N, N]`.</li></ul></td></tr></table></div><div class="doc"><p>Computes the eigen decomposition of one or more square self-adjoint matrices.</p><p>Computes the eigenvalues and (optionally) eigenvectors of each inner matrix in
+ <code>input</code> such that `input[..., :, :] = v[..., :, :] * diag(e[..., :])`.</p><p>```prettyprint
+ # a is a tensor.
+ # e is a tensor of eigenvalues.
+ # v is a tensor of eigenvectors.
+ e, v = self_adjoint_eig(a)
+ e = self_adjoint_eig(a, compute_v=False)
+ ```</p></div></div><div class="top"><p class="src"><a name="v:selfAdjointEig" class="def">selfAdjointEig</a> <a href="src/TensorFlow-GenOps-Core.html#selfAdjointEig" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Shape is `[..., M, M]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Shape is `[..., M+1, M]`.</p></td></tr></table></div><div class="doc"><p>Computes the Eigen Decomposition of a batch of square self-adjoint matrices.</p><p>The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+ form square matrices, with the same constraints as the single matrix
+ SelfAdjointEig.</p><p>The result is a [..., M+1, M] matrix with [..., 0,:] containing the
+ eigenvalues, and subsequent [...,1:, :] containing the eigenvectors.</p></div></div><div class="top"><p class="src"><a name="v:applyGradientDescent" class="def">applyGradientDescent</a> <a href="src/TensorFlow-GenOps-Core.html#applyGradientDescent" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>alpha</strong>: Scaling factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>delta</strong>: The change.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Update '*var' by subtracting <code>alpha</code> * <code>delta</code> from it.</p></div></div><div class="top"><p class="src"><a name="v:stackPush" class="def">stackPush</a> <a href="src/TensorFlow-GenOps-Core.html#stackPush" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a stack.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>elem</strong>: The tensor to be pushed onto the stack.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The same tensor as the input <code><a href="../base-4.8.2.0/Data-Foldable.html#v:elem">elem</a></code>.</p></td></tr></table></div><div class="doc"><p>Push an element onto the stack.</p></div></div><div class="top"><p class="src"><a name="v:cholesky" class="def">cholesky</a> <a href="src/TensorFlow-GenOps-Core.html#cholesky" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Shape is `[..., M, M]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Shape is `[..., M, M]`.</p></td></tr></table></div><div class="doc"><p>Computes the Cholesky decomposition of one or more square matrices.</p><p>The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+ form square matrices, with the same constraints as the single matrix Cholesky
+ decomposition above. The output is a tensor of the same shape as the input
+ containing the Cholesky decompositions for all input submatrices `[..., :, :]`.</p></div></div><div class="top"><p class="src"><a name="v:dynamicStitch" class="def">dynamicStitch</a> <a href="src/TensorFlow-GenOps-Core.html#dynamicStitch" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>]</td><td class="doc"><p><strong>indices</strong></p></td></tr><tr><td class="src">-&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t]</td><td class="doc"><p><strong>data</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>merged</strong></p></td></tr></table></div><div class="doc"><p>Interleave the values from the `data` tensors into a single tensor.</p><p>Builds a merged tensor such that</p><p>merged[indices[m][i, ..., j], ...] = data[m][i, ..., j, ...]</p><p>For example, if each `indices[m]` is scalar or vector, we have</p><p># Scalar indices
+     merged[indices[m], ...] = data[m][...]</p><p># Vector indices
+     merged[indices[m][i], ...] = data[m][i, ...]</p><p>Each `data[i].shape` must start with the corresponding `indices[i].shape`,
+ and the rest of `data[i].shape` must be constant w.r.t. <code>i</code>.  That is, we
+ must have `data[i].shape = indices[i].shape + constant`.  In terms of this
+ <code>constant</code>, the output shape is</p><p>merged.shape = [max(indices)] + constant</p><p>Values are merged in order, so if an index appears in both `indices[m][i]` and
+ `indices[n][j]` for `(m,i) &lt; (n,j)` the slice `data[n][j]` will appear in the
+ merged result.</p><p>For example:</p><p>indices[0] = 6
+     indices[1] = [4, 1]
+     indices[2] = [[5, 2], [0, 3]]
+     data[0] = [61, 62]
+     data[1] = [[41, 42], [11, 12]]
+     data[2] = [[[51, 52], [21, 22]], [[1, 2], [31, 32]]]
+     merged = [[1, 2], [11, 12], [21, 22], [31, 32], [41, 42],
+               [51, 52], [61, 62]]</p><p><a href="div">style=&quot;width:70%; margin:auto; margin-bottom:10px; margin-top:20px;&quot;</a>
+ <a href="img">style=&quot;width:100%&quot; src=&quot;../../images/DynamicStitch.png&quot; alt</a>
+ <a href="/div">/div</a></p></div></div><div class="top"><p class="src"><a name="v:readerNumWorkUnitsCompleted" class="def">readerNumWorkUnitsCompleted</a> <a href="src/TensorFlow-GenOps-Core.html#readerNumWorkUnitsCompleted" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>reader_handle</strong>: Handle to a Reader.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>units_completed</strong></p></td></tr></table></div><div class="doc"><p>Returns the number of work units this Reader has finished processing.</p></div></div><div class="top"><p class="src"><a name="v:readerRead" class="def">readerRead</a> <a href="src/TensorFlow-GenOps-Core.html#readerRead" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>reader_handle</strong>: Handle to a Reader.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>queue_handle</strong>: Handle to a Queue, with string work items.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>)</td><td class="doc"><p>(<strong>key</strong>, <strong>value</strong>)</p><ul><li><strong>key</strong>: A scalar.</li><li><strong>value</strong>: A scalar.</li></ul></td></tr></table></div><div class="doc"><p>Returns the next record (key, value pair) produced by a Reader.</p><p>Will dequeue from the input queue if necessary (e.g. when the
+ Reader needs to start reading from a new file since it has finished
+ with the previous file).</p></div></div><div class="top"><p class="src"><a name="v:fFT2D" class="def">fFT2D</a> <a href="src/TensorFlow-GenOps-Core.html#fFT2D" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>input</strong>: A complex64 tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>output</strong>: A complex64 tensor of the same shape as <code>input</code>. The inner-most 2
+ dimensions of <code>input</code> are replaced with their 2D Fourier Transform.</p></td></tr></table></div><div class="doc"><p>Compute the 2-dimensional discrete Fourier Transform over the inner-most</p><p>2 dimensions of <code>input</code>.</p></div></div><div class="top"><p class="src"><a name="v:fixedLengthRecordReader" class="def">fixedLengthRecordReader</a> <a href="src/TensorFlow-GenOps-Core.html#fixedLengthRecordReader" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>record_bytes</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>reader_handle</strong>: The handle to reference the Reader.</p></td></tr></table></div><div class="doc"><p>A Reader that outputs fixed-length records from a file.</p></div></div><div class="top"><p class="src"><a name="v:placeholder" class="def">placeholder</a> <a href="src/TensorFlow-GenOps-Core.html#placeholder" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</td><td class="doc"><p><strong>output</strong>: A placeholder tensor that must be replaced using the feed mechanism.</p></td></tr></table></div><div class="doc"><p>A placeholder op for a value that will be fed into the computation.</p><p>N.B. This operation will fail with an error if it is executed. It is
+ intended as a way to represent a value that will always be fed, and to
+ provide attrs that enable the fed value to be checked at runtime.</p></div></div><div class="top"><p class="src"><a name="v:scalarSummary" class="def">scalarSummary</a> <a href="src/TensorFlow-GenOps-Core.html#scalarSummary" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>tags</strong>: Tags for the summary.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>values</strong>: Same shape as `tags.  Values for the summary.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>summary</strong>: Scalar.  Serialized <code>Summary</code> protocol buffer.</p></td></tr></table></div><div class="doc"><p>Outputs a <code>Summary</code> protocol buffer with scalar values.</p><p>The input <code>tags</code> and <code>values</code> must have the same shape.  The generated summary
+ has a summary value for each tag-value pair in <code>tags</code> and <code>values</code>.</p></div></div><div class="top"><p class="src"><a name="v:softmax" class="def">softmax</a> <a href="src/TensorFlow-GenOps-Core.html#softmax" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>logits</strong>: 2-D with shape `[batch_size, num_classes]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>softmax</strong>: Same shape as <code>logits</code>.</p></td></tr></table></div><div class="doc"><p>Computes softmax activations.</p><p>For each batch <code>i</code> and class <code>j</code> we have</p><p>softmax[i, j] = exp(logits[i, j]) / sum_j(exp(logits[i, j]))</p></div></div><div class="top"><p class="src"><a name="v:shardedFilename" class="def">shardedFilename</a> <a href="src/TensorFlow-GenOps-Core.html#shardedFilename" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>basename</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>shard</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>num_shards</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>filename</strong></p></td></tr></table></div><div class="doc"><p>Generate a sharded filename. The filename is printf formatted as</p><p>%s-%05d-of-%05d, basename, shard, num_shards.</p></div></div><div class="top"><p class="src"><a name="v:_HostSend" class="def">_HostSend</a> <a href="src/TensorFlow-GenOps-Core.html#_HostSend" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>send_device_incarnation</strong>: The current incarnation of send_device.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>tensor</strong>: The tensor to send.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Sends the named tensor from send_device to recv_device.</p><p>_HostSend requires its input on host memory whereas _Send requires its
+ input on device memory.</p></div></div><div class="top"><p class="src"><a name="v:sigmoidGrad" class="def">sigmoidGrad</a> <a href="src/TensorFlow-GenOps-Core.html#sigmoidGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Computes the gradient of the sigmoid of <code>x</code> wrt its input.</p><p>Specifically, `grad = dy * y * (1 - y)`, where `y = sigmoid(x)`, and
+ <code>dy</code> is the corresponding input gradient.</p></div></div><div class="top"><p class="src"><a name="v:nonMaxSuppression" class="def">nonMaxSuppression</a> <a href="src/TensorFlow-GenOps-Core.html#nonMaxSuppression" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>boxes</strong>: A 2-D float tensor of shape `[num_boxes, 4]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>scores</strong>: A 1-D float tensor of shape `[num_boxes]` representing a single
+ score corresponding to each box (each row of boxes).</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>max_output_size</strong>: A scalar integer tensor representing the maximum number of
+ boxes to be selected by non max suppression.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>selected_indices</strong>: A 1-D integer tensor of shape `[M]` representing the selected
+ indices from the boxes tensor, where `M &lt;= max_output_size`.</p></td></tr></table></div><div class="doc"><p>Greedily selects a subset of bounding boxes in descending order of score,</p><p>pruning away boxes that have high intersection-over-union (IOU) overlap
+ with previously selected boxes.  Bounding boxes are supplied as
+ [y1, x1, y2, x2], where (y1, x1) and (y2, x2) are the coordinates of any
+ diagonal pair of box corners and the coordinates can be provided as normalized
+ (i.e., lying in the interval [0, 1]) or absolute.  Note that this algorithm
+ is agnostic to where the origin is in the coordinate system.  Note that this
+ algorithm is invariant to orthogonal transformations and translations
+ of the coordinate system; thus translating or reflections of the coordinate
+ system result in the same boxes being selected by the algorithm.</p><p>The output of this operation is a set of integers indexing into the input
+ collection of bounding boxes representing the selected boxes.  The bounding
+ box coordinates corresponding to the selected indices can then be obtained
+ using the tf.gather operation.  For example:</p><p>selected_indices = tf.image.non_max_suppression(
+       boxes, scores, max_output_size, iou_threshold)
+   selected_boxes = tf.gather(boxes, selected_indices)</p></div></div><div class="top"><p class="src"><a name="v:identityReader" class="def">identityReader</a> <a href="src/TensorFlow-GenOps-Core.html#identityReader" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>reader_handle</strong>: The handle to reference the Reader.</p></td></tr></table></div><div class="doc"><p>A Reader that outputs the queued work as both the key and value.</p><p>To use, enqueue strings in a Queue.  ReaderRead will take the front
+ work string and output (work, work).</p></div></div><div class="top"><p class="src"><a name="v:extractGlimpse" class="def">extractGlimpse</a> <a href="src/TensorFlow-GenOps-Core.html#extractGlimpse" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>input</strong>: A 4-D float tensor of shape `[batch_size, height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>size</strong>: A 1-D tensor of 2 elements containing the size of the glimpses
+ to extract.  The glimpse height must be specified first, following
+ by the glimpse width.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>offsets</strong>: A 2-D integer tensor of shape `[batch_size, 2]` containing
+ the x, y locations of the center of each window.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>glimpse</strong>: A tensor representing the glimpses `[batch_size,
+ glimpse_height, glimpse_width, channels]`.</p></td></tr></table></div><div class="doc"><p>Extracts a glimpse from the input tensor.</p><p>Returns a set of windows called glimpses extracted at location
+ <code>offsets</code> from the input tensor. If the windows only partially
+ overlaps the inputs, the non overlapping areas will be filled with
+ random noise.</p><p>The result is a 4-D tensor of shape `[batch_size, glimpse_height,
+ glimpse_width, channels]`. The channels and batch dimensions are the
+ same as that of the input tensor. The height and width of the output
+ windows are specified in the <code><a href="TensorFlow-GenOps-Core.html#v:size">size</a></code> parameter.</p><p>The argument <code>normalized</code> and <code>centered</code> controls how the windows are built:</p><ul><li>If the coordinates are normalized but not centered, 0.0 and 1.0
+   correspond to the minimum and maximum of each height and width
+   dimension.</li><li>If the coordinates are both normalized and centered, they range from</li><li>1.0 to 1.0. The coordinates (-1.0, -1.0) correspond to the upper
+   left corner, the lower right corner is located at (1.0, 1.0) and the
+   center is at (0, 0).</li><li>If the coordinates are not normalized they are interpreted as
+   numbers of pixels.</li></ul></div></div><div class="top"><p class="src"><a name="v:conv3DBackpropInput" class="def">conv3DBackpropInput</a> <a href="src/TensorFlow-GenOps-Core.html#conv3DBackpropInput" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Shape `[batch, depth, rows, cols, in_channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>filter</strong>: Shape `[depth, rows, cols, in_channels, out_channels]`.
+ <code>in_channels</code> must match between <code>input</code> and <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>out_backprop</strong>: Backprop signal of shape `[batch, out_depth, out_rows, out_cols,
+ out_channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Computes the gradients of 3-D convolution with respect to the input.</p></div></div><div class="top"><p class="src"><a name="v:matrixSolveLs" class="def">matrixSolveLs</a> <a href="src/TensorFlow-GenOps-Core.html#matrixSolveLs" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>matrix</strong>: Shape is `[..., M, N]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>rhs</strong>: Shape is `[..., M, K]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a></td><td class="doc"><p><strong>l2_regularizer</strong>: Scalar tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Shape is `[..., N, K]`.</p></td></tr></table></div><div class="doc"><p>Solves one or more linear least-squares problems.</p><p><code>matrix</code> is a tensor of shape `[..., M, N]` whose inner-most 2 dimensions
+ form matrices of size `[M, N]`. Rhs is a tensor of shape `[..., M, K]`.
+ The output is a tensor shape `[..., N, K]` where each output matrix solves
+ each of the equations matrix[..., :, :] * output[..., :, :] = rhs[..., :, :]
+ in the least squares sense.</p><p>matrix and right-hand sides in the batch:</p><p><code>matrix</code>=\(A in Re^{m times n}\),
+ <code>rhs</code>=\(B  in Re^{m times k}\),
+ <code>output</code>=\(X  in Re^{n times k}\),
+ <code>l2_regularizer</code>=\(lambda\).</p><p>If <code>fast</code> is <code><a href="../base-4.8.2.0/Data-Bool.html#v:True">True</a></code>, then the solution is computed by solving the normal
+ equations using Cholesky decomposition. Specifically, if \(m ge n\) then
+ \(X = (A^T A + lambda I)^{-1} A^T B\), which solves the least-squares
+ problem \(X = mathrm{argmin}_{Z in Re^{n times k}} ||A Z - B||_F^2 +
+ lambda ||Z||_F^2\). If \(m lt n\) then <code>output</code> is computed as
+ \(X = A^T (A A^T + lambda I)^{-1} B\), which (for \(lambda = 0\)) is the
+ minimum-norm solution to the under-determined linear system, i.e.
+ \(X = mathrm{argmin}_{Z in Re^{n times k}} ||Z||_F^2 \), subject to
+ \(A Z = B\). Notice that the fast path is only numerically stable when
+ \(A\) is numerically full rank and has a condition number
+ \(mathrm{cond}(A) lt frac{1}{sqrt{epsilon_{mach}}}\) or\(lambda\) is
+ sufficiently large.</p><p>If <code>fast</code> is <code><a href="../base-4.8.2.0/Data-Bool.html#v:False">False</a></code> an algorithm based on the numerically robust complete
+ orthogonal decomposition is used. This computes the minimum-norm
+ least-squares solution, even when \(A\) is rank deficient. This path is
+ typically 6-7 times slower than the fast path. If <code>fast</code> is <code><a href="../base-4.8.2.0/Data-Bool.html#v:False">False</a></code> then
+ <code>l2_regularizer</code> is ignored.</p></div></div><div class="top"><p class="src"><a name="v:rGBToHSV" class="def">rGBToHSV</a> <a href="src/TensorFlow-GenOps-Core.html#rGBToHSV" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>images</strong>: 1-D or higher rank. RGB data to convert. Last dimension must be size 3.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: <code>images</code> converted to HSV.</p></td></tr></table></div><div class="doc"><p>Converts one or more images from RGB to HSV.</p><p>Outputs a tensor of the same shape as the <code>images</code> tensor, containing the HSV
+ value of the pixels. The output is only well defined if the value in <code>images</code>
+ are in `[0,1]`.</p><p>`output[..., 0]` contains hue, `output[..., 1]` contains saturation, and
+ `output[..., 2]` contains value. All HSV values are in `[0,1]`. A hue of 0
+ corresponds to pure red, hue 1<em>3 is pure green, and 2</em>3 is pure blue.</p></div></div><div class="top"><p class="src"><a name="v:decodeGif" class="def">decodeGif</a> <a href="src/TensorFlow-GenOps-Core.html#decodeGif" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>contents</strong>: 0-D.  The GIF-encoded image.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a></td><td class="doc"><p><strong>image</strong>: 4-D with shape `[num_frames, height, width, 3]`. RGB order</p></td></tr></table></div><div class="doc"><p>Decode the first frame of a GIF-encoded image to a uint8 tensor.</p><p>GIF with frame or transparency compression are not supported
+ convert animated GIF from compressed to uncompressed by:</p><p>convert $src.gif -coalesce $dst.gif</p></div></div><div class="top"><p class="src"><a name="v:adjustContrast" class="def">adjustContrast</a> <a href="src/TensorFlow-GenOps-Core.html#adjustContrast" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>images</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>contrast_factor</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>min_value</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>max_value</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Deprecated. Disallowed in GraphDef version &gt;= 2.</p></div></div><div class="top"><p class="src"><a name="v:depthToSpace" class="def">depthToSpace</a> <a href="src/TensorFlow-GenOps-Core.html#depthToSpace" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>block_size</strong>: The size of the spatial block, same as in Space2Depth.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>DepthToSpace for tensors of type T.</p><p>Rearranges data from depth into blocks of spatial data.
+ This is the reverse transformation of SpaceToDepth. More specifically,
+ this op outputs a copy of the input tensor where values from the <code>depth</code>
+ dimension are moved in spatial blocks to the <code>height</code> and <code>width</code> dimensions.
+ The attr <code>block_size</code> indicates the input block size and how the data is moved.</p><ul><li>Chunks of data of size `block_size * block_size` from depth are rearranged
+     into non-overlapping blocks of size `block_size x block_size`</li><li>The width the output tensor is `input_depth * block_size`, whereas the
+     height is `input_height * block_size`.</li><li>The depth of the input tensor must be divisible by
+     `block_size * block_size`.</li></ul><p>That is, assuming the input is in the shape:
+ `[batch, height, width, depth]`,
+ the shape of the output will be:
+ `[batch, height*block_size, width*block_size, depth/(block_size*block_size)]`</p><p>This operation requires that the input tensor be of rank 4, and that
+ <code>block_size</code> be &gt;=1 and that `block_size * block_size` be a divisor of the
+ input depth.</p><p>This operation is useful for resizing the activations between convolutions
+ (but keeping all data), e.g. instead of pooling. It is also useful for training
+ purely convolutional models.</p><p>For example, given this input of shape `[1, 1, 1, 4]`, and a block size of 2:</p><p>```prettyprint
+ x = [[[[1, 2, 3, 4]]]]</p><p>```</p><p>This operation will output a tensor of shape `[1, 2, 2, 1]`:</p><p>```prettyprint
+    [[[[1], [2]],
+      [[3], [4]]]]
+ ```</p><p>Here, the input has a batch of 1 and each batch element has shape `[1, 1, 4]`,
+ the corresponding output will have 2x2 elements and will have a depth of
+ 1 channel (1 = `4 / (block_size * block_size)`).
+ The output element shape is `[2, 2, 1]`.</p><p>For an input tensor with larger depth, here of shape `[1, 1, 1, 12]`, e.g.</p><p>```prettyprint
+ x = [[[[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]]]]
+ ```</p><p>This operation, for block size of 2, will return the following tensor of shape
+ `[1, 2, 2, 3]`</p><p>```prettyprint
+    [[[[1, 2, 3], [4, 5, 6]],
+      [[7, 8, 9], [10, 11, 12]]]]</p><p>```</p><p>Similarly, for the following input of shape `[1 2 2 4]`, and a block size of 2:</p><p>```prettyprint
+ x =  [[[[1, 2, 3, 4],
+        [5, 6, 7, 8]],
+       [[9, 10, 11, 12],
+        [13, 14, 15, 16]]]]
+ ```</p><p>the operator will return the following tensor of shape `[1 4 4 1]`:</p><p>```prettyprint
+ x = [[ [1],   [2],  [5],  [6]],
+      [ [3],   [4],  [7],  [8]],
+      [ [9],  [10], [13],  [14]],
+      [ [11], [12], [15],  [16]]]</p><p>```</p></div></div><div class="top"><p class="src"><a name="v:batchMatrixSolve" class="def">batchMatrixSolve</a> <a href="src/TensorFlow-GenOps-Core.html#batchMatrixSolve" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>matrix</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>rhs</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:erfc" class="def">erfc</a> <a href="src/TensorFlow-GenOps-Core.html#erfc" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes the complementary error function of <code>x</code> element-wise.</p></div></div><div class="top"><p class="src"><a name="v:resizeBilinearGrad" class="def">resizeBilinearGrad</a> <a href="src/TensorFlow-GenOps-Core.html#resizeBilinearGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>grads</strong>: 4-D with shape `[batch, height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>original_image</strong>: 4-D with shape `[batch, orig_height, orig_width, channels]`,
+ The image tensor that was resized.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 4-D with shape `[batch, orig_height, orig_width, channels]`.
+ Gradients with respect to the input image. Input image must have been
+ float or double.</p></td></tr></table></div><div class="doc"><p>Computes the gradient of bilinear interpolation.</p></div></div><div class="top"><p class="src"><a name="v:fact" class="def">fact</a> <a href="src/TensorFlow-GenOps-Core.html#fact" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>fact</strong></p></td></tr></table></div><div class="doc"><p>Output a fact about factorials.</p></div></div><div class="top"><p class="src"><a name="v:deleteSessionTensor" class="def">deleteSessionTensor</a> <a href="src/TensorFlow-GenOps-Core.html#deleteSessionTensor" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle for a tensor stored in the session state.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Delete the tensor specified by its handle in the session.</p></div></div><div class="top"><p class="src"><a name="v:logicalOr" class="def">logicalOr</a> <a href="src/TensorFlow-GenOps-Core.html#logicalOr" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns the truth value of x OR y element-wise.</p><ul><li>NOTE*: <code>LogicalOr</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:getSessionTensor" class="def">getSessionTensor</a> <a href="src/TensorFlow-GenOps-Core.html#getSessionTensor" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle for a tensor stored in the session state.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</td><td class="doc"><p><strong>value</strong>: The tensor for the given handle.</p></td></tr></table></div><div class="doc"><p>Get the value of the tensor specified by its handle.</p></div></div><div class="top"><p class="src"><a name="v:batchMatrixInverse" class="def">batchMatrixInverse</a> <a href="src/TensorFlow-GenOps-Core.html#batchMatrixInverse" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:shardedFilespec" class="def">shardedFilespec</a> <a href="src/TensorFlow-GenOps-Core.html#shardedFilespec" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>basename</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>num_shards</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>filename</strong></p></td></tr></table></div><div class="doc"><p>Generate a glob pattern matching all sharded file names.</p></div></div><div class="top"><p class="src"><a name="v:decodeBase64" class="def">decodeBase64</a> <a href="src/TensorFlow-GenOps-Core.html#decodeBase64" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>input</strong>: Base64 strings to decode.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>output</strong>: Decoded strings.</p></td></tr></table></div><div class="doc"><p>Decode web-safe base64-encoded strings.</p><p>Input may or may not have padding at the end. See EncodeBase64 for padding.
+ Web-safe means that input must use - and _ instead of + and /.</p></div></div><div class="top"><p class="src"><a name="v:getSessionHandle" class="def">getSessionHandle</a> <a href="src/TensorFlow-GenOps-Core.html#getSessionHandle" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>value</strong>: The tensor to be stored.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle for the tensor stored in the session state.</p></td></tr></table></div><div class="doc"><p>Store the input tensor in the state of the current session.</p></div></div><div class="top"><p class="src"><a name="v:initializeTable" class="def">initializeTable</a> <a href="src/TensorFlow-GenOps-Core.html#initializeTable" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tkey, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tval)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>table_handle</strong>: Handle to a table which will be initialized.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tkey</td><td class="doc"><p><strong>keys</strong>: Keys of type Tkey.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tval</td><td class="doc"><p><strong>values</strong>: Values of type Tval.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Table initializer that takes two tensors for keys and values respectively.</p></div></div><div class="top"><p class="src"><a name="v:tan" class="def">tan</a> <a href="src/TensorFlow-GenOps-Core.html#tan" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes tan of x element-wise.</p></div></div><div class="top"><p class="src"><a name="v:tanh" class="def">tanh</a> <a href="src/TensorFlow-GenOps-Core.html#tanh" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes hyperbolic tangent of <code>x</code> element-wise.</p></div></div><div class="top"><p class="src"><a name="v:applyAdagradDA" class="def">applyAdagradDA</a> <a href="src/TensorFlow-GenOps-Core.html#applyAdagradDA" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>gradient_accumulator</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>gradient_squared_accumulator</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>grad</strong>: The gradient.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>lr</strong>: Scaling factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t</td><td class="doc"><p><strong>l1</strong>: L1 regularization. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t</td><td class="doc"><p><strong>l2</strong>: L2 regularization. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v8 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>global_step</strong>: Training step number. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Update '*var' according to the proximal adagrad scheme.</p></div></div><div class="top"><p class="src"><a name="v:stringToHashBucket" class="def">stringToHashBucket</a> <a href="src/TensorFlow-GenOps-Core.html#stringToHashBucket" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_buckets</strong>: The number of buckets.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>string_tensor</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>output</strong>: A Tensor of the same shape as the input <code>string_tensor</code>.</p></td></tr></table></div><div class="doc"><p>Converts each string in the input Tensor to its hash mod by a number of buckets.</p><p>The hash function is deterministic on the content of the string within the
+ process.</p><p>Note that the hash function may change from time to time.
+ This functionality will be deprecated and it's recommended to use
+ `tf.string_to_hash_bucket_fast()` or `tf.string_to_hash_bucket_strong()`.</p></div></div><div class="top"><p class="src"><a name="v:eluGrad" class="def">eluGrad</a> <a href="src/TensorFlow-GenOps-Core.html#eluGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>gradients</strong>: The backpropagated gradients to the corresponding Elu operation.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>outputs</strong>: The outputs of the corresponding Elu operation.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>backprops</strong>: The gradients: `gradients * (outputs + 1)` if outputs &lt; 0,
+ <code>gradients</code> otherwise.</p></td></tr></table></div><div class="doc"><p>Computes gradients for the exponential linear (Elu) operation.</p></div></div><div class="top"><p class="src"><a name="v:fractionalAvgPoolGrad" class="def">fractionalAvgPoolGrad</a> <a href="src/TensorFlow-GenOps-Core.html#fractionalAvgPoolGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>orig_input_tensor_shape</strong>: Original input tensor shape for <code>fractional_avg_pool</code></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>out_backprop</strong>: 4-D with shape `[batch, height, width, channels]`.  Gradients
+ w.r.t. the output of <code>fractional_avg_pool</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>row_pooling_sequence</strong>: row pooling sequence, form pooling region with
+ col_pooling_sequence.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>col_pooling_sequence</strong>: column pooling sequence, form pooling region with
+ row_pooling sequence.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 4-D.  Gradients w.r.t. the input of <code>fractional_avg_pool</code>.</p></td></tr></table></div><div class="doc"><p>Computes gradient of the FractionalAvgPool function.</p><p>Unlike FractionalMaxPoolGrad, we don't need to find arg_max for
+ FractionalAvgPoolGrad, we just need to evenly back-propagate each element of
+ out_backprop to those indices that form the same pooling cell. Therefore, we
+ just need to know the shape of original input tensor, instead of the whole
+ tensor.</p></div></div><div class="top"><p class="src"><a name="v:matrixTriangularSolve" class="def">matrixTriangularSolve</a> <a href="src/TensorFlow-GenOps-Core.html#matrixTriangularSolve" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>matrix</strong>: Shape is `[..., M, M]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>rhs</strong>: Shape is `[..., M, K]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Shape is `[..., M, K]`.</p></td></tr></table></div><div class="doc"><p>Solves systems of linear equations with upper or lower triangular matrices by</p><p>backsubstitution.</p><p><code>matrix</code> is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions form
+ square matrices. If <code>lower</code> is <code><a href="../base-4.8.2.0/Data-Bool.html#v:True">True</a></code> then the strictly upper triangular part
+ of each inner-most matrix is assumed to be zero and not accessed.
+ If <code>lower</code> is False then the strictly lower triangular part of each inner-most
+ matrix is assumed to be zero and not accessed.
+ <code>rhs</code> is a tensor of shape `[..., M, K]`.</p><p>The output is a tensor of shape `[..., M, K]`. If <code>adjoint</code> is
+ <code><a href="../base-4.8.2.0/Data-Bool.html#v:True">True</a></code> then the innermost matrices in output` satisfy matrix equations
+ `matrix[..., :, :] * output[..., :, :] = rhs[..., :, :]`.
+ If <code>adjoint</code> is <code><a href="../base-4.8.2.0/Data-Bool.html#v:False">False</a></code> then the strictly then the  innermost matrices in
+ <code>output</code> satisfy matrix equations
+ `adjoint(matrix[..., i, k]) * output[..., k, j] = rhs[..., i, j]`.</p></div></div><div class="top"><p class="src"><a name="v:editDistance" class="def">editDistance</a> <a href="src/TensorFlow-GenOps-Core.html#editDistance" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>hypothesis_indices</strong>: The indices of the hypothesis list SparseTensor.
+ This is an N x R int64 matrix.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>hypothesis_values</strong>: The values of the hypothesis list SparseTensor.
+ This is an N-length vector.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>hypothesis_shape</strong>: The shape of the hypothesis list SparseTensor.
+ This is an R-length vector.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>truth_indices</strong>: The indices of the truth list SparseTensor.
+ This is an M x R int64 matrix.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>truth_values</strong>: The values of the truth list SparseTensor.
+ This is an M-length vector.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>truth_shape</strong>: truth indices, vector.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>output</strong>: A dense float tensor with rank R - 1.</p><p>For the example input:</p><p>// hypothesis represents a 2x1 matrix with variable-length values:
+     //   (0,0) = [&quot;a&quot;]
+     //   (1,0) = [&quot;b&quot;]
+     hypothesis_indices = [[0, 0, 0],
+                           [1, 0, 0]]
+     hypothesis_values = [&quot;a&quot;, &quot;b&quot;]
+     hypothesis_shape = [2, 1, 1]</p><p>// truth represents a 2x2 matrix with variable-length values:
+     //   (0,0) = []
+     //   (0,1) = [&quot;a&quot;]
+     //   (1,0) = [&quot;b&quot;, &quot;c&quot;]
+     //   (1,1) = [&quot;a&quot;]
+     truth_indices = [[0, 1, 0],
+                      [1, 0, 0],
+                      [1, 0, 1],
+                      [1, 1, 0]]
+     truth_values = [&quot;a&quot;, &quot;b&quot;, &quot;c&quot;, &quot;a&quot;]
+     truth_shape = [2, 2, 2]
+     normalize = true</p><p>The output will be:</p><p>// output is a 2x2 matrix with edit distances normalized by truth lengths.
+     output = [[inf, 1.0],  // (0,0): no truth, (0,1): no hypothesis
+               [0.5, 1.0]]  // (1,0): addition, (1,1): no hypothesis</p></td></tr></table></div><div class="doc"><p>Computes the (possibly normalized) Levenshtein Edit Distance.</p><p>The inputs are variable-length sequences provided by SparseTensors
+   (hypothesis_indices, hypothesis_values, hypothesis_shape)
+ and
+   (truth_indices, truth_values, truth_shape).</p><p>The inputs are:</p></div></div><div class="top"><p class="src"><a name="v:barrierIncompleteSize" class="def">barrierIncompleteSize</a> <a href="src/TensorFlow-GenOps-Core.html#barrierIncompleteSize" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a barrier.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>size</strong>: The number of incomplete elements (i.e. those with some of their value
+ components not set) in the barrier.</p></td></tr></table></div><div class="doc"><p>Computes the number of incomplete elements in the given barrier.</p></div></div><div class="top"><p class="src"><a name="v:threadUnsafeUnigramCandidateSampler" class="def">threadUnsafeUnigramCandidateSampler</a> <a href="src/TensorFlow-GenOps-Core.html#threadUnsafeUnigramCandidateSampler" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_sampled</strong>: Number of candidates to randomly sample per batch.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_true</strong>: Number of true labels per context.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>range_max</strong>: The sampler will sample integers from the interval [0, range_max).</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>unique</strong>: If unique is true, we sample with rejection, so that all sampled
+ candidates in a batch are unique. This requires some approximation to
+ estimate the post-rejection sampling probabilities.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>true_classes</strong>: A batch_size * num_true matrix, in which each row contains the
+ IDs of the num_true target_classes in the corresponding original label.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p>(<strong>sampled_candidates</strong>, <strong>true_expected_count</strong>, <strong>sampled_expected_count</strong>)</p><ul><li><strong>sampled_candidates</strong>: A vector of length num_sampled, in which each element is
+ the ID of a sampled candidate.</li><li><strong>true_expected_count</strong>: A batch_size * num_true matrix, representing
+ the number of times each candidate is expected to occur in a batch
+ of sampled candidates. If unique=true, then this is a probability.</li><li><strong>sampled_expected_count</strong>: A vector of length num_sampled, for each sampled
+ candidate representing the number of times the candidate is expected
+ to occur in a batch of sampled candidates.  If unique=true, then this is a
+ probability.</li></ul></td></tr></table></div><div class="doc"><p>Generates labels for candidate sampling with a learned unigram distribution.</p><p>See explanations of candidate sampling and the data formats at
+ go/candidate-sampling.</p><p>For each batch, this op picks a single set of sampled candidate labels.</p><p>The advantages of sampling candidates per-batch are simplicity and the
+ possibility of efficient dense matrix multiplication. The disadvantage is that
+ the sampled candidates must be chosen independently of the context and of the
+ true labels.</p></div></div><div class="top"><p class="src"><a name="v:barrierReadySize" class="def">barrierReadySize</a> <a href="src/TensorFlow-GenOps-Core.html#barrierReadySize" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a barrier.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>size</strong>: The number of complete elements (i.e. those with all of their value
+ components set) in the barrier.</p></td></tr></table></div><div class="doc"><p>Computes the number of complete elements in the given barrier.</p></div></div><div class="top"><p class="src"><a name="v:barrierClose" class="def">barrierClose</a> <a href="src/TensorFlow-GenOps-Core.html#barrierClose" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a barrier.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Closes the given barrier.</p><p>This operation signals that no more new elements will be inserted in the
+ given barrier. Subsequent InsertMany that try to introduce a new key will fail.
+ Subsequent InsertMany operations that just add missing components to already
+ existing elements will continue to succeed. Subsequent TakeMany operations will
+ continue to succeed if sufficient completed elements remain in the barrier.
+ Subsequent TakeMany operations that would block will fail immediately.</p></div></div><div class="top"><p class="src"><a name="v:textLineReader" class="def">textLineReader</a> <a href="src/TensorFlow-GenOps-Core.html#textLineReader" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>reader_handle</strong>: The handle to reference the Reader.</p></td></tr></table></div><div class="doc"><p>A Reader that outputs the lines of a file delimited by '\n'.</p></div></div><div class="top"><p class="src"><a name="v:fFT3D" class="def">fFT3D</a> <a href="src/TensorFlow-GenOps-Core.html#fFT3D" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>input</strong>: A complex64 tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>output</strong>: A complex64 tensor of the same shape as <code>input</code>. The inner-most 3
+ dimensions of <code>input</code> are replaced with their 3D Fourier Transform.</p></td></tr></table></div><div class="doc"><p>Compute the 3-dimensional discrete Fourier Transform over the inner-most 3</p><p>dimensions of <code>input</code>.</p></div></div><div class="top"><p class="src"><a name="v:refExit" class="def">refExit</a> <a href="src/TensorFlow-GenOps-Core.html#refExit" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong>: The tensor to be made available to the parent frame.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The same tensor as `data`.</p></td></tr></table></div><div class="doc"><p>Exits the current frame to its parent frame.</p><p>Exit makes its input `data` available to the parent frame.</p></div></div><div class="top"><p class="src"><a name="v:exp" class="def">exp</a> <a href="src/TensorFlow-GenOps-Core.html#exp" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes exponential of x element-wise.  \(y = e^x\).</p></div></div><div class="top"><p class="src"><a name="v:restoreSlice" class="def">restoreSlice</a> <a href="src/TensorFlow-GenOps-Core.html#restoreSlice" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dt</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>file_pattern</strong>: Must have a single element. The pattern of the files from
+ which we read the tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>tensor_name</strong>: Must have a single element. The name of the tensor to be
+ restored.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>shape_and_slice</strong>: Scalar. The shapes and slice specifications to use when
+ restoring a tensors.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dt</td><td class="doc"><p><strong>tensor</strong>: The restored tensor.</p></td></tr></table></div><div class="doc"><p>Restores a tensor from checkpoint files.</p><p>This is like <code>Restore</code> except that restored tensor can be listed as filling
+ only a slice of a larger tensor.  <code>shape_and_slice</code> specifies the shape of the
+ larger tensor and the slice that the restored tensor covers.</p><p>The <code>shape_and_slice</code> input has the same format as the
+ elements of the <code>shapes_and_slices</code> input of the <code>SaveSlices</code> op.</p></div></div><div class="top"><p class="src"><a name="v:conj" class="def">conj</a> <a href="src/TensorFlow-GenOps-Core.html#conj" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Returns the complex conjugate of a complex number.</p><p>Given a tensor <code>input</code> of complex numbers, this operation returns a tensor of
+ complex numbers that are the complex conjugate of each element in <code>input</code>. The
+ complex numbers in <code>input</code> must be of the form \(a + bj\), where *a* is the
+ real part and *b* is the imaginary part.</p><p>The complex conjugate returned by this operation is of the form \(a - bj\).</p><p>For example:</p><p>```
+ # tensor <code>input</code> is [-2.25 + 4.75j, 3.25 + 5.75j]
+ tf.conj(input) ==&gt; [-2.25 - 4.75j, 3.25 - 5.75j]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:resizeNearestNeighborGrad" class="def">resizeNearestNeighborGrad</a> <a href="src/TensorFlow-GenOps-Core.html#resizeNearestNeighborGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>grads</strong>: 4-D with shape `[batch, height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>size</strong>: = A 1-D int32 Tensor of 2 elements: `orig_height, orig_width`. The
+ original input size.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 4-D with shape `[batch, orig_height, orig_width, channels]`. Gradients
+ with respect to the input image.</p></td></tr></table></div><div class="doc"><p>Computes the gradient of nearest neighbor interpolation.</p></div></div><div class="top"><p class="src"><a name="v:tensorArrayClose" class="def">tensorArrayClose</a> <a href="src/TensorFlow-GenOps-Core.html#tensorArrayClose" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a TensorArray (output of TensorArray or TensorArrayGrad).</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Delete the TensorArray from its resource container.  This enables</p><p>the user to close and release the resource in the middle of a step/run.</p></div></div><div class="top"><p class="src"><a name="v:atan" class="def">atan</a> <a href="src/TensorFlow-GenOps-Core.html#atan" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes atan of x element-wise.</p></div></div><div class="top"><p class="src"><a name="v:tensorArraySize" class="def">tensorArraySize</a> <a href="src/TensorFlow-GenOps-Core.html#tensorArraySize" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a TensorArray (output of TensorArray or TensorArrayGrad).</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>flow_in</strong>: A float scalar that enforces proper chaining of operations.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>size</strong>: The current size of the TensorArray.</p></td></tr></table></div><div class="doc"><p>Get the current size of the TensorArray.</p></div></div><div class="top"><p class="src"><a name="v:tensorArrayConcat" class="def">tensorArrayConcat</a> <a href="src/TensorFlow-GenOps-Core.html#tensorArrayConcat" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a TensorArray.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>flow_in</strong>: A float scalar that enforces proper chaining of operations.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</td><td class="doc"><p>(<strong>value</strong>, <strong>lengths</strong>)</p><ul><li><strong>value</strong>: All of the elements in the TensorArray, concatenated along the first
+ axis.</li><li><strong>lengths</strong>: A vector of the row sizes of the original T elements in the
+ value output.  In the example above, this would be the values:
+ `(n1, n2, ..., n(T-1))`.</li></ul></td></tr></table></div><div class="doc"><p>Concat the elements from the TensorArray into value <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>.</p><p>Takes <code>T</code> elements of shapes</p><p>```
+   (n0 x d0 x d1 x ...), (n1 x d0 x d1 x ...), ..., (n(T-1) x d0 x d1 x ...)
+   ```</p><p>and concatenates them into a Tensor of shape:</p><p>```(n0 + n1 + ... + n(T-1) x d0 x d1 x ...)```</p><p>All elements must have the same shape (excepting the first dimension).</p></div></div><div class="top"><p class="src"><a name="v:lRN" class="def">lRN</a> <a href="src/TensorFlow-GenOps-Core.html#lRN" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: 4-D.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Local Response Normalization.</p><p>The 4-D <code>input</code> tensor is treated as a 3-D array of 1-D vectors (along the last
+ dimension), and each vector is normalized independently.  Within a given vector,
+ each component is divided by the weighted, squared sum of inputs within
+ <code>depth_radius</code>.  In detail,</p><p>sqr_sum[a, b, c, d] =
+         sum(input[a, b, c, d - depth_radius : d + depth_radius + 1] ** 2)
+     output = input / (bias + alpha * sqr_sum) ** beta</p><p>For details, see <a href="http://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks">Krizhevsky et al., ImageNet classification with deep
+ convolutional neural networks (NIPS 2012)</a>.</p></div></div><div class="top"><p class="src"><a name="v:stringToHashBucketFast" class="def">stringToHashBucketFast</a> <a href="src/TensorFlow-GenOps-Core.html#stringToHashBucketFast" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_buckets</strong>: The number of buckets.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>input</strong>: The strings to assign a hash bucket.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>output</strong>: A Tensor of the same shape as the input <code>string_tensor</code>.</p></td></tr></table></div><div class="doc"><p>Converts each string in the input Tensor to its hash mod by a number of buckets.</p><p>The hash function is deterministic on the content of the string within the
+ process and will never change. However, it is not suitable for cryptography.
+ This function may be used when CPU time is scarce and inputs are trusted or
+ unimportant. There is a risk of adversaries constructing inputs that all hash
+ to the same bucket. To prevent this problem, use a strong hash function with
+ `tf.string_to_hash_bucket_strong`.</p></div></div><div class="top"><p class="src"><a name="v:tensorArrayPack" class="def">tensorArrayPack</a> <a href="src/TensorFlow-GenOps-Core.html#tensorArrayPack" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a TensorArray.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>flow_in</strong>: A float scalar that enforces proper chaining of operations.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</td><td class="doc"><p><strong>value</strong>: All of the elements in the TensorArray, concatenated along a new
+ axis (the new dimension 0).</p></td></tr></table></div><div class="doc"><p>Pack the elements from the TensorArray into output <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>.</p><ul><li>*WARNING: This op is deprecated.**</li></ul><p>Instead of this op, use <code>TensorArrayGather</code> with
+ `indices = RangeOp(0, TensorArraySizeOp)`.</p><p>All elements must have the same shape.</p></div></div><div class="top"><p class="src"><a name="v:concatOffset" class="def">concatOffset</a> <a href="src/TensorFlow-GenOps-Core.html#concatOffset" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>concat_dim</strong>: The dimension along which to concatenate.</p></td></tr><tr><td class="src">-&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>]</td><td class="doc"><p><strong>shape</strong>: The <code>N</code> int32 vectors representing shape of tensors being concatenated.</p></td></tr><tr><td class="src">-&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>]</td><td class="doc"><p><strong>offset</strong>: The <code>N</code> int32 vectors representing the starting offset
+         of input tensors within the concatenated output.</p><p>This is typically used by gradient computations for a concat operation.</p></td></tr></table></div><div class="doc"><p>Computes offsets of concat inputs within its output.</p><p>For example:</p><p>```prettyprint
+ # <code>x</code> is [2, 2, 7]
+ # <code>y</code> is [2, 3, 7]
+ # <code>z</code> is [2, 5, 7]
+ concat_offset(2, [x, y, z]) =&gt; [0, 0, 0], [0, 2, 0], [0, 5, 0]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:refEnter" class="def">refEnter</a> <a href="src/TensorFlow-GenOps-Core.html#refEnter" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong>: The tensor to be made available to the child frame.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The same tensor as `data`.</p></td></tr></table></div><div class="doc"><p>Creates or finds a child frame, and makes `data` available to the child frame.</p><p>The unique <code>frame_name</code> is used by the <code>Executor</code> to identify frames. If
+ <code>is_constant</code> is true, <code>output</code> is a constant in the child frame; otherwise
+ it may be changed in the child frame. At most <code>parallel_iterations</code> iterations
+ are run in parallel in the child frame.</p></div></div><div class="top"><p class="src"><a name="v:softsign" class="def">softsign</a> <a href="src/TensorFlow-GenOps-Core.html#softsign" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>features</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>activations</strong></p></td></tr></table></div><div class="doc"><p>Computes softsign: `features / (abs(features) + 1)`.</p></div></div><div class="top"><p class="src"><a name="v:tensorArrayWrite" class="def">tensorArrayWrite</a> <a href="src/TensorFlow-GenOps-Core.html#tensorArrayWrite" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a TensorArray.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>index</strong>: The position to write to inside the TensorArray.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>value</strong>: The tensor to write to the TensorArray.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>flow_in</strong>: A float scalar that enforces proper chaining of operations.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>flow_out</strong>: A float scalar that enforces proper chaining of operations.</p></td></tr></table></div><div class="doc"><p>Push an element onto the tensor_array.</p></div></div><div class="top"><p class="src"><a name="v:diag" class="def">diag</a> <a href="src/TensorFlow-GenOps-Core.html#diag" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>diagonal</strong>: Rank k tensor where k is at most 3.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Returns a diagonal tensor with a given diagonal values.</p><p>Given a <code>diagonal</code>, this operation returns a tensor with the <code>diagonal</code> and
+ everything else padded with zeros. The diagonal is computed as follows:</p><p>Assume <code>diagonal</code> has dimensions [D1,..., Dk], then the output is a tensor of
+ rank 2k with dimensions [D1,..., Dk, D1,..., Dk] where:</p><p>`output[i1,..., ik, i1,..., ik] = diagonal[i1, ..., ik]` and 0 everywhere else.</p><p>For example:</p><p>```prettyprint
+ # <code>diagonal</code> is [1, 2, 3, 4]
+ tf.diag(diagonal) ==&gt; [[1, 0, 0, 0]
+                        [0, 2, 0, 0]
+                        [0, 0, 3, 0]
+                        [0, 0, 0, 4]]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:matrixDiagPart" class="def">matrixDiagPart</a> <a href="src/TensorFlow-GenOps-Core.html#matrixDiagPart" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Rank <code>k</code> tensor where `k &gt;= 2` and the last two dimensions are equal.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>diagonal</strong>: The extracted diagonal(s) having shape
+ `diagonal.shape = input.shape[:-1]`.</p></td></tr></table></div><div class="doc"><p>Returns the batched diagonal part of a batched tensor.</p><p>This operation returns a tensor with the <code>diagonal</code> part
+ of the batched <code>input</code>. The <code>diagonal</code> part is computed as follows:</p><p>Assume <code>input</code> has <code>k</code> dimensions `[I, J, K, ..., N, N]`, then the output is a
+ tensor of rank `k - 1` with dimensions `[I, J, K, ..., N]` where:</p><p>`diagonal[i, j, k, ..., n] = input[i, j, k, ..., n, n]`.</p><p>The input must be at least a matrix.</p><p>For example:</p><p>```prettyprint
+ # <code>input</code> is [[[1, 0, 0, 0]
+                [0, 2, 0, 0]
+                [0, 0, 3, 0]
+                [0, 0, 0, 4]],
+               [[5, 0, 0, 0]
+                [0, 6, 0, 0]
+                [0, 0, 7, 0]
+                [0, 0, 0, 8]]]</p><p>and input.shape = (2, 4, 4)</p><p>tf.matrix_diag_part(input) ==&gt; [[1, 2, 3, 4], [5, 6, 7, 8]]</p><p>which has shape (2, 4)
+ ```</p></div></div><div class="top"><p class="src"><a name="v:queueSize" class="def">queueSize</a> <a href="src/TensorFlow-GenOps-Core.html#queueSize" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a queue.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>size</strong>: The number of elements in the given queue.</p></td></tr></table></div><div class="doc"><p>Computes the number of elements in the given queue.</p></div></div><div class="top"><p class="src"><a name="v:decodePng" class="def">decodePng</a> <a href="src/TensorFlow-GenOps-Core.html#decodePng" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>]` dtype)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>contents</strong>: 0-D.  The PNG-encoded image.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</td><td class="doc"><p><strong>image</strong>: 3-D with shape `[height, width, channels]`.</p></td></tr></table></div><div class="doc"><p>Decode a PNG-encoded image to a uint8 or uint16 tensor.</p><p>The attr <code>channels</code> indicates the desired number of color channels for the
+ decoded image.</p><p>Accepted values are:</p><ul><li>0: Use the number of channels in the PNG-encoded image.</li><li>1: output a grayscale image.</li><li>3: output an RGB image.</li><li>4: output an RGBA image.</li></ul><p>If needed, the PNG-encoded image is transformed to match the requested number
+ of color channels.</p></div></div><div class="top"><p class="src"><a name="v:ceil" class="def">ceil</a> <a href="src/TensorFlow-GenOps-Core.html#ceil" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Returns element-wise smallest integer in not less than x.</p></div></div><div class="top"><p class="src"><a name="v:priorityQueue" class="def">priorityQueue</a> <a href="src/TensorFlow-GenOps-Core.html#priorityQueue" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to the queue.</p></td></tr></table></div><div class="doc"><p>A queue that produces elements sorted by the first component value.</p><p>Note that the PriorityQueue requires the first component of any element
+ to be a scalar int64, in addition to the other elements declared by
+ component_types.  Therefore calls to Enqueue and EnqueueMany (resp. Dequeue
+ and DequeueMany) on a PriorityQueue will all require (resp. output) one extra
+ entry in their input (resp. output) lists.</p></div></div><div class="top"><p class="src"><a name="v:placeholderWithDefault" class="def">placeholderWithDefault</a> <a href="src/TensorFlow-GenOps-Core.html#placeholderWithDefault" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 dtype</td><td class="doc"><p><strong>input</strong>: The default value to produce when <code>output</code> is not fed.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</td><td class="doc"><p><strong>output</strong>: A placeholder tensor that defaults to <code>input</code> if it is not fed.</p></td></tr></table></div><div class="doc"><p>A placeholder op that passes though <code>input</code> when its output is not fed.</p></div></div><div class="top"><p class="src"><a name="v:cropAndResizeGradImage" class="def">cropAndResizeGradImage</a> <a href="src/TensorFlow-GenOps-Core.html#cropAndResizeGradImage" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>grads</strong>: A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>boxes</strong>: A 2-D tensor of shape `[num_boxes, 4]`. The <code>i</code>-th row of the tensor
+ specifies the coordinates of a box in the `box_ind[i]` image and is specified
+ in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of
+ <code>y</code> is mapped to the image coordinate at `y * (image_height - 1)`, so as the
+ `[0, 1]` interval of normalized image height is mapped to
+ `[0, image_height - 1] in image height coordinates. We do allow y1 &gt; y2, in
+ which case the sampled crop is an up-down flipped version of the original
+ image. The width dimension is treated similarly. Normalized coordinates
+ outside the `[0, 1]` range are allowed, in which case we use
+ <code>extrapolation_value</code> to extrapolate the input image values.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>box_ind</strong>: A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.
+ The value of `box_ind[i]` specifies the image that the <code>i</code>-th box refers to.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>image_size</strong>: A 1-D tensor with value `[batch, image_height, image_width, depth]`
+ containing the original image size. Both <code>image_height</code> and <code>image_width</code> need
+ to be positive.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: A 4-D tensor of shape `[batch, image_height, image_width, depth]`.</p></td></tr></table></div><div class="doc"><p>Computes the gradient of the crop_and_resize op wrt the input image tensor.</p></div></div><div class="top"><p class="src"><a name="v:readerReset" class="def">readerReset</a> <a href="src/TensorFlow-GenOps-Core.html#readerReset" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>reader_handle</strong>: Handle to a Reader.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Restore a Reader to its initial clean state.</p></div></div><div class="top"><p class="src"><a name="v:extractImagePatches" class="def">extractImagePatches</a> <a href="src/TensorFlow-GenOps-Core.html#extractImagePatches" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>images</strong>: 4-D Tensor with shape `[batch, in_rows, in_cols, depth]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>patches</strong>: 4-D Tensor with shape `[batch, out_rows, out_cols, ksize_rows *
+ ksize_cols * depth]` containing image patches with size
+ `ksize_rows x ksize_cols x depth` vectorized in the &quot;depth&quot; dimension.</p></td></tr></table></div><div class="doc"><p>Extract <code>patches</code> from <code>images</code> and put them in the &quot;depth&quot; output dimension.</p></div></div><div class="top"><p class="src"><a name="v:batchMatrixSetDiag" class="def">batchMatrixSetDiag</a> <a href="src/TensorFlow-GenOps-Core.html#batchMatrixSetDiag" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>diagonal</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:stackClose" class="def">stackClose</a> <a href="src/TensorFlow-GenOps-Core.html#stackClose" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a stack.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Delete the stack from its resource container.</p></div></div><div class="top"><p class="src"><a name="v:quantizeAndDequantize" class="def">quantizeAndDequantize</a> <a href="src/TensorFlow-GenOps-Core.html#quantizeAndDequantize" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Tensor to quantize and then dequantize.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Quantizes then dequantizes a tensor.</p><p>This op simulates the precision loss from the quantized forward pass by:
+ 1. Quantizing the tensor to fixed point numbers, which should match the target
+    quantization method when it is used in inference.
+ 2. Dequantizing it back to floating point numbers for the following ops, most
+    likely matmul.</p><p>There are different ways to quantize. This version does not use the full range
+ of the output type, choosing to elide the lowest possible value for symmetry
+ (e.g., output range is -127 to 127, not -128 to 127 for signed 8 bit
+ quantization), so that 0.0 maps to 0.</p><p>To perform this op, we first find the range of values in our tensor. The range
+ we use is always centered on 0, so we find m such that</p><ol><li>m = max(abs(input_min), abs(input_max)) if range_given is true,</li><li>m = max(max(abs(min_elem(input)), abs(max_elem(input))) otherwise.</li></ol><p>Our input tensor range is then [-m, m].</p><p>Next, we choose our fixed-point quantization buckets, [min_fixed, max_fixed].
+ If signed_input is true, this is</p><dl><dt>min_fixed, max_fixed </dt><dd>=</dd><dt>-(1 &lt;&lt; (num_bits - 1) - 1), (1 &lt;&lt; (num_bits - 1)) - 1</dt><dd>.</dd></dl><p>Otherwise, if signed_input is false, the fixed-point range is</p><dl><dt>min_fixed, max_fixed</dt><dd>= [0, (1 &lt;&lt; num_bits) - 1].</dd></dl><p>From this we compute our scaling factor, s:</p><p>s = (max_fixed - min_fixed) / (2 * m).</p><p>Now we can quantize and dequantize the elements of our tensor.  An element e
+ is transformed into e':</p><p>e' = (e * s).round_to_nearest() / s.</p><p>Note that we have a different number of buckets in the signed vs. unsigned
+ cases.  For example, if num_bits == 8, we get 254 buckets in the signed case
+ vs. 255 in the unsigned case.</p><p>For example, suppose num_bits = 8 and m = 1.  Then</p><dl><dt>min_fixed, max_fixed</dt><dd>= [-127, 127], and
+   s = (127 + 127) / 2 = 127.</dd></dl><p>Given the vector {-1, -0.5, 0, 0.3}, this is quantized to
+ {-127, -63, 0, 38}, and dequantized to {-1, -63.0<em>127, 0, 38.0</em>127}.</p></div></div><div class="top"><p class="src"><a name="v:isNan" class="def">isNan</a> <a href="src/TensorFlow-GenOps-Core.html#isNan" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Returns which elements of x are NaN.</p></div></div><div class="top"><p class="src"><a name="v:where-39-" class="def">where'</a> <a href="src/TensorFlow-GenOps-Core.html#where%27" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>index</strong></p></td></tr></table></div><div class="doc"><p>Returns locations of true values in a boolean tensor.</p><p>This operation returns the coordinates of true elements in <code>input</code>. The
+ coordinates are returned in a 2-D tensor where the first dimension (rows)
+ represents the number of true elements, and the second dimension (columns)
+ represents the coordinates of the true elements. Keep in mind, the shape of
+ the output tensor can vary depending on how many true values there are in
+ <code>input</code>. Indices are output in row-major order.</p><p>For example:</p><p>```prettyprint
+ # <code>input</code> tensor is [[True, False]
+ #                    [True, False]]
+ # <code>input</code> has two true values, so output has two coordinates.
+ # <code>input</code> has rank of 2, so coordinates have two indices.
+ where(input) ==&gt; [[0, 0],
+                   [1, 0]]</p><p># <code>input</code> tensor is [[[True, False]
+ #                     [True, False]]
+ #                    [[False, True]
+ #                     [False, True]]
+ #                    [[False, False]
+ #                     [False, True]]]
+ # <code>input</code> has 5 true values, so output has 5 coordinates.
+ # <code>input</code> has rank of 3, so coordinates have three indices.
+ where(input) ==&gt; [[0, 0, 0],
+                   [0, 1, 0],
+                   [1, 0, 1],
+                   [1, 1, 1],
+                   [2, 1, 1]]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:listDiff" class="def">listDiff</a> <a href="src/TensorFlow-GenOps-Core.html#listDiff" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_idx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` out_idx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong>: 1-D. Values to keep.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong>: 1-D. Values to remove.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_idx)</td><td class="doc"><p>(<strong>out</strong>, <strong>idx</strong>)</p><ul><li><strong>out</strong>: 1-D. Values present in <code>x</code> but not in <code>y</code>.</li><li><strong>idx</strong>: 1-D. Positions of <code>x</code> values preserved in <code>out</code>.</li></ul></td></tr></table></div><div class="doc"><p>Computes the difference between two lists of numbers or strings.</p><p>Given a list <code>x</code> and a list <code>y</code>, this operation returns a list <code>out</code> that
+ represents all values that are in <code>x</code> but not in <code>y</code>. The returned list <code>out</code>
+ is sorted in the same order that the numbers appear in <code>x</code> (duplicates are
+ preserved). This operation also returns a list <code>idx</code> that represents the
+ position of each <code>out</code> element in <code>x</code>. In other words:</p><p>`out[i] = x[idx[i]] for i in [0, 1, ..., len(out) - 1]`</p><p>For example, given this input:</p><p>```prettyprint
+ x = [1, 2, 3, 4, 5, 6]
+ y = [1, 3, 5]
+ ```</p><p>This operation would return:</p><p>```prettyprint
+ out ==&gt; [2, 4, 6]
+ idx ==&gt; [1, 3, 5]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:stridedSlice" class="def">stridedSlice</a> <a href="src/TensorFlow-GenOps-Core.html#stridedSlice" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> index, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` index, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 index</td><td class="doc"><p><strong>begin</strong>: `begin[i]` specifies the offset into the <code>i</code>th dimension of
+ <code>input</code> to slice from.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 index</td><td class="doc"><p><strong>end</strong>: `end[i]` specifies the first offset into the <code>i</code>th dimension of
+ <code>input</code> that will not be extracted. Out or range values are
+ clamped to `[0,dim[i]) if slice[i] &gt; 0` or `[-1,dim[i]-1]`
+ `if slice[i] &lt; 0`</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 index</td><td class="doc"><p><strong>strides</strong>: `strides[i]` specifies the increment in the <code>i</code>th dimension
+ after extracting a given element. Negative indices will reverse
+ the original order. Out or range values are
+ clamped to `[0,dim[i]) if slice[i]&gt;0` or `[-1,dim[i]-1] if slice[i] &lt; 0`</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Return a strided slice from <code>input</code>.</p><p>The output tensor is a tensor with dimensions implied by <code>begin</code>,
+ <code>end</code>, and <code>strides</code>, whose values are extracted from <code>begin</code>.</p><p>Specifically, the result tensor at index `(i[0], i[1], ..., i[n-1])`
+ will obtain the value `input[begin[0] + i[0] * stride[0], ..., `
+                             `begin[n-1] + i[n-1] * stride[n-1])]`.</p><ul><li>Requirements*:
+   `0 != strides[i] for i in [0, n)`</li></ul></div></div><div class="top"><p class="src"><a name="v:randomShuffleQueue" class="def">randomShuffleQueue</a> <a href="src/TensorFlow-GenOps-Core.html#randomShuffleQueue" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to the queue.</p></td></tr></table></div><div class="doc"><p>A queue that randomizes the order of elements.</p></div></div><div class="top"><p class="src"><a name="v:tileGrad" class="def">tileGrad</a> <a href="src/TensorFlow-GenOps-Core.html#tileGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>multiples</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Returns the gradient of <code>Tile</code>.</p><p>Since <code>Tile</code> takes an input and repeats the input <code>multiples</code> times
+ along each dimension, <code>TileGrad</code> takes in <code>multiples</code> and aggregates
+ each repeated tile of <code>input</code> into <code>output</code>.</p></div></div><div class="top"><p class="src"><a name="v:stridedSliceAssign" class="def">stridedSliceAssign</a> <a href="src/TensorFlow-GenOps-Core.html#stridedSliceAssign" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> index, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` index, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>ref</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 index</td><td class="doc"><p><strong>begin</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 index</td><td class="doc"><p><strong>end</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 index</td><td class="doc"><p><strong>strides</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>value</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output_ref</strong></p></td></tr></table></div><div class="doc"><p>Assign <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code> to the sliced l-value reference of <code>ref</code>.</p><p>The values of <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code> are assigned to the positions in the variable
+ <code>ref</code> that are selected by the slice parameters. The slice parameters
+ `begin, <code>end</code>, <code>strides</code>, etc. work exactly as in <code>StridedSlice</code>.</p><p>NOTE this op currently does not support broadcasting and so <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>'s
+ shape must be exactly the shape produced by the slice of <code>ref</code>.</p></div></div><div class="top"><p class="src"><a name="v:reshape" class="def">reshape</a> <a href="src/TensorFlow-GenOps-Core.html#reshape" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tshape, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tshape)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>tensor</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tshape</td><td class="doc"><p><strong>shape</strong>: Defines the shape of the output tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Reshapes a tensor.</p><p>Given <code>tensor</code>, this operation returns a tensor that has the same values
+ as <code>tensor</code> with shape <code><a href="TensorFlow-GenOps-Core.html#v:shape">shape</a></code>.</p><p>If one component of <code><a href="TensorFlow-GenOps-Core.html#v:shape">shape</a></code> is the special value -1, the size of that dimension
+ is computed so that the total size remains constant.  In particular, a <code><a href="TensorFlow-GenOps-Core.html#v:shape">shape</a></code>
+ of `[-1]` flattens into 1-D.  At most one component of <code><a href="TensorFlow-GenOps-Core.html#v:shape">shape</a></code> can be -1.</p><p>If <code><a href="TensorFlow-GenOps-Core.html#v:shape">shape</a></code> is 1-D or higher, then the operation returns a tensor with shape
+ <code><a href="TensorFlow-GenOps-Core.html#v:shape">shape</a></code> filled with the values of <code>tensor</code>. In this case, the number of elements
+ implied by <code><a href="TensorFlow-GenOps-Core.html#v:shape">shape</a></code> must be the same as the number of elements in <code>tensor</code>.</p><p>For example:</p><p>```prettyprint
+ # tensor <code>t</code> is [1, 2, 3, 4, 5, 6, 7, 8, 9]
+ # tensor <code>t</code> has shape [9]
+ reshape(t, [3, 3]) ==&gt; [[1, 2, 3],
+                         [4, 5, 6],
+                         [7, 8, 9]]</p><p># tensor <code>t</code> is [[[1, 1], [2, 2]],
+ #                [[3, 3], [4, 4]]]
+ # tensor <code>t</code> has shape [2, 2, 2]
+ reshape(t, [2, 4]) ==&gt; [[1, 1, 2, 2],
+                         [3, 3, 4, 4]]</p><p># tensor <code>t</code> is [[[1, 1, 1],
+ #                 [2, 2, 2]],
+ #                [[3, 3, 3],
+ #                 [4, 4, 4]],
+ #                [[5, 5, 5],
+ #                 [6, 6, 6]]]
+ # tensor <code>t</code> has shape [3, 2, 3]
+ # pass '[-1]' to flatten <code>t</code>
+ reshape(t, [-1]) ==&gt; [1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6, 6, 6]</p><p># -1 can also be used to infer the shape</p><p># -1 is inferred to be 9:
+ reshape(t, [2, -1]) ==&gt; [[1, 1, 1, 2, 2, 2, 3, 3, 3],
+                          [4, 4, 4, 5, 5, 5, 6, 6, 6]]
+ # -1 is inferred to be 2:
+ reshape(t, [-1, 9]) ==&gt; [[1, 1, 1, 2, 2, 2, 3, 3, 3],
+                          [4, 4, 4, 5, 5, 5, 6, 6, 6]]
+ # -1 is inferred to be 3:
+ reshape(t, [ 2, -1, 3]) ==&gt; [[[1, 1, 1],
+                               [2, 2, 2],
+                               [3, 3, 3]],
+                              [[4, 4, 4],
+                               [5, 5, 5],
+                               [6, 6, 6]]]</p><p># tensor <code>t</code> is [7]
+ # shape `[]` reshapes to a scalar
+ reshape(t, []) ==&gt; 7
+ ```</p></div></div><div class="top"><p class="src"><a name="v:fIFOQueue" class="def">fIFOQueue</a> <a href="src/TensorFlow-GenOps-Core.html#fIFOQueue" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to the queue.</p></td></tr></table></div><div class="doc"><p>A queue that produces elements in first-in first-out order.</p></div></div><div class="top"><p class="src"><a name="v:learnedUnigramCandidateSampler" class="def">learnedUnigramCandidateSampler</a> <a href="src/TensorFlow-GenOps-Core.html#learnedUnigramCandidateSampler" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_sampled</strong>: Number of candidates to randomly sample per batch.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_true</strong>: Number of true labels per context.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>range_max</strong>: The sampler will sample integers from the interval [0, range_max).</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>unique</strong>: If unique is true, we sample with rejection, so that all sampled
+ candidates in a batch are unique. This requires some approximation to
+ estimate the post-rejection sampling probabilities.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>true_classes</strong>: A batch_size * num_true matrix, in which each row contains the
+ IDs of the num_true target_classes in the corresponding original label.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p>(<strong>sampled_candidates</strong>, <strong>true_expected_count</strong>, <strong>sampled_expected_count</strong>)</p><ul><li><strong>sampled_candidates</strong>: A vector of length num_sampled, in which each element is
+ the ID of a sampled candidate.</li><li><strong>true_expected_count</strong>: A batch_size * num_true matrix, representing
+ the number of times each candidate is expected to occur in a batch
+ of sampled candidates. If unique=true, then this is a probability.</li><li><strong>sampled_expected_count</strong>: A vector of length num_sampled, for each sampled
+ candidate representing the number of times the candidate is expected
+ to occur in a batch of sampled candidates.  If unique=true, then this is a
+ probability.</li></ul></td></tr></table></div><div class="doc"><p>Generates labels for candidate sampling with a learned unigram distribution.</p><p>See explanations of candidate sampling and the data formats at
+ go/candidate-sampling.</p><p>For each batch, this op picks a single set of sampled candidate labels.</p><p>The advantages of sampling candidates per-batch are simplicity and the
+ possibility of efficient dense matrix multiplication. The disadvantage is that
+ the sampled candidates must be chosen independently of the context and of the
+ true labels.</p></div></div><div class="top"><p class="src"><a name="v:fractionalAvgPool" class="def">fractionalAvgPool</a> <a href="src/TensorFlow-GenOps-Core.html#fractionalAvgPool" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>value</strong>: 4-D with shape `[batch, height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</td><td class="doc"><p>(<strong>output</strong>, <strong>row_pooling_sequence</strong>, <strong>col_pooling_sequence</strong>)</p><ul><li><strong>output</strong>: output tensor after fractional avg pooling.</li><li><strong>row_pooling_sequence</strong>: row pooling sequence, needed to calculate gradient.</li><li><strong>col_pooling_sequence</strong>: column pooling sequence, needed to calculate gradient.</li></ul></td></tr></table></div><div class="doc"><p>Performs fractional average pooling on the input.</p><p>Fractional average pooling is similar to Fractional max pooling in the pooling
+ region generation step. The only difference is that after pooling regions are
+ generated, a mean operation is performed instead of a max operation in each
+ pooling region.</p></div></div><div class="top"><p class="src"><a name="v:randomCrop" class="def">randomCrop</a> <a href="src/TensorFlow-GenOps-Core.html#randomCrop" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>image</strong>: 3-D of shape `[height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>size</strong>: 1-D of length 2 containing: <code>crop_height</code>, <code>crop_width</code>..</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 3-D of shape `[crop_height, crop_width, channels].`</p></td></tr></table></div><div class="doc"><p>Randomly crop <code>image</code>.</p><p><code><a href="TensorFlow-GenOps-Core.html#v:size">size</a></code> is a 1-D int64 tensor with 2 elements representing the crop height and
+ width.  The values must be non negative.</p><p>This Op picks a random location in <code>image</code> and crops a <code>height</code> by <code>width</code>
+ rectangle from that location.  The random location is picked so the cropped
+ area will fit inside the original image.</p></div></div><div class="top"><p class="src"><a name="v:_HostCast" class="def">_HostCast</a> <a href="src/TensorFlow-GenOps-Core.html#_HostCast" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dstT, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> srcT)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 srcT</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dstT</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Cast x of type SrcT to y of DstT.</p><p>_HostCast requires its input and produces its output in host memory.</p></div></div><div class="top"><p class="src"><a name="v:queueClose" class="def">queueClose</a> <a href="src/TensorFlow-GenOps-Core.html#queueClose" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a queue.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Closes the given queue.</p><p>This operation signals that no more elements will be enqueued in the
+ given queue. Subsequent Enqueue(Many) operations will fail.
+ Subsequent Dequeue(Many) operations will continue to succeed if
+ sufficient elements remain in the queue. Subsequent Dequeue(Many)
+ operations that would block will fail immediately.</p></div></div><div class="top"><p class="src"><a name="v:slice" class="def">slice</a> <a href="src/TensorFlow-GenOps-Core.html#slice" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> index, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` index, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 index</td><td class="doc"><p><strong>begin</strong>: begin[i] specifies the offset into the <code>i</code>th dimension of
+ <code>input</code> to slice from.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 index</td><td class="doc"><p><strong>size</strong>: size[i] specifies the number of elements of the <code>i</code>th dimension
+ of <code>input</code> to slice. If size[i] is -1, all remaining elements in dimension
+ i are included in the slice (i.e. this is equivalent to setting
+ size[i] = input.dim_size(i) - begin[i]).</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Return a slice from <code>input</code>.</p><p>The output tensor is a tensor with dimensions described by <code><a href="TensorFlow-GenOps-Core.html#v:size">size</a></code>
+ whose values are extracted from <code>input</code> starting at the offsets in
+ <code>begin</code>.</p><ul><li>Requirements*:
+   0 &lt;= begin[i] &lt;= begin[i] + size[i] &lt;= Di  for i in [0, n)</li></ul></div></div><div class="top"><p class="src"><a name="v:stridedSliceGrad" class="def">stridedSliceGrad</a> <a href="src/TensorFlow-GenOps-Core.html#stridedSliceGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> index, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` index, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 index</td><td class="doc"><p><strong>shape</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 index</td><td class="doc"><p><strong>begin</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 index</td><td class="doc"><p><strong>end</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 index</td><td class="doc"><p><strong>strides</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>dy</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Returns the gradient of <code>StridedSlice</code>.</p><p>Since <code>StridedSlice</code> cuts out pieces of its <code>input</code> which is size
+ <code><a href="TensorFlow-GenOps-Core.html#v:shape">shape</a></code>, its gradient will have the same shape (which is passed here
+ as <code><a href="TensorFlow-GenOps-Core.html#v:shape">shape</a></code>). The gradient will be zero in any element that the slice
+ does not select.</p><p>Arguments are the same as StridedSliceGrad with the exception that
+ <code>dy</code> is the input gradient to be propagated and <code><a href="TensorFlow-GenOps-Core.html#v:shape">shape</a></code> is the
+ shape of <code>StridedSlice</code>'s <code>input</code>.</p></div></div><div class="top"><p class="src"><a name="v:sparseTensorDenseAdd" class="def">sparseTensorDenseAdd</a> <a href="src/TensorFlow-GenOps-Core.html#sparseTensorDenseAdd" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tindices</td><td class="doc"><p><strong>a_indices</strong>: 2-D.  The <code>indices</code> of the <code>SparseTensor</code>, with shape `[nnz, ndims]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>a_values</strong>: 1-D.  The <code>values</code> of the <code>SparseTensor</code>, with shape `[nnz]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tindices</td><td class="doc"><p><strong>a_shape</strong>: 1-D.  The <code><a href="TensorFlow-GenOps-Core.html#v:shape">shape</a></code> of the <code>SparseTensor</code>, with shape `[ndims]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>b</strong>: <code>ndims</code>-D Tensor.  With shape <code>a_shape</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Adds up a <code>SparseTensor</code> and a dense <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code>, producing a dense <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code>.</p><p>This Op does not require <code>a_indices</code> be sorted in standard lexicographic order.</p></div></div><div class="top"><p class="src"><a name="v:size" class="def">size</a> <a href="src/TensorFlow-GenOps-Core.html#size" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_type, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` out_type)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_type</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Returns the size of a tensor.</p><p>This operation returns an integer representing the number of elements in
+ <code>input</code>.</p><p>For example:</p><p>```prettyprint
+ # <code>t</code> is [[[1, 1,, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]]
+ size(t) ==&gt; 12
+ ```</p></div></div><div class="top"><p class="src"><a name="v:barrier" class="def">barrier</a> <a href="src/TensorFlow-GenOps-Core.html#barrier" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to the barrier.</p></td></tr></table></div><div class="doc"><p>Defines a barrier that persists across different graph executions.</p><p>A barrier represents a key-value map, where each key is a string, and
+ each value is a tuple of tensors.</p><p>At runtime, the barrier contains <code>complete</code> and <code>incomplete</code>
+ elements. A complete element has defined tensors for all components of
+ its value tuple, and may be accessed using BarrierTakeMany. An
+ incomplete element has some undefined components in its value tuple,
+ and may be updated using BarrierInsertMany.</p></div></div><div class="top"><p class="src"><a name="v:lgamma" class="def">lgamma</a> <a href="src/TensorFlow-GenOps-Core.html#lgamma" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes the log of the absolute value of `Gamma(x)` element-wise.</p></div></div><div class="top"><p class="src"><a name="v:decodeJpeg" class="def">decodeJpeg</a> <a href="src/TensorFlow-GenOps-Core.html#decodeJpeg" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>contents</strong>: 0-D.  The JPEG-encoded image.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a></td><td class="doc"><p><strong>image</strong>: 3-D with shape `[height, width, channels]`..</p></td></tr></table></div><div class="doc"><p>Decode a JPEG-encoded image to a uint8 tensor.</p><p>The attr <code>channels</code> indicates the desired number of color channels for the
+ decoded image.</p><p>Accepted values are:</p><ul><li>0: Use the number of channels in the JPEG-encoded image.</li><li>1: output a grayscale image.</li><li>3: output an RGB image.</li></ul><p>If needed, the JPEG-encoded image is transformed to match the requested number
+ of color channels.</p><p>The attr <code>ratio</code> allows downscaling the image by an integer factor during
+ decoding.  Allowed values are: 1, 2, 4, and 8.  This is much faster than
+ downscaling the image later.</p></div></div><div class="top"><p class="src"><a name="v:shapeN" class="def">shapeN</a> <a href="src/TensorFlow-GenOps-Core.html#shapeN" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_type, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` out_type)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t]</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_type]</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Returns shape of tensors.</p><p>This operation returns N 1-D integer tensors representing shape of `input[i]s`.</p></div></div><div class="top"><p class="src"><a name="v:uniformCandidateSampler" class="def">uniformCandidateSampler</a> <a href="src/TensorFlow-GenOps-Core.html#uniformCandidateSampler" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_sampled</strong>: Number of candidates to randomly sample per batch.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_true</strong>: Number of true labels per context.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>range_max</strong>: The sampler will sample integers from the interval [0, range_max).</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>unique</strong>: If unique is true, we sample with rejection, so that all sampled
+ candidates in a batch are unique. This requires some approximation to
+ estimate the post-rejection sampling probabilities.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>true_classes</strong>: A batch_size * num_true matrix, in which each row contains the
+ IDs of the num_true target_classes in the corresponding original label.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p>(<strong>sampled_candidates</strong>, <strong>true_expected_count</strong>, <strong>sampled_expected_count</strong>)</p><ul><li><strong>sampled_candidates</strong>: A vector of length num_sampled, in which each element is
+ the ID of a sampled candidate.</li><li><strong>true_expected_count</strong>: A batch_size * num_true matrix, representing
+ the number of times each candidate is expected to occur in a batch
+ of sampled candidates. If unique=true, then this is a probability.</li><li><strong>sampled_expected_count</strong>: A vector of length num_sampled, for each sampled
+ candidate representing the number of times the candidate is expected
+ to occur in a batch of sampled candidates.  If unique=true, then this is a
+ probability.</li></ul></td></tr></table></div><div class="doc"><p>Generates labels for candidate sampling with a uniform distribution.</p><p>See explanations of candidate sampling and the data formats at
+ go/candidate-sampling.</p><p>For each batch, this op picks a single set of sampled candidate labels.</p><p>The advantages of sampling candidates per-batch are simplicity and the
+ possibility of efficient dense matrix multiplication. The disadvantage is that
+ the sampled candidates must be chosen independently of the context and of the
+ true labels.</p></div></div><div class="top"><p class="src"><a name="v:unique" class="def">unique</a> <a href="src/TensorFlow-GenOps-Core.html#unique" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_idx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` out_idx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong>: 1-D.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_idx)</td><td class="doc"><p>(<strong>y</strong>, <strong>idx</strong>)</p><ul><li><strong>y</strong>: 1-D.</li><li><strong>idx</strong>: 1-D.</li></ul></td></tr></table></div><div class="doc"><p>Finds unique elements in a 1-D tensor.</p><p>This operation returns a tensor <code>y</code> containing all of the unique elements of <code>x</code>
+ sorted in the same order that they occur in <code>x</code>. This operation also returns a
+ tensor <code>idx</code> the same size as <code>x</code> that contains the index of each value of <code>x</code>
+ in the unique output <code>y</code>. In other words:</p><p>`y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`</p><p>For example:</p><p>```prettyprint
+ # tensor <code>x</code> is [1, 1, 2, 4, 4, 4, 7, 8, 8]
+ y, idx = unique(x)
+ y ==&gt; [1, 2, 4, 7, 8]
+ idx ==&gt; [0, 0, 1, 2, 2, 2, 3, 4, 4]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:drawBoundingBoxes" class="def">drawBoundingBoxes</a> <a href="src/TensorFlow-GenOps-Core.html#drawBoundingBoxes" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>images</strong>: 4-D with shape `[batch, height, width, depth]`. A batch of images.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>boxes</strong>: 3-D with shape `[batch, num_bounding_boxes, 4]` containing bounding
+ boxes.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 4-D with the same shape as <code>images</code>. The batch of input images with
+ bounding boxes drawn on the images.</p></td></tr></table></div><div class="doc"><p>Draw bounding boxes on a batch of images.</p><p>Outputs a copy of <code>images</code> but draws on top of the pixels zero or more bounding
+ boxes specified by the locations in <code>boxes</code>. The coordinates of the each
+ bounding box in <code>boxes</code> are encoded as `[y_min, x_min, y_max, x_max]`. The
+ bounding box coordinates are floats in `[0.0, 1.0]` relative to the width and
+ height of the underlying image.</p><p>For example, if an image is 100 x 200 pixels and the bounding box is
+ `[0.1, 0.2, 0.5, 0.9]`, the bottom-left and upper-right coordinates of the
+ bounding box will be `(10, 40)` to `(50, 180)`.</p><p>Parts of the bounding box may fall outside the image.</p></div></div><div class="top"><p class="src"><a name="v:tensorArraySplit" class="def">tensorArraySplit</a> <a href="src/TensorFlow-GenOps-Core.html#tensorArraySplit" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a TensorArray.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>value</strong>: The concatenated tensor to write to the TensorArray.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>lengths</strong>: The vector of lengths, how to split the rows of value into the
+ TensorArray.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>flow_in</strong>: A float scalar that enforces proper chaining of operations.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>flow_out</strong>: A float scalar that enforces proper chaining of operations.</p></td></tr></table></div><div class="doc"><p>Split the data from the input value into TensorArray elements.</p><p>Assuming that <code>lengths</code> takes on values</p><p>```(n0, n1, ..., n(T-1))```</p><p>and that <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code> has shape</p><p>```(n0 + n1 + ... + n(T-1) x d0 x d1 x ...)```,</p><p>this splits values into a TensorArray with T tensors.</p><p>TensorArray index t will be the subtensor of values with starting position</p><p>```(n0 + n1 + ... + n(t-1), 0, 0, ...)```</p><p>and having size</p><p>```nt x d0 x d1 x ...```</p></div></div><div class="top"><p class="src"><a name="v:split" class="def">split</a> <a href="src/TensorFlow-GenOps-Core.html#split" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_split</strong>: The number of ways to split.  Must evenly divide
+ `value.shape[split_dim]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>split_dim</strong>: 0-D.  The dimension along which to split.  Must be in the range
+ `[0, rank(value))`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>value</strong>: The tensor to split.</p></td></tr><tr><td class="src">-&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t]</td><td class="doc"><p><strong>output</strong>: They are identically shaped tensors, whose shape matches that of <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>
+ except along <code>split_dim</code>, where their sizes are
+ `values.shape[split_dim] / num_split`.</p></td></tr></table></div><div class="doc"><p>Splits a tensor into <code>num_split</code> tensors along one dimension.</p></div></div><div class="top"><p class="src"><a name="v:segmentMax" class="def">segmentMax</a> <a href="src/TensorFlow-GenOps-Core.html#segmentMax" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices</td><td class="doc"><p><strong>segment_ids</strong>: A 1-D tensor whose rank is equal to the rank of `data`'s
+ first dimension.  Values should be sorted and can be repeated.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Has same shape as data, except for dimension 0 which
+ has size <code>k</code>, the number of segments.</p></td></tr></table></div><div class="doc"><p>Computes the maximum along segments of a tensor.</p><p>Read <a href="../../api_docs/python/math_ops.md#segmentation">the section on Segmentation</a>
+ for an explanation of segments.</p><p>Computes a tensor such that
+ \(output_i = max_j(data_j)\) where <code><a href="../base-4.8.2.0/Data-Ord.html#v:max">max</a></code> is over <code>j</code> such
+ that `segment_ids[j] == i`.</p><p><a href="div">style=&quot;width:70%; margin:auto; margin-bottom:10px; margin-top:20px;&quot;</a>
+ <a href="img">style=&quot;width:100%&quot; src=&quot;../../images/SegmentMax.png&quot; alt</a>
+ <a href="/div">/div</a></p></div></div><div class="top"><p class="src"><a name="v:abort" class="def">abort</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a> <a href="src/TensorFlow-GenOps-Core.html#abort" class="link">Source</a></p><div class="doc"><p>Raise a exception to abort the process when called.</p><p>Returns nothing but an exception.</p></div></div><div class="top"><p class="src"><a name="v:sparseReorder" class="def">sparseReorder</a> <a href="src/TensorFlow-GenOps-Core.html#sparseReorder" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>input_indices</strong>: 2-D.  `N x R` matrix with the indices of non-empty values in a
+ SparseTensor, possibly not in canonical ordering.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>input_values</strong>: 1-D.  <code>N</code> non-empty values corresponding to <code>input_indices</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>input_shape</strong>: 1-D.  Shape of the input SparseTensor.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><td class="doc"><p>(<strong>output_indices</strong>, <strong>output_values</strong>)</p><ul><li><strong>output_indices</strong>: 2-D.  `N x R` matrix with the same indices as input_indices, but
+ in canonical row-major ordering.</li><li><strong>output_values</strong>: 1-D.  <code>N</code> non-empty values corresponding to <code>output_indices</code>.</li></ul></td></tr></table></div><div class="doc"><p>Reorders a SparseTensor into the canonical, row-major ordering.</p><p>Note that by convention, all sparse ops preserve the canonical ordering along
+ increasing dimension number. The only time ordering can be violated is during
+ manual manipulation of the indices and values vectors to add entries.</p><p>Reordering does not affect the shape of the SparseTensor.</p><p>If the tensor has rank <code>R</code> and <code>N</code> non-empty values, <code>input_indices</code> has
+ shape `[N, R]`, input_values has length <code>N</code>, and input_shape has length <code>R</code>.</p></div></div><div class="top"><p class="src"><a name="v:rsqrtGrad" class="def">rsqrtGrad</a> <a href="src/TensorFlow-GenOps-Core.html#rsqrtGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Computes the gradient for the rsqrt of <code>x</code> wrt its input.</p><p>Specifically, `grad = dy * -0.5 * y^3`, where `y = rsqrt(x)`, and <code>dy</code>
+ is the corresponding input gradient.</p></div></div><div class="top"><p class="src"><a name="v:reverseSequence" class="def">reverseSequence</a> <a href="src/TensorFlow-GenOps-Core.html#reverseSequence" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tlen, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tlen)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>seq_dim</strong>: The dimension which is partially reversed.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: The input to reverse.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tlen</td><td class="doc"><p><strong>seq_lengths</strong>: 1-D with length `input.dims(batch_dim)` and
+ `max(seq_lengths) &lt; input.dims(seq_dim)`</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The partially reversed input. It has the same shape as <code>input</code>.</p></td></tr></table></div><div class="doc"><p>Reverses variable length slices.</p><p>This op first slices <code>input</code> along the dimension <code>batch_dim</code>, and for each
+ slice <code>i</code>, reverses the first `seq_lengths[i]` elements along
+ the dimension <code>seq_dim</code>.</p><p>The elements of <code>seq_lengths</code> must obey `seq_lengths[i] &lt; input.dims[seq_dim]`,
+ and <code>seq_lengths</code> must be a vector of length `input.dims[batch_dim]`.</p><p>The output slice <code>i</code> along dimension <code>batch_dim</code> is then given by input
+ slice <code>i</code>, with the first `seq_lengths[i]` slices along dimension
+ <code>seq_dim</code> reversed.</p><p>For example:</p><p>```prettyprint
+ # Given this:
+ batch_dim = 0
+ seq_dim = 1
+ input.dims = (4, 8, ...)
+ seq_lengths = [7, 2, 3, 5]</p><p># then slices of input are reversed on seq_dim, but only up to seq_lengths:
+ output[0, 0:7, :, ...] = input[0, 7:0:-1, :, ...]
+ output[1, 0:2, :, ...] = input[1, 2:0:-1, :, ...]
+ output[2, 0:3, :, ...] = input[2, 3:0:-1, :, ...]
+ output[3, 0:5, :, ...] = input[3, 5:0:-1, :, ...]</p><p># while entries past seq_lens are copied through:
+ output[0, 7:, :, ...] = input[0, 7:, :, ...]
+ output[1, 2:, :, ...] = input[1, 2:, :, ...]
+ output[2, 3:, :, ...] = input[2, 3:, :, ...]
+ output[3, 2:, :, ...] = input[3, 2:, :, ...]
+ ```</p><p>In contrast, if:</p><p>```prettyprint
+ # Given this:
+ batch_dim = 2
+ seq_dim = 0
+ input.dims = (8, ?, 4, ...)
+ seq_lengths = [7, 2, 3, 5]</p><p># then slices of input are reversed on seq_dim, but only up to seq_lengths:
+ output[0:7, :, 0, :, ...] = input[7:0:-1, :, 0, :, ...]
+ output[0:2, :, 1, :, ...] = input[2:0:-1, :, 1, :, ...]
+ output[0:3, :, 2, :, ...] = input[3:0:-1, :, 2, :, ...]
+ output[0:5, :, 3, :, ...] = input[5:0:-1, :, 3, :, ...]</p><p># while entries past seq_lens are copied through:
+ output[7:, :, 0, :, ...] = input[7:, :, 0, :, ...]
+ output[2:, :, 1, :, ...] = input[2:, :, 1, :, ...]
+ output[3:, :, 2, :, ...] = input[3:, :, 2, :, ...]
+ output[2:, :, 3, :, ...] = input[2:, :, 3, :, ...]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:readerNumRecordsProduced" class="def">readerNumRecordsProduced</a> <a href="src/TensorFlow-GenOps-Core.html#readerNumRecordsProduced" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>reader_handle</strong>: Handle to a Reader.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>records_produced</strong></p></td></tr></table></div><div class="doc"><p>Returns the number of records this Reader has produced.</p><p>This is the same as the number of ReaderRead executions that have
+ succeeded.</p></div></div><div class="top"><p class="src"><a name="v:deserializeManySparse" class="def">deserializeManySparse</a> <a href="src/TensorFlow-GenOps-Core.html#deserializeManySparse" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>serialized_sparse</strong>: 2-D, The <code>N</code> serialized <code>SparseTensor</code> objects.
+ Must have 3 columns.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</td><td class="doc"><p>(<strong>sparse_indices</strong>, <strong>sparse_values</strong>, <strong>sparse_shape</strong>)</p><ul><li><strong>sparse_indices</strong></li><li><strong>sparse_values</strong></li><li><strong>sparse_shape</strong></li></ul></td></tr></table></div><div class="doc"><p>Deserialize and concatenate <code>SparseTensors</code> from a serialized minibatch.</p><p>The input <code>serialized_sparse</code> must be a string matrix of shape `[N x 3]` where
+ <code>N</code> is the minibatch size and the rows correspond to packed outputs of
+ <code>SerializeSparse</code>.  The ranks of the original <code>SparseTensor</code> objects
+ must all match.  When the final <code>SparseTensor</code> is created, it has rank one
+ higher than the ranks of the incoming <code>SparseTensor</code> objects
+ (they have been concatenated along a new row dimension).</p><p>The output <code>SparseTensor</code> object's shape values for all dimensions but the
+ first are the max across the input <code>SparseTensor</code> objects' shape values
+ for the corresponding dimensions.  Its first shape value is <code>N</code>, the minibatch
+ size.</p><p>The input <code>SparseTensor</code> objects' indices are assumed ordered in
+ standard lexicographic order.  If this is not the case, after this
+ step run <code>SparseReorder</code> to restore index ordering.</p><p>For example, if the serialized input is a `[2 x 3]` matrix representing two
+ original <code>SparseTensor</code> objects:</p><p>index = [ 0]
+             [10]
+             [20]
+     values = [1, 2, 3]
+     shape = [50]</p><p>and</p><p>index = [ 2]
+             [10]
+     values = [4, 5]
+     shape = [30]</p><p>then the final deserialized <code>SparseTensor</code> will be:</p><p>index = [0  0]
+             [0 10]
+             [0 20]
+             [1  2]
+             [1 10]
+     values = [1, 2, 3, 4, 5]
+     shape = [2 50]</p></div></div><div class="top"><p class="src"><a name="v:immutableConst" class="def">immutableConst</a> <a href="src/TensorFlow-GenOps-Core.html#immutableConst" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</td><td class="doc"><p><strong>tensor</strong></p></td></tr></table></div><div class="doc"><p>Returns immutable tensor from memory region.</p><p>The current implementation memmaps the tensor from a file.</p></div></div><div class="top"><p class="src"><a name="v:minimum" class="def">minimum</a> <a href="src/TensorFlow-GenOps-Core.html#minimum" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns the min of x and y (i.e. x &lt; y ? x : y) element-wise.</p><ul><li>NOTE*: <code>Minimum</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:initializeTableFromTextFile" class="def">initializeTableFromTextFile</a> <a href="src/TensorFlow-GenOps-Core.html#initializeTableFromTextFile" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>key_index</strong>: Column index in a line to get the table <code>key</code> values from.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>value_index</strong>: Column index that represents information of a line to get the table
+ <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code> values from.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>table_handle</strong>: Handle to a table which will be initialized.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>filename</strong>: Filename of a vocabulary text file.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Initializes a table from a text file.</p><p>It inserts one key-value pair into the table for each line of the file.
+ The key and value is extracted from the whole line content, elements from the
+ split line based on <code>delimiter</code> or the line number (starting from zero).
+ Where to extract the key and value from a line is specified by <code>key_index</code> and
+ <code>value_index</code>.</p><ul><li>A value of -1 means use the line number(starting from zero), expects <code>int64</code>.</li><li>A value of -2 means use the whole line content, expects <code>string</code>.</li><li>A value &gt;= 0 means use the index (starting at zero) of the split line based
+   on <code>delimiter</code>.</li></ul></div></div><div class="top"><p class="src"><a name="v:diagPart" class="def">diagPart</a> <a href="src/TensorFlow-GenOps-Core.html#diagPart" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Rank k tensor where k is 2, 4, or 6.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>diagonal</strong>: The extracted diagonal.</p></td></tr></table></div><div class="doc"><p>Returns the diagonal part of the tensor.</p><p>This operation returns a tensor with the <code>diagonal</code> part
+ of the <code>input</code>. The <code>diagonal</code> part is computed as follows:</p><p>Assume <code>input</code> has dimensions `[D1,..., Dk, D1,..., Dk]`, then the output is a
+ tensor of rank <code>k</code> with dimensions `[D1,..., Dk]` where:</p><p>`diagonal[i1,..., ik] = input[i1, ..., ik, i1,..., ik]`.</p><p>For example:</p><p>```prettyprint
+ # <code>input</code> is [[1, 0, 0, 0]
+               [0, 2, 0, 0]
+               [0, 0, 3, 0]
+               [0, 0, 0, 4]]</p><p>tf.diag_part(input) ==&gt; [1, 2, 3, 4]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:log" class="def">log</a> <a href="src/TensorFlow-GenOps-Core.html#log" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes natural logarithm of x element-wise.</p><p>I.e., \(y = log_e x\).</p></div></div><div class="top"><p class="src"><a name="v:tensorArrayScatter" class="def">tensorArrayScatter</a> <a href="src/TensorFlow-GenOps-Core.html#tensorArrayScatter" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a TensorArray.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>indices</strong>: The locations at which to write the tensor elements.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>value</strong>: The concatenated tensor to write to the TensorArray.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>flow_in</strong>: A float scalar that enforces proper chaining of operations.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>flow_out</strong>: A float scalar that enforces proper chaining of operations.</p></td></tr></table></div><div class="doc"><p>Scatter the data from the input value into specific TensorArray elements.</p><p><code>indices</code> must be a vector, its length must match the first dim of <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>.</p></div></div><div class="top"><p class="src"><a name="v:rank" class="def">rank</a> <a href="src/TensorFlow-GenOps-Core.html#rank" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Returns the rank of a tensor.</p><p>This operation returns an integer representing the rank of <code>input</code>.</p><p>For example:</p><p>```prettyprint
+ # <code>t</code> is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
+ # shape of tensor <code>t</code> is [2, 2, 3]
+ rank(t) ==&gt; 3
+ ```</p><ul><li>*Note**: The rank of a tensor is not the same as the rank of a matrix. The rank
+ of a tensor is the number of indices required to uniquely select each element
+ of the tensor. Rank is also known as &quot;order&quot;, &quot;degree&quot;, or &quot;ndims.&quot;</li></ul></div></div><div class="top"><p class="src"><a name="v:identity" class="def">identity</a> <a href="src/TensorFlow-GenOps-Core.html#identity" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Return a tensor with the same shape and contents as the input tensor or value.</p></div></div><div class="top"><p class="src"><a name="v:adjustContrastv2" class="def">adjustContrastv2</a> <a href="src/TensorFlow-GenOps-Core.html#adjustContrastv2" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>images</strong>: Images to adjust.  At least 3-D.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>contrast_factor</strong>: A float multiplier for adjusting contrast.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>output</strong>: The contrast-adjusted image or images.</p></td></tr></table></div><div class="doc"><p>Adjust the contrast of one or more images.</p><p><code>images</code> is a tensor of at least 3 dimensions.  The last 3 dimensions are
+ interpreted as `[height, width, channels]`.  The other dimensions only
+ represent a collection of images, such as `[batch, height, width, channels].`</p><p>Contrast is adjusted independently for each channel of each image.</p><p>For each channel, the Op first computes the mean of the image pixels in the
+ channel and then adjusts each component of each pixel to
+ `(x - mean) * contrast_factor + mean`.</p></div></div><div class="top"><p class="src"><a name="v:sparseApplyProximalAdagrad" class="def">sparseApplyProximalAdagrad</a> <a href="src/TensorFlow-GenOps-Core.html#sparseApplyProximalAdagrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>accum</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>lr</strong>: Learning rate. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>l1</strong>: L1 regularization. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>l2</strong>: L2 regularization. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t</td><td class="doc"><p><strong>grad</strong>: The gradient.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 tindices</td><td class="doc"><p><strong>indices</strong>: A vector of indices into the first dimension of var and accum.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Sparse update entries in '*var' and '*accum' according to FOBOS algorithm.</p><p>That is for rows we have grad for, we update var and accum as follows:
+ accum += grad * grad
+ prox_v = var
+ prox_v -= lr * grad * (1 / sqrt(accum))
+ var = sign(prox_v)/(1+lr*l2) * max{|prox_v|-lr*l1,0}</p></div></div><div class="top"><p class="src"><a name="v:gather" class="def">gather</a> <a href="src/TensorFlow-GenOps-Core.html#gather" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tparams)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tparams</td><td class="doc"><p><strong>params</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices</td><td class="doc"><p><strong>indices</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tparams</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Gather slices from <code>params</code> according to <code>indices</code>.</p><p><code>indices</code> must be an integer tensor of any dimension (usually 0-D or 1-D).
+ Produces an output tensor with shape `indices.shape + params.shape[1:]` where:</p><p># Scalar indices
+     output[:, ..., :] = params[indices, :, ... :]</p><p># Vector indices
+     output[i, :, ..., :] = params[indices[i], :, ... :]</p><p># Higher rank indices
+     output[i, ..., j, :, ... :] = params[indices[i, ..., j], :, ..., :]</p><p>If <code>indices</code> is a permutation and `len(indices) == params.shape[0]` then
+ this operation will permute <code>params</code> accordingly.</p><p><a href="div">style=&quot;width:70%; margin:auto; margin-bottom:10px; margin-top:20px;&quot;</a>
+ <a href="img">style=&quot;width:100%&quot; src=&quot;../../images/Gather.png&quot; alt</a>
+ <a href="/div">/div</a></p></div></div><div class="top"><p class="src"><a name="v:isVariableInitialized" class="def">isVariableInitialized</a> <a href="src/TensorFlow-GenOps-Core.html#isVariableInitialized" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 dtype</td><td class="doc"><p><strong>ref</strong>: Should be from a <code>Variable</code> node. May be uninitialized.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>is_initialized</strong></p></td></tr></table></div><div class="doc"><p>Checks whether a tensor has been initialized.</p><p>Outputs boolean scalar indicating whether the tensor has been initialized.</p></div></div><div class="top"><p class="src"><a name="v:concat" class="def">concat</a> <a href="src/TensorFlow-GenOps-Core.html#concat" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>concat_dim</strong>: 0-D.  The dimension along which to concatenate.  Must be in the
+ range [0, rank(values)).</p></td></tr><tr><td class="src">-&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t]</td><td class="doc"><p><strong>values</strong>: The <code>N</code> Tensors to concatenate. Their ranks and types must match,
+ and their sizes must match in all dimensions except <code>concat_dim</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: A <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> with the concatenation of values stacked along the
+ <code>concat_dim</code> dimension.  This tensor's shape matches that of <code>values</code> except
+ in <code>concat_dim</code> where it has the sum of the sizes.</p></td></tr></table></div><div class="doc"><p>Concatenates tensors along one dimension.</p></div></div><div class="top"><p class="src"><a name="v:randomUniformInt" class="def">randomUniformInt</a> <a href="src/TensorFlow-GenOps-Core.html#randomUniformInt" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tout, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tout)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>shape</strong>: The shape of the output tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tout</td><td class="doc"><p><strong>minval</strong>: 0-D.  Inclusive lower bound on the generated integers.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tout</td><td class="doc"><p><strong>maxval</strong>: 0-D.  Exclusive upper bound on the generated integers.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tout</td><td class="doc"><p><strong>output</strong>: A tensor of the specified shape filled with uniform random integers.</p></td></tr></table></div><div class="doc"><p>Outputs random integers from a uniform distribution.</p><p>The generated values are uniform integers in the range `[minval, maxval)`.
+ The lower bound <code>minval</code> is included in the range, while the upper bound
+ <code>maxval</code> is excluded.</p><p>The random integers are slightly biased unless `maxval - minval` is an exact
+ power of two.  The bias is small for values of `maxval - minval` significantly
+ smaller than the range of the output (either `2^32` or `2^64`).</p></div></div><div class="top"><p class="src"><a name="v:stopGradient" class="def">stopGradient</a> <a href="src/TensorFlow-GenOps-Core.html#stopGradient" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Stops gradient computation.</p><p>When executed in a graph, this op outputs its input tensor as-is.</p><p>When building ops to compute gradients, this op prevents the contribution of
+ its inputs to be taken into account.  Normally, the gradient generator adds ops
+ to a graph to compute the derivatives of a specified <code>loss</code> by recursively
+ finding out inputs that contributed to its computation.  If you insert this op
+ in the graph it inputs are masked from the gradient generator.  They are not
+ taken into account for computing gradients.</p><p>This is useful any time you want to compute a value with TensorFlow but need
+ to pretend that the value was a constant. Some examples include:</p><ul><li>The *EM* algorithm where the *M-step* should not involve backpropagation
+    through the output of the *E-step*.</li><li>Contrastive divergence training of Boltzmann machines where, when
+    differentiating the energy function, the training must not backpropagate
+    through the graph that generated the samples from the model.</li><li>Adversarial training, where no backprop should happen through the adversarial
+    example generation process.</li></ul></div></div><div class="top"><p class="src"><a name="v:avgPool" class="def">avgPool</a> <a href="src/TensorFlow-GenOps-Core.html#avgPool" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>value</strong>: 4-D with shape `[batch, height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The average pooled output tensor.</p></td></tr></table></div><div class="doc"><p>Performs average pooling on the input.</p><p>Each entry in <code>output</code> is the mean of the corresponding size <code>ksize</code>
+ window in <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>.</p></div></div><div class="top"><p class="src"><a name="v:wholeFileReader" class="def">wholeFileReader</a> <a href="src/TensorFlow-GenOps-Core.html#wholeFileReader" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>reader_handle</strong>: The handle to reference the Reader.</p></td></tr></table></div><div class="doc"><p>A Reader that outputs the entire contents of a file as a value.</p><p>To use, enqueue filenames in a Queue.  The output of ReaderRead will
+ be a filename (key) and the contents of that file (value).</p></div></div><div class="top"><p class="src"><a name="v:switch" class="def">switch</a> <a href="src/TensorFlow-GenOps-Core.html#switch" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong>: The tensor to be forwarded to the appropriate output.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>pred</strong>: A scalar that specifies which output port will receive data.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><td class="doc"><p>(<strong>output_false</strong>, <strong>output_true</strong>)</p><ul><li><strong>output_false</strong>: If <code><a href="../base-4.8.2.0/Prelude.html#v:pred">pred</a></code> is false, data will be forwarded to this output.</li><li><strong>output_true</strong>: If <code><a href="../base-4.8.2.0/Prelude.html#v:pred">pred</a></code> is true, data will be forwarded to this output.</li></ul></td></tr></table></div><div class="doc"><p>Forwards `data` to the output port determined by <code><a href="../base-4.8.2.0/Prelude.html#v:pred">pred</a></code>.</p><p>If <code><a href="../base-4.8.2.0/Prelude.html#v:pred">pred</a></code> is true, the `data` input is forwarded to <code>output_true</code>. Otherwise,
+ the data goes to <code>output_false</code>.</p><p>See also <code>RefSwitch</code> and <code>Merge</code>.</p></div></div><div class="top"><p class="src"><a name="v:randomStandardNormal" class="def">randomStandardNormal</a> <a href="src/TensorFlow-GenOps-Core.html#randomStandardNormal" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` dtype)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>shape</strong>: The shape of the output tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</td><td class="doc"><p><strong>output</strong>: A tensor of the specified shape filled with random normal values.</p></td></tr></table></div><div class="doc"><p>Outputs random values from a normal distribution.</p><p>The generated values will have mean 0 and standard deviation 1.</p></div></div><div class="top"><p class="src"><a name="v:sigmoid" class="def">sigmoid</a> <a href="src/TensorFlow-GenOps-Core.html#sigmoid" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes sigmoid of <code>x</code> element-wise.</p><p>Specifically, `y = 1 / (1 + exp(-x))`.</p></div></div><div class="top"><p class="src"><a name="v:sampleDistortedBoundingBox" class="def">sampleDistortedBoundingBox</a> <a href="src/TensorFlow-GenOps-Core.html#sampleDistortedBoundingBox" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>image_size</strong>: 1-D, containing `[height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>bounding_boxes</strong>: 3-D with shape `[batch, N, 4]` describing the N bounding boxes
+ associated with the image.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p>(<strong>begin</strong>, <strong>size</strong>, <strong>bboxes</strong>)</p><ul><li><strong>begin</strong>: 1-D, containing `[offset_height, offset_width, 0]`. Provide as input to
+ `tf.slice`.</li><li><strong>size</strong>: 1-D, containing `[target_height, target_width, -1]`. Provide as input to
+ `tf.slice`.</li><li><strong>bboxes</strong>: 3-D with shape `[1, 1, 4]` containing the distorted bounding box.
+ Provide as input to `tf.image.draw_bounding_boxes`.</li></ul></td></tr></table></div><div class="doc"><p>Generate a single randomly distorted bounding box for an image.</p><p>Bounding box annotations are often supplied in addition to ground-truth labels
+ in image recognition or object localization tasks. A common technique for
+ training such a system is to randomly distort an image while preserving
+ its content, i.e. *data augmentation*. This Op outputs a randomly distorted
+ localization of an object, i.e. bounding box, given an <code>image_size</code>,
+ <code>bounding_boxes</code> and a series of constraints.</p><p>The output of this Op is a single bounding box that may be used to crop the
+ original image. The output is returned as 3 tensors: <code>begin</code>, <code><a href="TensorFlow-GenOps-Core.html#v:size">size</a></code> and
+ <code>bboxes</code>. The first 2 tensors can be fed directly into `tf.slice` to crop the
+ image. The latter may be supplied to `tf.image.draw_bounding_box` to visualize
+ what the bounding box looks like.</p><p>Bounding boxes are supplied and returned as `[y_min, x_min, y_max, x_max]`. The
+ bounding box coordinates are floats in `[0.0, 1.0]` relative to the width and
+ height of the underlying image.</p><p>For example,</p><p># Generate a single distorted bounding box.
+     begin, size, bbox_for_draw = tf.image.sample_distorted_bounding_box(
+         tf.shape(image),
+         bounding_boxes=bounding_boxes)</p><p># Draw the bounding box in an image summary.
+     image_with_box = tf.image.draw_bounding_boxes(tf.expand_dims(image, 0),
+                                                   bbox_for_draw)
+     tf.image_summary(<code>images_with_box</code>, image_with_box)</p><p># Employ the bounding box to distort the image.
+     distorted_image = tf.slice(image, begin, size)</p><p>Note that if no bounding box information is available, setting
+ `use_image_if_no_bounding_boxes = true` will assume there is a single implicit
+ bounding box covering the whole image. If <code>use_image_if_no_bounding_boxes</code> is
+ false and no bounding boxes are supplied, an error is raised.</p></div></div><div class="top"><p class="src"><a name="v:greater" class="def">greater</a> <a href="src/TensorFlow-GenOps-Core.html#greater" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns the truth value of (x &gt; y) element-wise.</p><ul><li>NOTE*: <code>Greater</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:refNextIteration" class="def">refNextIteration</a> <a href="src/TensorFlow-GenOps-Core.html#refNextIteration" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong>: The tensor to be made available to the next iteration.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The same tensor as `data`.</p></td></tr></table></div><div class="doc"><p>Makes its input available to the next iteration.</p></div></div><div class="top"><p class="src"><a name="v:spaceToDepth" class="def">spaceToDepth</a> <a href="src/TensorFlow-GenOps-Core.html#spaceToDepth" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>block_size</strong>: The size of the spatial block.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>SpaceToDepth for tensors of type T.</p><p>Rearranges blocks of spatial data, into depth. More specifically,
+ this op outputs a copy of the input tensor where values from the <code>height</code>
+ and <code>width</code> dimensions are moved to the <code>depth</code> dimension.
+ The attr <code>block_size</code> indicates the input block size and how the data is moved.</p><ul><li>Non-overlapping blocks of size `block_size x block size` are rearranged
+     into depth at each location.</li><li>The depth of the output tensor is `input_depth * block_size * block_size`.</li><li>The input tensor's height and width must be divisible by block_size.</li></ul><p>That is, assuming the input is in the shape:
+ `[batch, height, width, depth]`,
+ the shape of the output will be:
+ `[batch, height<em>block_size, width</em>block_size, depth*block_size*block_size]`</p><p>This operation requires that the input tensor be of rank 4, and that
+ <code>block_size</code> be &gt;=1 and a divisor of both the input <code>height</code> and <code>width</code>.</p><p>This operation is useful for resizing the activations between convolutions
+ (but keeping all data), e.g. instead of pooling. It is also useful for training
+ purely convolutional models.</p><p>For example, given this input of shape `[1, 2, 2, 1]`, and block_size of 2:</p><p>```prettyprint
+ x = [[[[1], [2]],
+       [[3], [4]]]]
+ ```</p><p>This operation will output a tensor of shape `[1, 1, 1, 4]`:</p><p>```prettyprint
+ [[[[1, 2, 3, 4]]]]
+ ```</p><p>Here, the input has a batch of 1 and each batch element has shape `[2, 2, 1]`,
+ the corresponding output will have a single element (i.e. width and height are
+ both 1) and will have a depth of 4 channels (1 * block_size * block_size).
+ The output element shape is `[1, 1, 4]`.</p><p>For an input tensor with larger depth, here of shape `[1, 2, 2, 3]`, e.g.</p><p>```prettyprint
+ x = [[[[1, 2, 3], [4, 5, 6]],
+       [[7, 8, 9], [10, 11, 12]]]]
+ ```</p><p>This operation, for block_size of 2, will return the following tensor of shape
+ `[1, 1, 1, 12]`</p><p>```prettyprint
+ [[[[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]]]]
+ ```</p><p>Similarly, for the following input of shape `[1 4 4 1]`, and a block size of 2:</p><p>```prettyprint
+ x = [[[[1],   [2],  [5],  [6]],
+       [[3],   [4],  [7],  [8]],
+       [[9],  [10], [13],  [14]],
+       [[11], [12], [15],  [16]]]]
+ ```</p><p>the operator will return the following tensor of shape `[1 2 2 4]`:</p><p>```prettyprint
+ x = [[[[1, 2, 3, 4],
+        [5, 6, 7, 8]],
+       [[9, 10, 11, 12],
+        [13, 14, 15, 16]]]]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:controlTrigger" class="def">controlTrigger</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a> <a href="src/TensorFlow-GenOps-Core.html#controlTrigger" class="link">Source</a></p><div class="doc"><p>Does nothing. Serves as a control trigger for scheduling.</p><p>Only useful as a placeholder for control edges.</p></div></div><div class="top"><p class="src"><a name="v:scatterDiv" class="def">scatterDiv</a> <a href="src/TensorFlow-GenOps-Core.html#scatterDiv" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>ref</strong>: Should be from a <code>Variable</code> node.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices</td><td class="doc"><p><strong>indices</strong>: A tensor of indices into the first dimension of <code>ref</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>updates</strong>: A tensor of values that <code>ref</code> is divided by.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output_ref</strong>: = Same as <code>ref</code>.  Returned as a convenience for operations that want
+ to use the updated values after the update is done.</p></td></tr></table></div><div class="doc"><p>Divides a variable reference by sparse updates.</p><p>This operation computes</p><p># Scalar indices
+     ref[indices, ...] /= updates[...]</p><p># Vector indices (for each i)
+     ref[indices[i], ...] /= updates[i, ...]</p><p># High rank indices (for each i, ..., j)
+     ref[indices[i, ..., j], ...] /= updates[i, ..., j, ...]</p><p>This operation outputs <code>ref</code> after the update is done.
+ This makes it easier to chain operations that need to use the reset value.</p><p>Duplicate entries are handled correctly: if multiple <code>indices</code> reference
+ the same location, their contributions divide.</p><p>Requires `updates.shape = indices.shape + ref.shape[1:]`.</p></div></div><div class="top"><p class="src"><a name="v:copy" class="def">copy</a> <a href="src/TensorFlow-GenOps-Core.html#copy" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Input tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Output tensor, deep-copied from input.</p></td></tr></table></div><div class="doc"><p>Copy Op.</p><p>Performs CPU-to-CPU or GPU-to-GPU deep-copying of tensor, depending on the
+ device on which the tensor is allocated.</p><p>Unlike the CopyHost Op, this op does not have HostMemory constraint on its
+ input or output.</p></div></div><div class="top"><p class="src"><a name="v:cropAndResizeGradBoxes" class="def">cropAndResizeGradBoxes</a> <a href="src/TensorFlow-GenOps-Core.html#cropAndResizeGradBoxes" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>grads</strong>: A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>image</strong>: A 4-D tensor of shape `[batch, image_height, image_width, depth]`.
+ Both <code>image_height</code> and <code>image_width</code> need to be positive.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>boxes</strong>: A 2-D tensor of shape `[num_boxes, 4]`. The <code>i</code>-th row of the tensor
+ specifies the coordinates of a box in the `box_ind[i]` image and is specified
+ in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of
+ <code>y</code> is mapped to the image coordinate at `y * (image_height - 1)`, so as the
+ `[0, 1]` interval of normalized image height is mapped to
+ `[0, image_height - 1] in image height coordinates. We do allow y1 &gt; y2, in
+ which case the sampled crop is an up-down flipped version of the original
+ image. The width dimension is treated similarly. Normalized coordinates
+ outside the `[0, 1]` range are allowed, in which case we use
+ <code>extrapolation_value</code> to extrapolate the input image values.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>box_ind</strong>: A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.
+ The value of `box_ind[i]` specifies the image that the <code>i</code>-th box refers to.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>output</strong>: A 2-D tensor of shape `[num_boxes, 4]`.</p></td></tr></table></div><div class="doc"><p>Computes the gradient of the crop_and_resize op wrt the input boxes tensor.</p></div></div><div class="top"><p class="src"><a name="v:sparseSegmentMean" class="def">sparseSegmentMean</a> <a href="src/TensorFlow-GenOps-Core.html#sparseSegmentMean" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>indices</strong>: A 1-D tensor. Has same rank as <code>segment_ids</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>segment_ids</strong>: A 1-D tensor. Values should be sorted and can be repeated.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Has same shape as data, except for dimension 0 which
+ has size <code>k</code>, the number of segments.</p></td></tr></table></div><div class="doc"><p>Computes the mean along sparse segments of a tensor.</p><p>Read <a href="../../api_docs/python/math_ops.md#segmentation">the section on
+ Segmentation</a> for an explanation
+ of segments.</p><p>Like <code>SegmentMean</code>, but <code>segment_ids</code> can have rank less than `data`'s first
+ dimension, selecting a subset of dimension 0, specified by <code>indices</code>.</p></div></div><div class="top"><p class="src"><a name="v:assign" class="def">assign</a> <a href="src/TensorFlow-GenOps-Core.html#assign" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>ref</strong>: Should be from a <code>Variable</code> node. May be uninitialized.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>value</strong>: The value to be assigned to the variable.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output_ref</strong>: = Same as &quot;ref&quot;.  Returned as a convenience for operations that want
+ to use the new value after the variable has been reset.</p></td></tr></table></div><div class="doc"><p>Update <code>ref</code> by assigning <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code> to it.</p><p>This operation outputs &quot;ref&quot; after the assignment is done.
+ This makes it easier to chain operations that need to use the reset value.</p></div></div><div class="top"><p class="src"><a name="v:restore" class="def">restore</a> <a href="src/TensorFlow-GenOps-Core.html#restore" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dt</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>file_pattern</strong>: Must have a single element. The pattern of the files from
+ which we read the tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>tensor_name</strong>: Must have a single element. The name of the tensor to be
+ restored.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dt</td><td class="doc"><p><strong>tensor</strong>: The restored tensor.</p></td></tr></table></div><div class="doc"><p>Restores a tensor from checkpoint files.</p><p>Reads a tensor stored in one or several files. If there are several files (for
+ instance because a tensor was saved as slices), <code>file_pattern</code> may contain
+ wildcard symbols (<code><a href="../base-4.8.2.0/Prelude.html#v:-42-">*</a></code> and <code>?</code>) in the filename portion only, not in the
+ directory portion.</p><p>If a <code>file_pattern</code> matches several files, <code>preferred_shard</code> can be used to hint
+ in which file the requested tensor is likely to be found. This op will first
+ open the file at index <code>preferred_shard</code> in the list of matching files and try
+ to restore tensors from that file.  Only if some tensors or tensor slices are
+ not found in that first file, then the Op opens all the files. Setting
+ <code>preferred_shard</code> to match the value passed as the <code>shard</code> input
+ of a matching <code>Save</code> Op may speed up Restore.  This attribute only affects
+ performance, not correctness.  The default value -1 means files are processed in
+ order.</p><p>See also <code>RestoreSlice</code>.</p></div></div><div class="top"><p class="src"><a name="v:maxPoolGradWithArgmax" class="def">maxPoolGradWithArgmax</a> <a href="src/TensorFlow-GenOps-Core.html#maxPoolGradWithArgmax" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> targmax, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` targmax)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: The original input.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>grad</strong>: 4-D with shape `[batch, height, width, channels]`.  Gradients w.r.t. the
+ output of <code>max_pool</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 targmax</td><td class="doc"><p><strong>argmax</strong>: The indices of the maximum values chosen for each output of <code>max_pool</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Gradients w.r.t. the input of <code>max_pool</code>.</p></td></tr></table></div><div class="doc"><p>Computes gradients of the maxpooling function.</p></div></div><div class="top"><p class="src"><a name="v:checkNumerics" class="def">checkNumerics</a> <a href="src/TensorFlow-GenOps-Core.html#checkNumerics" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>tensor</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Checks a tensor for NaN and Inf values.</p><p>When run, reports an <code>InvalidArgument</code> error if <code>tensor</code> has any values
+ that are not a number (NaN) or infinity (Inf). Otherwise, passes <code>tensor</code> as-is.</p></div></div><div class="top"><p class="src"><a name="v:zerosLike" class="def">zerosLike</a> <a href="src/TensorFlow-GenOps-Core.html#zerosLike" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong>: a tensor of type T.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong>: a tensor of the same shape and type as x but filled with zeros.</p></td></tr></table></div><div class="doc"><p>Returns a tensor of zeros with the same shape and type as x.</p></div></div><div class="top"><p class="src"><a name="v:readFile" class="def">readFile</a> <a href="src/TensorFlow-GenOps-Core.html#readFile" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>filename</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>contents</strong></p></td></tr></table></div><div class="doc"><p>Reads and outputs the entire contents of the input filename.</p></div></div><div class="top"><p class="src"><a name="v:transpose" class="def">transpose</a> <a href="src/TensorFlow-GenOps-Core.html#transpose" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tperm, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tperm)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tperm</td><td class="doc"><p><strong>perm</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Shuffle dimensions of x according to a permutation.</p><p>The output <code>y</code> has the same rank as <code>x</code>. The shapes of <code>x</code> and <code>y</code> satisfy:
+   `y.shape[i] == x.shape[perm[i]] for i in [0, 1, ..., rank(x) - 1]`</p></div></div><div class="top"><p class="src"><a name="v:parseTensor" class="def">parseTensor</a> <a href="src/TensorFlow-GenOps-Core.html#parseTensor" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_type</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>serialized</strong>: A scalar string containing a serialized TensorProto proto.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_type</td><td class="doc"><p><strong>output</strong>: A Tensor of type <code>out_type</code>.</p></td></tr></table></div><div class="doc"><p>Transforms a serialized tensorflow.TensorProto proto into a Tensor.</p></div></div><div class="top"><p class="src"><a name="v:acos" class="def">acos</a> <a href="src/TensorFlow-GenOps-Core.html#acos" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes acos of x element-wise.</p></div></div><div class="top"><p class="src"><a name="v:bitcast" class="def">bitcast</a> <a href="src/TensorFlow-GenOps-Core.html#bitcast" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> type', <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` type')</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> type'</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Bitcasts a tensor from one type to another without copying data.</p><p>Given a tensor <code>input</code>, this operation returns a tensor that has the same buffer
+ data as <code>input</code> with datatype `type`.</p><p>If the input datatype <code>T</code> is larger than the output datatype `type` then the
+ shape changes from [...] to [..., sizeof(<code>T</code>)/sizeof(`type`)].</p><p>If <code>T</code> is smaller than `type`, the operator requires that the rightmost
+ dimension be equal to sizeof(`type`)/sizeof(<code>T</code>). The shape then goes from
+ [..., sizeof(`type`)/sizeof(<code>T</code>)] to [...].</p><ul><li>NOTE*: Bitcast is implemented as a low-level cast, so machines with different
+ endian orderings will give different results.</li></ul></div></div><div class="top"><p class="src"><a name="v:lookupTableImport" class="def">lookupTableImport</a> <a href="src/TensorFlow-GenOps-Core.html#lookupTableImport" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tin, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tout)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>table_handle</strong>: Handle to the table.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tin</td><td class="doc"><p><strong>keys</strong>: Any shape.  Keys to look up.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tout</td><td class="doc"><p><strong>values</strong>: Values to associate with keys.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Replaces the contents of the table with the specified keys and values.</p><p>The tensor <code>keys</code> must be of the same type as the keys of the table.
+ The tensor <code>values</code> must be of the type of the table values.</p></div></div><div class="top"><p class="src"><a name="v:biasAddGrad" class="def">biasAddGrad</a> <a href="src/TensorFlow-GenOps-Core.html#biasAddGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>out_backprop</strong>: Any number of dimensions.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 1-D with size the feature dimension of <code>out_backprop</code>.</p></td></tr></table></div><div class="doc"><p>The backward operation for <a href="BiasAdd.html">BiasAdd</a> on the &quot;bias&quot; tensor.</p><p>It accumulates all the values from out_backprop into the feature dimension.
+ For NHWC data format, the feature dimension is the last. For NCHW data format,
+ the feature dimension is the third-to-last.</p></div></div><div class="top"><p class="src"><a name="v:batchSelfAdjointEig" class="def">batchSelfAdjointEig</a> <a href="src/TensorFlow-GenOps-Core.html#batchSelfAdjointEig" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:prod" class="def">prod</a> <a href="src/TensorFlow-GenOps-Core.html#prod" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: The tensor to reduce.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>reduction_indices</strong>: The dimensions to reduce.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The reduced tensor.</p></td></tr></table></div><div class="doc"><p>Computes the product of elements across dimensions of a tensor.</p><p>Reduces <code>input</code> along the dimensions given in <code>reduction_indices</code>. Unless
+ <code>keep_dims</code> is true, the rank of the tensor is reduced by 1 for each entry in
+ <code>reduction_indices</code>. If <code>keep_dims</code> is true, the reduced dimensions are
+ retained with length 1.</p></div></div><div class="top"><p class="src"><a name="v:resizeBilinear" class="def">resizeBilinear</a> <a href="src/TensorFlow-GenOps-Core.html#resizeBilinear" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>images</strong>: 4-D with shape `[batch, height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>size</strong>: = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
+ new size for the images.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>resized_images</strong>: 4-D with shape
+ `[batch, new_height, new_width, channels]`.</p></td></tr></table></div><div class="doc"><p>Resize <code>images</code> to <code><a href="TensorFlow-GenOps-Core.html#v:size">size</a></code> using bilinear interpolation.</p><p>Input images can be of different types but output images are always float.</p></div></div><div class="top"><p class="src"><a name="v:tensorArrayUnpack" class="def">tensorArrayUnpack</a> <a href="src/TensorFlow-GenOps-Core.html#tensorArrayUnpack" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a TensorArray.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>value</strong>: The concatenated tensor to write to the TensorArray.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>flow_in</strong>: A float scalar that enforces proper chaining of operations.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>flow_out</strong>: A float scalar that enforces proper chaining of operations.</p></td></tr></table></div><div class="doc"><p>Unpack the data from the input value into TensorArray elements.</p><ul><li>*WARNING: This op is deprecated.**</li></ul><p>Instead of this op, use <code>TensorArrayScatter</code> with
+ `indices = RangeOp(0, SizeOp(value)[0])`.</p></div></div><div class="top"><p class="src"><a name="v:batchMatrixDeterminant" class="def">batchMatrixDeterminant</a> <a href="src/TensorFlow-GenOps-Core.html#batchMatrixDeterminant" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:sum" class="def">sum</a> <a href="src/TensorFlow-GenOps-Core.html#sum" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: The tensor to reduce.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>reduction_indices</strong>: The dimensions to reduce.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The reduced tensor.</p></td></tr></table></div><div class="doc"><p>Computes the sum of elements across dimensions of a tensor.</p><p>Reduces <code>input</code> along the dimensions given in <code>reduction_indices</code>. Unless
+ <code>keep_dims</code> is true, the rank of the tensor is reduced by 1 for each entry in
+ <code>reduction_indices</code>. If <code>keep_dims</code> is true, the reduced dimensions are
+ retained with length 1.</p></div></div><div class="top"><p class="src"><a name="v:iFFT2D" class="def">iFFT2D</a> <a href="src/TensorFlow-GenOps-Core.html#iFFT2D" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>input</strong>: A complex64 tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>output</strong>: A complex64 tensor of the same shape as <code>input</code>. The inner-most 2
+ dimensions of <code>input</code> are replaced with their inverse 2D Fourier Transform.</p></td></tr></table></div><div class="doc"><p>Compute the inverse 2-dimensional discrete Fourier Transform over the inner-most</p><p>2 dimensions of <code>input</code>.</p></div></div><div class="top"><p class="src"><a name="v:fill" class="def">fill</a> <a href="src/TensorFlow-GenOps-Core.html#fill" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>dims</strong>: 1-D. Represents the shape of the output tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>value</strong>: 0-D (scalar). Value to fill the returned tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Creates a tensor filled with a scalar value.</p><p>This operation creates a tensor of shape <code>dims</code> and fills it with <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>.</p><p>For example:</p><p>```prettyprint
+ # Output tensor has shape [2, 3].
+ fill([2, 3], 9) ==&gt; [[9, 9, 9]
+                      [9, 9, 9]]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:fixedUnigramCandidateSampler" class="def">fixedUnigramCandidateSampler</a> <a href="src/TensorFlow-GenOps-Core.html#fixedUnigramCandidateSampler" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_sampled</strong>: Number of candidates to randomly sample per batch.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_true</strong>: Number of true labels per context.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>range_max</strong>: The sampler will sample integers from the interval [0, range_max).</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>unique</strong>: If unique is true, we sample with rejection, so that all sampled
+ candidates in a batch are unique. This requires some approximation to
+ estimate the post-rejection sampling probabilities.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>true_classes</strong>: A batch_size * num_true matrix, in which each row contains the
+ IDs of the num_true target_classes in the corresponding original label.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p>(<strong>sampled_candidates</strong>, <strong>true_expected_count</strong>, <strong>sampled_expected_count</strong>)</p><ul><li><strong>sampled_candidates</strong>: A vector of length num_sampled, in which each element is
+ the ID of a sampled candidate.</li><li><strong>true_expected_count</strong>: A batch_size * num_true matrix, representing
+ the number of times each candidate is expected to occur in a batch
+ of sampled candidates. If unique=true, then this is a probability.</li><li><strong>sampled_expected_count</strong>: A vector of length num_sampled, for each sampled
+ candidate representing the number of times the candidate is expected
+ to occur in a batch of sampled candidates.  If unique=true, then this is a
+ probability.</li></ul></td></tr></table></div><div class="doc"><p>Generates labels for candidate sampling with a learned unigram distribution.</p><p>A unigram sampler could use a fixed unigram distribution read from a
+ file or passed in as an in-memory array instead of building up the distribution
+ from data on the fly. There is also an option to skew the distribution by
+ applying a distortion power to the weights.</p><p>The vocabulary file should be in CSV-like format, with the last field
+ being the weight associated with the word.</p><p>For each batch, this op picks a single set of sampled candidate labels.</p><p>The advantages of sampling candidates per-batch are simplicity and the
+ possibility of efficient dense matrix multiplication. The disadvantage is that
+ the sampled candidates must be chosen independently of the context and of the
+ true labels.</p></div></div><div class="top"><p class="src"><a name="v:dilation2D" class="def">dilation2D</a> <a href="src/TensorFlow-GenOps-Core.html#dilation2D" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: 4-D with shape `[batch, in_height, in_width, depth]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>filter</strong>: 3-D with shape `[filter_height, filter_width, depth]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 4-D with shape `[batch, out_height, out_width, depth]`.</p></td></tr></table></div><div class="doc"><p>Computes the grayscale dilation of 4-D <code>input</code> and 3-D <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code> tensors.</p><p>The <code>input</code> tensor has shape `[batch, in_height, in_width, depth]` and the
+ <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code> tensor has shape `[filter_height, filter_width, depth]`, i.e., each
+ input channel is processed independently of the others with its own structuring
+ function. The <code>output</code> tensor has shape
+ `[batch, out_height, out_width, depth]`. The spatial dimensions of the output
+ tensor depend on the <code>padding</code> algorithm. We currently only support the default
+ <a href="NHWC.html">NHWC</a> <code>data_format</code>.</p><p>In detail, the grayscale morphological 2-D dilation is the max-sum correlation
+ (for consistency with <code>conv2d</code>, we use unmirrored filters):</p><p>output[b, y, x, c] =
+        max_{dy, dx} input[b,
+                           strides[1] * y + rates[1] * dy,
+                           strides[2] * x + rates[2] * dx,
+                           c] +
+                     filter[dy, dx, c]</p><p>Max-pooling is a special case when the filter has size equal to the pooling
+ kernel size and contains all zeros.</p><p>Note on duality: The dilation of <code>input</code> by the <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code> is equal to the
+ negation of the erosion of `-input` by the reflected <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code>.</p></div></div><div class="top"><p class="src"><a name="v:polygamma" class="def">polygamma</a> <a href="src/TensorFlow-GenOps-Core.html#polygamma" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>a</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Compute the polygamma function \(psi^{(n)}(x)\).</p><p>The polygamma function is defined as:</p><p>```
+ psi^{(n)}(x) = frac{d^n}{dx^n} psi(x)
+ ```
+ where \(psi(x)\) is the digamma function.</p></div></div><div class="top"><p class="src"><a name="v:refIdentity" class="def">refIdentity</a> <a href="src/TensorFlow-GenOps-Core.html#refIdentity" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Return the same ref tensor as the input ref tensor.</p></div></div><div class="top"><p class="src"><a name="v:encodePng" class="def">encodePng</a> <a href="src/TensorFlow-GenOps-Core.html#encodePng" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>image</strong>: 3-D with shape `[height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>contents</strong>: 0-D. PNG-encoded image.</p></td></tr></table></div><div class="doc"><p>PNG-encode an image.</p><p><code>image</code> is a 3-D uint8 or uint16 Tensor of shape `[height, width, channels]`
+ where <code>channels</code> is:</p><ul><li>1: for grayscale.</li><li>2: for grayscale + alpha.</li><li>3: for RGB.</li><li>4: for RGBA.</li></ul><p>The ZLIB compression level, <code>compression</code>, can be -1 for the PNG-encoder
+ default or a value from 0 to 9.  9 is the highest compression level, generating
+ the smallest output, but is slower.</p></div></div><div class="top"><p class="src"><a name="v:lookupTableInsert" class="def">lookupTableInsert</a> <a href="src/TensorFlow-GenOps-Core.html#lookupTableInsert" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tin, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tout)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>table_handle</strong>: Handle to the table.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tin</td><td class="doc"><p><strong>keys</strong>: Any shape.  Keys to look up.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tout</td><td class="doc"><p><strong>values</strong>: Values to associate with keys.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Updates the table to associates keys with values.</p><p>The tensor <code>keys</code> must be of the same type as the keys of the table.
+ The tensor <code>values</code> must be of the type of the table values.</p></div></div><div class="top"><p class="src"><a name="v:batchIFFT2D" class="def">batchIFFT2D</a> <a href="src/TensorFlow-GenOps-Core.html#batchIFFT2D" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:uniqueWithCounts" class="def">uniqueWithCounts</a> <a href="src/TensorFlow-GenOps-Core.html#uniqueWithCounts" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_idx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` out_idx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong>: 1-D.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_idx, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_idx)</td><td class="doc"><p>(<strong>y</strong>, <strong>idx</strong>, <strong>count</strong>)</p><ul><li><strong>y</strong>: 1-D.</li><li><strong>idx</strong>: 1-D.</li><li><strong>count</strong>: 1-D.</li></ul></td></tr></table></div><div class="doc"><p>Finds unique elements in a 1-D tensor.</p><p>This operation returns a tensor <code>y</code> containing all of the unique elements of <code>x</code>
+ sorted in the same order that they occur in <code>x</code>. This operation also returns a
+ tensor <code>idx</code> the same size as <code>x</code> that contains the index of each value of <code>x</code>
+ in the unique output <code>y</code>. Finally, it returns a third tensor <code>count</code> that
+ contains the count of each element of <code>y</code> in <code>x</code>. In other words:</p><p>`y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`</p><p>For example:</p><p>```prettyprint
+ # tensor <code>x</code> is [1, 1, 2, 4, 4, 4, 7, 8, 8]
+ y, idx, count = unique_with_counts(x)
+ y ==&gt; [1, 2, 4, 7, 8]
+ idx ==&gt; [0, 0, 1, 2, 2, 2, 3, 4, 4]
+ count ==&gt; [2, 1, 3, 1, 2]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:gatherNd" class="def">gatherNd</a> <a href="src/TensorFlow-GenOps-Core.html#gatherNd" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tparams)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tparams</td><td class="doc"><p><strong>params</strong>: `M-D`.  The tensor from which to gather values.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices</td><td class="doc"><p><strong>indices</strong>: `(N+1)-D`.  Index tensor having shape `[d_0, ..., d_N, R]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tparams</td><td class="doc"><p><strong>output</strong>: `(N+M-R)-D`.  Values from <code>params</code> gathered from indices given by
+ <code>indices</code>.</p></td></tr></table></div><div class="doc"><p>Gather values or slices from <code>params</code> according to <code>indices</code>.</p><p><code>params</code> is a Tensor of rank <code>R</code> and <code>indices</code> is a Tensor of rank <code>M</code>.</p><p><code>indices</code> must be integer tensor, containing indices into <code>params</code>.
+ It must be shape `[d_0, ..., d_N, R]` where `0 &lt; R &lt;= M`.</p><p>The innermost dimension of <code>indices</code> (with length <code>R</code>) corresponds to
+ indices into elements (if `R = M`) or slices (if `R &lt; M`) along the <code>N</code>th
+ dimension of <code>params</code>.</p><p>Produces an output tensor with shape</p><dl><dt>d_0, ..., d_{n-1}, params.shape[R</dt><dd>, ..., params.shape[M-1]].</dd></dl><p>Some examples below.</p><p>Simple indexing into a matrix:</p><p>indices = [[0, 0], [1, 1]]
+     params = [[<code>a</code>, <code>b</code>], [<code>c</code>, <code>d</code>]]
+     output = [<code>a</code>, <code>d</code>]</p><p>Slice indexing into a matrix:</p><p>indices = [[1], [0]]
+     params = [[<code>a</code>, <code>b</code>], [<code>c</code>, <code>d</code>]]
+     output = [[<code>c</code>, <code>d</code>], [<code>a</code>, <code>b</code>]]</p><p>Indexing into a 3-tensor:</p><p>indices = [[1]]
+     params = [[[<code>a0</code>, <code>b0</code>], [<code>c0</code>, <code>d0</code>]],
+               [[<code>a1</code>, <code>b1</code>], [<code>c1</code>, <code>d1</code>]]]
+     output = [[[<code>a1</code>, <code>b1</code>], [<code>c1</code>, <code>d1</code>]]]</p><p>indices = [[0, 1], [1, 0]]
+     params = [[[<code>a0</code>, <code>b0</code>], [<code>c0</code>, <code>d0</code>]],
+               [[<code>a1</code>, <code>b1</code>], [<code>c1</code>, <code>d1</code>]]]
+     output = [[<code>c0</code>, <code>d0</code>], [<code>a1</code>, <code>b1</code>]]</p><p>indices = [[0, 0, 1], [1, 0, 1]]
+     params = [[[<code>a0</code>, <code>b0</code>], [<code>c0</code>, <code>d0</code>]],
+               [[<code>a1</code>, <code>b1</code>], [<code>c1</code>, <code>d1</code>]]]
+     output = [<code>b0</code>, <code>b1</code>]</p><p>Batched indexing into a matrix:</p><p>indices = [[[0, 0]], [[0, 1]]]
+     params = [[<code>a</code>, <code>b</code>], [<code>c</code>, <code>d</code>]]
+     output = [[<code>a</code>], [<code>b</code>]]</p><p>Batched slice indexing into a matrix:</p><p>indices = [[[1]], [[0]]]
+     params = [[<code>a</code>, <code>b</code>], [<code>c</code>, <code>d</code>]]
+     output = [[[<code>c</code>, <code>d</code>]], [[<code>a</code>, <code>b</code>]]]</p><p>Batched indexing into a 3-tensor:</p><p>indices = [[[1]], [[0]]]
+     params = [[[<code>a0</code>, <code>b0</code>], [<code>c0</code>, <code>d0</code>]],
+               [[<code>a1</code>, <code>b1</code>], [<code>c1</code>, <code>d1</code>]]]
+     output = [[[[<code>a1</code>, <code>b1</code>], [<code>c1</code>, <code>d1</code>]]],
+               [[[<code>a0</code>, <code>b0</code>], [<code>c0</code>, <code>d0</code>]]]]</p><p>indices = [[[0, 1], [1, 0]], [[0, 0], [1, 1]]]
+     params = [[[<code>a0</code>, <code>b0</code>], [<code>c0</code>, <code>d0</code>]],
+               [[<code>a1</code>, <code>b1</code>], [<code>c1</code>, <code>d1</code>]]]
+     output = [[[<code>c0</code>, <code>d0</code>], [<code>a1</code>, <code>b1</code>]],
+               [[<code>a0</code>, <code>b0</code>], [<code>c1</code>, <code>d1</code>]]]</p><p>indices = [[[0, 0, 1], [1, 0, 1]], [[0, 1, 1], [1, 1, 0]]]
+     params = [[[<code>a0</code>, <code>b0</code>], [<code>c0</code>, <code>d0</code>]],
+               [[<code>a1</code>, <code>b1</code>], [<code>c1</code>, <code>d1</code>]]]
+     output = [[<code>b0</code>, <code>b1</code>], [<code>d0</code>, <code>c1</code>]]</p></div></div><div class="top"><p class="src"><a name="v:tensorArrayRead" class="def">tensorArrayRead</a> <a href="src/TensorFlow-GenOps-Core.html#tensorArrayRead" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a TensorArray.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>index</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>flow_in</strong>: A float scalar that enforces proper chaining of operations.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</td><td class="doc"><p><strong>value</strong>: The tensor that is read from the TensorArray.</p></td></tr></table></div><div class="doc"><p>Read an element from the TensorArray into output <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>.</p></div></div><div class="top"><p class="src"><a name="v:readerReadUpTo" class="def">readerReadUpTo</a> <a href="src/TensorFlow-GenOps-Core.html#readerReadUpTo" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>reader_handle</strong>: Handle to a <code>Reader</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>queue_handle</strong>: Handle to a <code>Queue</code>, with string work items.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_records</strong>: number of records to read from <code>Reader</code>.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>)</td><td class="doc"><p>(<strong>keys</strong>, <strong>values</strong>)</p><ul><li><strong>keys</strong>: A 1-D tensor.</li><li><strong>values</strong>: A 1-D tensor.</li></ul></td></tr></table></div><div class="doc"><p>Returns up to <code>num_records</code> (key, value) pairs produced by a Reader.</p><p>Will dequeue from the input queue if necessary (e.g. when the
+ Reader needs to start reading from a new file since it has finished
+ with the previous file).
+ It may return less than <code>num_records</code> even before the last batch.</p></div></div><div class="top"><p class="src"><a name="v:betainc" class="def">betainc</a> <a href="src/TensorFlow-GenOps-Core.html#betainc" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>a</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>b</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Compute the regularized incomplete beta integral \(I_x(a, b)\).</p><p>The regularized incomplete beta integral is defined as:</p><p>```
+ I_x(a, b) = frac{B(x; a, b)}{B(a, b)}
+ ```
+ where</p><p>```
+ B(x; a, b) = int_0^x t^{a-1} (1 - t)^{b-1} dt
+ ```</p><p>is the incomplete beta function and \(B(a, b)\) is the *complete*
+ beta function.</p></div></div><div class="top"><p class="src"><a name="v:batchMatrixBandPart" class="def">batchMatrixBandPart</a> <a href="src/TensorFlow-GenOps-Core.html#batchMatrixBandPart" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_lower</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_upper</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>band</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:depthwiseConv2dNativeBackpropInput" class="def">depthwiseConv2dNativeBackpropInput</a> <a href="src/TensorFlow-GenOps-Core.html#depthwiseConv2dNativeBackpropInput" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>input_sizes</strong>: An integer vector representing the shape of <code>input</code>,
+ where <code>input</code> is a 4-D `[batch, height, width, channels]` tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>filter</strong>: 4-D with shape
+ `[filter_height, filter_width, in_channels, depthwise_multiplier]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>out_backprop</strong>: 4-D with shape `[batch, out_height, out_width, out_channels]`.
+ Gradients w.r.t. the output of the convolution.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 4-D with shape `[batch, in_height, in_width, in_channels]`.  Gradient
+ w.r.t. the input of the convolution.</p></td></tr></table></div><div class="doc"><p>Computes the gradients of depthwise convolution with respect to the input.</p></div></div><div class="top"><p class="src"><a name="v:refSelect" class="def">refSelect</a> <a href="src/TensorFlow-GenOps-Core.html#refSelect" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>index</strong>: A scalar that determines the input that gets selected.</p></td></tr><tr><td class="src">-&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t]</td><td class="doc"><p><strong>inputs</strong>: A list of ref tensors, one of which will be forwarded to <code>output</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The forwarded tensor.</p></td></tr></table></div><div class="doc"><p>Forwards the <code>index</code>th element of <code>inputs</code> to <code>output</code>.</p></div></div><div class="top"><p class="src"><a name="v:exit" class="def">exit</a> <a href="src/TensorFlow-GenOps-Core.html#exit" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong>: The tensor to be made available to the parent frame.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The same tensor as `data`.</p></td></tr></table></div><div class="doc"><p>Exits the current frame to its parent frame.</p><p>Exit makes its input `data` available to the parent frame.</p></div></div><div class="top"><p class="src"><a name="v:lookupTableFind" class="def">lookupTableFind</a> <a href="src/TensorFlow-GenOps-Core.html#lookupTableFind" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tin, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tout)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>table_handle</strong>: Handle to the table.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tin</td><td class="doc"><p><strong>keys</strong>: Any shape.  Keys to look up.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tout</td><td class="doc"><p><strong>default_value</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tout</td><td class="doc"><p><strong>values</strong>: Same shape as <code>keys</code>.  Values found in the table, or <code>default_values</code>
+ for missing keys.</p></td></tr></table></div><div class="doc"><p>Looks up keys in a table, outputs the corresponding values.</p><p>The tensor <code>keys</code> must of the same type as the keys of the table.
+ The output <code>values</code> is of the type of the table values.</p><p>The scalar <code>default_value</code> is the value output for keys not present in the
+ table. It must also be of the same type as the table values.</p></div></div><div class="top"><p class="src"><a name="v:squeeze" class="def">squeeze</a> <a href="src/TensorFlow-GenOps-Core.html#squeeze" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: The <code>input</code> to squeeze.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Contains the same data as <code>input</code>, but has one or more dimensions of
+ size 1 removed.</p></td></tr></table></div><div class="doc"><p>Removes dimensions of size 1 from the shape of a tensor.</p><p>Given a tensor <code>input</code>, this operation returns a tensor of the same type with
+ all dimensions of size 1 removed. If you don't want to remove all size 1
+ dimensions, you can remove specific size 1 dimensions by specifying
+ <code>squeeze_dims</code>.</p><p>For example:</p><p>```prettyprint
+ # <code>t</code> is a tensor of shape [1, 2, 1, 3, 1, 1]
+ shape(squeeze(t)) ==&gt; [2, 3]
+ ```</p><p>Or, to remove specific size 1 dimensions:</p><p>```prettyprint
+ # <code>t</code> is a tensor of shape [1, 2, 1, 3, 1, 1]
+ shape(squeeze(t, [2, 4])) ==&gt; [1, 2, 3, 1]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:mean" class="def">mean</a> <a href="src/TensorFlow-GenOps-Core.html#mean" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: The tensor to reduce.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>reduction_indices</strong>: The dimensions to reduce.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The reduced tensor.</p></td></tr></table></div><div class="doc"><p>Computes the mean of elements across dimensions of a tensor.</p><p>Reduces <code>input</code> along the dimensions given in <code>reduction_indices</code>. Unless
+ <code>keep_dims</code> is true, the rank of the tensor is reduced by 1 for each entry in
+ <code>reduction_indices</code>. If <code>keep_dims</code> is true, the reduced dimensions are
+ retained with length 1.</p></div></div><div class="top"><p class="src"><a name="v:spaceToBatchND" class="def">spaceToBatchND</a> <a href="src/TensorFlow-GenOps-Core.html#spaceToBatchND" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tblock_shape, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tblock_shape, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tpaddings, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tpaddings)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: N-D with shape `input_shape = [batch] + spatial_shape + remaining_shape`,
+ where spatial_shape has <code>M</code> dimensions.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tblock_shape</td><td class="doc"><p><strong>block_shape</strong>: 1-D with shape `[M]`, all values must be &gt;= 1.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tpaddings</td><td class="doc"><p><strong>paddings</strong>: 2-D with shape `[M, 2]`, all values must be &gt;= 0.
+   `paddings[i] = [pad_start, pad_end]` specifies the padding for input dimension
+   `i + 1`, which corresponds to spatial dimension <code>i</code>.  It is required that
+   `block_shape[i]` divides `input_shape[i + 1] + pad_start + pad_end`.</p><p>This operation is equivalent to the following steps:</p><ol><li>Zero-pad the start and end of dimensions `[1, ..., M]` of the
+    input according to <code>paddings</code> to produce <code>padded</code> of shape <code>padded_shape</code>.</li><li>Reshape <code>padded</code> to <code>reshaped_padded</code> of shape:
+      [batch] +
+      [padded_shape[1] / block_shape[0],
+        block_shape[0],
+       ...,
+       padded_shape[M] / block_shape[M-1],
+       block_shape[M-1]] +
+      remaining_shape</li><li>Permute dimensions of <code>reshaped_padded</code> to produce
+    <code>permuted_reshaped_padded</code> of shape:
+      block_shape +
+      [batch] +
+      [padded_shape[1] / block_shape[0],
+       ...,
+       padded_shape[M] / block_shape[M-1]] +
+      remaining_shape</li><li>Reshape <code>permuted_reshaped_padded</code> to flatten <code>block_shape</code> into the batch
+    dimension, producing an output tensor of shape:
+      [batch * prod(block_shape)] +
+      [padded_shape[1] / block_shape[0],
+       ...,
+       padded_shape[M] / block_shape[M-1]] +
+      remaining_shape</li></ol><p>Some examples:</p><ol><li>For the following input of shape `[1, 2, 2, 1]`, `block_shape = [2, 2]`, and
+     `paddings = [[0, 0], [0, 0]]`:</li></ol><p>```prettyprint
+ x = [[[[1], [2]], [[3], [4]]]]
+ ```</p><p>The output tensor has shape `[4, 1, 1, 1]` and value:</p><p>```prettyprint
+ [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
+ ```</p><ol><li>For the following input of shape `[1, 2, 2, 3]`, `block_shape = [2, 2]`, and
+     `paddings = [[0, 0], [0, 0]]`:</li></ol><p>```prettyprint
+ x = [[[[1, 2, 3], [4, 5, 6]],
+       [[7, 8, 9], [10, 11, 12]]]]
+ ```</p><p>The output tensor has shape `[4, 1, 1, 3]` and value:</p><p>```prettyprint
+ [[[1, 2, 3]], [[4, 5, 6]], [[7, 8, 9]], [[10, 11, 12]]]
+ ```</p><ol><li>For the following input of shape `[1, 4, 4, 1]`, `block_shape = [2, 2]`, and
+     `paddings = [[0, 0], [0, 0]]`:</li></ol><p>```prettyprint
+ x = [[[[1],   [2],  [3],  [4]],
+       [[5],   [6],  [7],  [8]],
+       [[9],  [10], [11],  [12]],
+       [[13], [14], [15],  [16]]]]
+ ```</p><p>The output tensor has shape `[4, 2, 2, 1]` and value:</p><p>```prettyprint
+ x = [[[[1], [3]], [[5], [7]]],
+      [[[2], [4]], [[10], [12]]],
+      [[[5], [7]], [[13], [15]]],
+      [[[6], [8]], [[14], [16]]]]
+ ```</p><ol><li>For the following input of shape `[2, 2, 4, 1]`, block_shape = `[2, 2]`, and
+     paddings = `[[0, 0], [2, 0]]`:</li></ol><p>```prettyprint
+ x = [[[[1],   [2],  [3],  [4]],
+       [[5],   [6],  [7],  [8]]],
+      [[[9],  [10], [11],  [12]],
+       [[13], [14], [15],  [16]]]]
+ ```</p><p>The output tensor has shape `[8, 1, 3, 1]` and value:</p><p>```prettyprint
+ x = [[[[0], [1], [3]]], [[[0], [9], [11]]],
+      [[[0], [2], [4]]], [[[0], [10], [12]]],
+      [[[0], [5], [7]]], [[[0], [13], [15]]],
+      [[[0], [6], [8]]], [[[0], [14], [16]]]]
+ ```</p><p>Among others, this operation is useful for reducing atrous convolution into
+ regular convolution.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>SpaceToBatch for N-D tensors of type T.</p><p>This operation divides &quot;spatial&quot; dimensions `[1, ..., M]` of the input into a
+ grid of blocks of shape <code>block_shape</code>, and interleaves these blocks with the
+ &quot;batch&quot; dimension (0) such that in the output, the spatial dimensions
+ `[1, ..., M]` correspond to the position within the grid, and the batch
+ dimension combines both the position within a spatial block and the original
+ batch position.  Prior to division into blocks, the spatial dimensions of the
+ input are optionally zero padded according to <code>paddings</code>.  See below for a
+ precise description.</p></div></div><div class="top"><p class="src"><a name="v:spaceToBatch" class="def">spaceToBatch</a> <a href="src/TensorFlow-GenOps-Core.html#spaceToBatch" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tpaddings, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tpaddings)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>block_size</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: 4-D with shape `[batch, height, width, depth]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tpaddings</td><td class="doc"><p><strong>paddings</strong>: 2-D tensor of non-negative integers with shape `[2, 2]`. It specifies
+   the padding of the input with zeros across the spatial dimensions as follows:</p><p>paddings = [[pad_top, pad_bottom], [pad_left, pad_right]]</p><p>The effective spatial dimensions of the zero-padded input tensor will be:</p><p>height_pad = pad_top + height + pad_bottom
+       width_pad = pad_left + width + pad_right</p><p>The attr <code>block_size</code> must be greater than one. It indicates the block size.</p><ul><li>Non-overlapping blocks of size `block_size x block size` in the height and
+     width dimensions are rearranged into the batch dimension at each location.</li><li>The batch of the output tensor is `batch * block_size * block_size`.</li><li>Both height_pad and width_pad must be divisible by block_size.</li></ul><p>The shape of the output will be:</p><p>[batch*block_size*block_size, height_pad<em>block_size, width_pad</em>block_size,
+      depth]</p><p>Some examples:</p><ol><li>For the following input of shape `[1, 2, 2, 1]` and block_size of 2:</li></ol><p>```prettyprint
+ x = [[[[1], [2]], [[3], [4]]]]
+ ```</p><p>The output tensor has shape `[4, 1, 1, 1]` and value:</p><p>```prettyprint
+ [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
+ ```</p><ol><li>For the following input of shape `[1, 2, 2, 3]` and block_size of 2:</li></ol><p>```prettyprint
+ x = [[[[1, 2, 3], [4, 5, 6]],
+       [[7, 8, 9], [10, 11, 12]]]]
+ ```</p><p>The output tensor has shape `[4, 1, 1, 3]` and value:</p><p>```prettyprint
+ [[[1, 2, 3]], [[4, 5, 6]], [[7, 8, 9]], [[10, 11, 12]]]
+ ```</p><ol><li>For the following input of shape `[1, 4, 4, 1]` and block_size of 2:</li></ol><p>```prettyprint
+ x = [[[[1],   [2],  [3],  [4]],
+       [[5],   [6],  [7],  [8]],
+       [[9],  [10], [11],  [12]],
+       [[13], [14], [15],  [16]]]]
+ ```</p><p>The output tensor has shape `[4, 2, 2, 1]` and value:</p><p>```prettyprint
+ x = [[[[1], [3]], [[5], [7]]],
+      [[[2], [4]], [[10], [12]]],
+      [[[5], [7]], [[13], [15]]],
+      [[[6], [8]], [[14], [16]]]]
+ ```</p><ol><li>For the following input of shape `[2, 2, 4, 1]` and block_size of 2:</li></ol><p>```prettyprint
+ x = [[[[1],   [2],  [3],  [4]],
+       [[5],   [6],  [7],  [8]]],
+      [[[9],  [10], [11],  [12]],
+       [[13], [14], [15],  [16]]]]
+ ```</p><p>The output tensor has shape `[8, 1, 2, 1]` and value:</p><p>```prettyprint
+ x = [[[[1], [3]]], [[[9], [11]]], [[[2], [4]]], [[[10], [12]]],
+      [[[5], [7]]], [[[13], [15]]], [[[6], [8]]], [[[14], [16]]]]
+ ```</p><p>Among others, this operation is useful for reducing atrous convolution into
+ regular convolution.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>SpaceToBatch for 4-D tensors of type T.</p><p>This is a legacy version of the more general SpaceToBatchND.</p><p>Zero-pads and then rearranges (permutes) blocks of spatial data into batch.
+ More specifically, this op outputs a copy of the input tensor where values from
+ the <code>height</code> and <code>width</code> dimensions are moved to the <code>batch</code> dimension. After
+ the zero-padding, both <code>height</code> and <code>width</code> of the input must be divisible by the
+ block size.</p></div></div><div class="top"><p class="src"><a name="v:cTCGreedyDecoder" class="def">cTCGreedyDecoder</a> <a href="src/TensorFlow-GenOps-Core.html#cTCGreedyDecoder" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>inputs</strong>: 3-D, shape: `(max_time x batch_size x num_classes)`, the logits.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>sequence_length</strong>: A vector containing sequence lengths, size `(batch_size)`.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p>(<strong>decoded_indices</strong>, <strong>decoded_values</strong>, <strong>decoded_shape</strong>, <strong>log_probability</strong>)</p><ul><li><strong>decoded_indices</strong>: Indices matrix, size `(total_decoded_outputs x 2)`,
+ of a `SparseTensor<a href="int64,">2</a>`.  The rows store: [batch, time].</li><li><strong>decoded_values</strong>: Values vector, size: `(total_decoded_outputs)`,
+ of a `SparseTensor<a href="int64,">2</a>`.  The vector stores the decoded classes.</li><li><strong>decoded_shape</strong>: Shape vector, size `(2)`, of the decoded SparseTensor.
+ Values are: `[batch_size, max_decoded_length]`.</li><li><strong>log_probability</strong>: Matrix, size `(batch_size x 1)`, containing sequence
+ log-probabilities.</li></ul></td></tr></table></div><div class="doc"><p>Performs greedy decoding on the logits given in inputs.</p><p>A note about the attribute merge_repeated: if enabled, when
+ consecutive logits' maximum indices are the same, only the first of
+ these is emitted.  Labeling the blank <code><a href="../base-4.8.2.0/Prelude.html#v:-42-">*</a></code>, the sequence &quot;A B B * B B&quot;
+ becomes &quot;A B&quot; if merge_repeated = True and &quot;A B B B B&quot; if
+ merge_repeated = False.</p><p>Regardless of the value of merge_repeated, if the maximum index of a given
+ time and batch corresponds to the blank, index `(num_classes - 1)`, no new
+ element is emitted.</p></div></div><div class="top"><p class="src"><a name="v:batchToSpaceND" class="def">batchToSpaceND</a> <a href="src/TensorFlow-GenOps-Core.html#batchToSpaceND" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tblock_shape, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tblock_shape, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tcrops, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tcrops)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: N-D with shape `input_shape = [batch] + spatial_shape + remaining_shape`,
+ where spatial_shape has M dimensions.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tblock_shape</td><td class="doc"><p><strong>block_shape</strong>: 1-D with shape `[M]`, all values must be &gt;= 1.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tcrops</td><td class="doc"><p><strong>crops</strong>: 2-D with shape `[M, 2]`, all values must be &gt;= 0.
+   `crops[i] = [crop_start, crop_end]` specifies the amount to crop from input
+   dimension `i + 1`, which corresponds to spatial dimension <code>i</code>.  It is
+   required that
+   `crop_start[i] + crop_end[i] &lt;= block_shape[i] * input_shape[i + 1]`.</p><p>This operation is equivalent to the following steps:</p><ol><li>Reshape <code>input</code> to <code>reshaped</code> of shape:
+      [block_shape[0], ..., block_shape[M-1],
+       batch / prod(block_shape),
+       input_shape[1], ..., input_shape[N-1]]</li><li>Permute dimensions of <code>reshaped</code> to produce <code>permuted</code> of shape
+      [batch / prod(block_shape),</li></ol><p>input_shape[1], block_shape[0],
+       ...,
+       input_shape[M], block_shape[M-1],</p><p>input_shape[M+1], ..., input_shape[N-1]]</p><ol><li>Reshape <code>permuted</code> to produce <code>reshaped_permuted</code> of shape
+      [batch / prod(block_shape),</li></ol><p>input_shape[1] * block_shape[0],
+       ...,
+       input_shape[M] * block_shape[M-1],</p><p>input_shape[M+1],
+       ...,
+       input_shape[N-1]]</p><ol><li>Crop the start and end of dimensions `[1, ..., M]` of
+    <code>reshaped_permuted</code> according to <code>crops</code> to produce the output of shape:
+      [batch / prod(block_shape),</li></ol><p>input_shape[1] * block_shape[0] - crops[0,0] - crops[0,1],
+       ...,
+       input_shape[M] * block_shape[M-1] - crops[M-1,0] - crops[M-1,1],</p><p>input_shape[M+1], ..., input_shape[N-1]]</p><p>Some examples:</p><ol><li>For the following input of shape `[4, 1, 1, 1]`, `block_shape = [2, 2]`, and
+     `crops = [[0, 0], [0, 0]]`:</li></ol><p>```prettyprint
+ [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
+ ```</p><p>The output tensor has shape `[1, 2, 2, 1]` and value:</p><p>```prettyprint
+ x = [[[[1], [2]], [[3], [4]]]]
+ ```</p><ol><li>For the following input of shape `[4, 1, 1, 3]`, `block_shape = [2, 2]`, and
+     `crops = [[0, 0], [0, 0]]`:</li></ol><p>```prettyprint
+ [[[1, 2, 3]], [[4, 5, 6]], [[7, 8, 9]], [[10, 11, 12]]]
+ ```</p><p>The output tensor has shape `[1, 2, 2, 3]` and value:</p><p>```prettyprint
+ x = [[[[1, 2, 3], [4, 5, 6]],
+       [[7, 8, 9], [10, 11, 12]]]]
+ ```</p><ol><li>For the following input of shape `[4, 2, 2, 1]`, `block_shape = [2, 2]`, and
+     `crops = [[0, 0], [0, 0]]`:</li></ol><p>```prettyprint
+ x = [[[[1], [3]], [[5], [7]]],
+      [[[2], [4]], [[10], [12]]],
+      [[[5], [7]], [[13], [15]]],
+      [[[6], [8]], [[14], [16]]]]
+ ```</p><p>The output tensor has shape `[1, 4, 4, 1]` and value:</p><p>```prettyprint
+ x = [[[1],   [2],  [3],  [4]],
+      [[5],   [6],  [7],  [8]],
+      [[9],  [10], [11],  [12]],
+      [[13], [14], [15],  [16]]]
+ ```</p><ol><li>For the following input of shape `[8, 1, 3, 1]`, `block_shape = [2, 2]`, and
+     `crops = [[0, 0], [2, 0]]`:</li></ol><p>```prettyprint
+ x = [[[[0], [1], [3]]], [[[0], [9], [11]]],
+      [[[0], [2], [4]]], [[[0], [10], [12]]],
+      [[[0], [5], [7]]], [[[0], [13], [15]]],
+      [[[0], [6], [8]]], [[[0], [14], [16]]]]
+ ```</p><p>The output tensor has shape `[2, 2, 4, 1]` and value:</p><p>```prettyprint
+ x = [[[[1],   [2],  [3],  [4]],
+       [[5],   [6],  [7],  [8]]],
+      [[[9],  [10], [11],  [12]],
+       [[13], [14], [15],  [16]]]]
+ ```</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>BatchToSpace for N-D tensors of type T.</p><p>This operation reshapes the &quot;batch&quot; dimension 0 into `M + 1` dimensions of shape
+ `block_shape + [batch]`, interleaves these blocks back into the grid defined by
+ the spatial dimensions `[1, ..., M]`, to obtain a result with the same rank as
+ the input.  The spatial dimensions of this intermediate result are then
+ optionally cropped according to <code>crops</code> to produce the output.  This is the
+ reverse of SpaceToBatch.  See below for a precise description.</p></div></div><div class="top"><p class="src"><a name="v:pack" class="def">pack</a> <a href="src/TensorFlow-GenOps-Core.html#pack" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t]</td><td class="doc"><p><strong>values</strong>: Must be of same shape and type.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The packed tensor.</p></td></tr></table></div><div class="doc"><p>Packs a list of <code>N</code> rank-<code>R</code> tensors into one rank-`(R+1)` tensor.</p><p>Packs the <code>N</code> tensors in <code>values</code> into a tensor with rank one higher than each
+ tensor in <code>values</code>, by packing them along the <code>axis</code> dimension.
+ Given a list of tensors of shape `(A, B, C)`;</p><p>if `axis == 0` then the <code>output</code> tensor will have the shape `(N, A, B, C)`.
+ if `axis == 1` then the <code>output</code> tensor will have the shape `(A, N, B, C)`.
+ Etc.</p><p>For example:</p><p>```prettyprint
+ # <code>x</code> is [1, 4]
+ # <code>y</code> is [2, 5]
+ # <code>z</code> is [3, 6]
+ pack([x, y, z]) =&gt; [[1, 4], [2, 5], [3, 6]]  # Pack along first dim.
+ pack([x, y, z], axis=1) =&gt; [[1, 2, 3], [4, 5, 6]]
+ ```</p><p>This is the opposite of <code><a href="TensorFlow-GenOps-Core.html#v:unpack">unpack</a></code>.</p></div></div><div class="top"><p class="src"><a name="v:oneHot" class="def">oneHot</a> <a href="src/TensorFlow-GenOps-Core.html#oneHot" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tI, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>]` tI)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tI</td><td class="doc"><p><strong>indices</strong>: A tensor of indices.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>depth</strong>: A scalar defining the depth of the one hot dimension.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>on_value</strong>: A scalar defining the value to fill in output when `indices[j] = i`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>off_value</strong>: A scalar defining the value to fill in output when `indices[j] != i`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The one-hot tensor.</p></td></tr></table></div><div class="doc"><p>Returns a one-hot tensor.</p><p>The locations represented by indices in <code>indices</code> take value <code>on_value</code>,
+ while all other locations take value <code>off_value</code>.</p><p>If the input <code>indices</code> is rank <code>N</code>, the output will have rank `N+1`,
+ The new axis is created at dimension <code>axis</code> (default: the new axis is
+ appended at the end).</p><p>If <code>indices</code> is a scalar the output shape will be a vector of length <code>depth</code>.</p><p>If <code>indices</code> is a vector of length <code>features</code>, the output shape will be:
+ ```
+   features x depth if axis == -1
+   depth x features if axis == 0
+ ```</p><p>If <code>indices</code> is a matrix (batch) with shape `[batch, features]`,
+ the output shape will be:
+ ```
+   batch x features x depth if axis == -1
+   batch x depth x features if axis == 1
+   depth x batch x features if axis == 0
+ ```</p><p>Examples
+ =========</p><p>Suppose that</p><p>```
+   indices = [0, 2, -1, 1]
+   depth = 3
+   on_value = 5.0
+   off_value = 0.0
+   axis = -1
+ ```</p><p>Then output is `[4 x 3]`:</p><p>```output =
+       [5.0 0.0 0.0]  // one_hot(0)
+       [0.0 0.0 5.0]  // one_hot(2)
+       [0.0 0.0 0.0]  // one_hot(-1)
+       [0.0 5.0 0.0]  // one_hot(1)
+     ```</p><p>Suppose that</p><p>```
+   indices = [0, 2, -1, 1]
+   depth = 3
+   on_value = 0.0
+   off_value = 3.0
+   axis = 0
+ ```</p><p>Then output is `[3 x 4]`:</p><p>```output =
+       [0.0 3.0 3.0 3.0]
+       [3.0 3.0 3.0 0.0]
+       [3.0 3.0 3.0 3.0]
+       [3.0 0.0 3.0 3.0]
+     //  ^                one_hot(0)
+     //      ^            one_hot(2)
+     //          ^        one_hot(-1)
+     //              ^    one_hot(1)
+     ```
+ Suppose that</p><p>```
+   indices = [[0, 2], [1, -1]]
+   depth = 3
+   on_value = 1.0
+   off_value = 0.0
+   axis = -1
+ ```</p><p>Then output is `[2 x 2 x 3]`:</p><p>```output =
+       [
+         [1.0, 0.0, 0.0]  // one_hot(0)
+         [0.0, 0.0, 1.0]  // one_hot(2)
+       ][
+         [0.0, 1.0, 0.0]  // one_hot(1)
+         [0.0, 0.0, 0.0]  // one_hot(-1)
+       ]```</p></div></div><div class="top"><p class="src"><a name="v:broadcastGradientArgs" class="def">broadcastGradientArgs</a> <a href="src/TensorFlow-GenOps-Core.html#broadcastGradientArgs" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>s0</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>s1</strong></p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><td class="doc"><p>(<strong>r0</strong>, <strong>r1</strong>)</p><ul><li><strong>r0</strong></li><li><strong>r1</strong></li></ul></td></tr></table></div><div class="doc"><p>Return the reduction indices for computing gradients of s0 op s1 with broadcast.</p><p>This is typically used by gradient computations for a broadcasting operation.</p></div></div><div class="top"><p class="src"><a name="v:matrixSetDiag" class="def">matrixSetDiag</a> <a href="src/TensorFlow-GenOps-Core.html#matrixSetDiag" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Rank `k+1`, where `k &gt;= 1`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>diagonal</strong>: Rank <code>k</code>, where `k &gt;= 1`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Rank `k+1`, with `output.shape = input.shape`.</p></td></tr></table></div><div class="doc"><p>Returns a batched matrix tensor with new batched diagonal values.</p><p>Given <code>input</code> and <code>diagonal</code>, this operation returns a tensor with the
+ same shape and values as <code>input</code>, except for the diagonals of the innermost
+ matrices.  These will be overwritten by the values in <code>diagonal</code>.
+ The batched matrices must be square.</p><p>The output is computed as follows:</p><p>Assume <code>input</code> has `k+1` dimensions `[I, J, K, ..., N, N]` and <code>diagonal</code> has
+ <code>k</code> dimensions `[I, J, K, ..., N]`.  Then the output is a
+ tensor of rank `k+1` with dimensions [I, J, K, ..., N, N]` where:</p><ul><li>`output[i, j, k, ..., m, n] = diagonal[i, j, k, ..., n]` for `m == n`.</li><li>`output[i, j, k, ..., m, n] = input[i, j, k, ..., m, n]` for `m != n`.</li></ul></div></div><div class="top"><p class="src"><a name="v:applyRMSProp" class="def">applyRMSProp</a> <a href="src/TensorFlow-GenOps-Core.html#applyRMSProp" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>ms</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>mom</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>lr</strong>: Scaling factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>rho</strong>: Decay rate. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t</td><td class="doc"><p><strong>momentum</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t</td><td class="doc"><p><strong>epsilon</strong>: Ridge term. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v8 t</td><td class="doc"><p><strong>grad</strong>: The gradient.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Update '*var' according to the RMSProp algorithm.</p><p>Note that in dense implement of this algorithm, ms and mom will
+ update even if the grad is zero, but in this sparse implement, ms
+ and mom will not update in iterations the grad is zero.</p><p>mean_square = decay * mean_square + (1-decay) * gradient ** 2
+ Delta = learning_rate * gradient / sqrt(mean_square + epsilon)</p><p>ms &lt;- rho * ms_{t-1} + (1-rho) * grad * grad
+ mom &lt;- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
+ var &lt;- var - mom</p></div></div><div class="top"><p class="src"><a name="v:const" class="def">const</a> <a href="src/TensorFlow-GenOps-Core.html#const" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Returns a constant tensor.</p></div></div><div class="top"><p class="src"><a name="v:enter" class="def">enter</a> <a href="src/TensorFlow-GenOps-Core.html#enter" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong>: The tensor to be made available to the child frame.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The same tensor as `data`.</p></td></tr></table></div><div class="doc"><p>Creates or finds a child frame, and makes `data` available to the child frame.</p><p>This op is used together with <code>Exit</code> to create loops in the graph.
+ The unique <code>frame_name</code> is used by the <code>Executor</code> to identify frames. If
+ <code>is_constant</code> is true, <code>output</code> is a constant in the child frame; otherwise
+ it may be changed in the child frame. At most <code>parallel_iterations</code> iterations
+ are run in parallel in the child frame.</p></div></div><div class="top"><p class="src"><a name="v:debugIdentity" class="def">debugIdentity</a> <a href="src/TensorFlow-GenOps-Core.html#debugIdentity" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Input tensor, non-Reference type.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Output tensor that equals the input tensor.</p></td></tr></table></div><div class="doc"><p>Debug Identity Op.</p><p>Provides an identity mapping of the non-Ref type input tensor for debugging.</p></div></div><div class="top"><p class="src"><a name="v:debugNanCount" class="def">debugNanCount</a> <a href="src/TensorFlow-GenOps-Core.html#debugNanCount" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Input tensor, non-Reference type.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>output</strong>: An integer output tensor that is the number of NaNs in the input.</p></td></tr></table></div><div class="doc"><p>Debug NaN Value Counter Op</p><p>Counts number of NaNs in the input tensor, for debugging.</p></div></div><div class="top"><p class="src"><a name="v:batchNormWithGlobalNormalization" class="def">batchNormWithGlobalNormalization</a> <a href="src/TensorFlow-GenOps-Core.html#batchNormWithGlobalNormalization" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>scale_after_normalization</strong>: A bool indicating whether the resulted tensor
+ needs to be multiplied with gamma.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>variance_epsilon</strong>: A small float number to avoid dividing by 0.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>t</strong>: A 4D input Tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>m</strong>: A 1D mean Tensor with size matching the last dimension of t.
+ This is the first output from tf.nn.moments,
+ or a saved moving average thereof.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>v</strong>: A 1D variance Tensor with size matching the last dimension of t.
+ This is the second output from tf.nn.moments,
+ or a saved moving average thereof.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>beta</strong>: A 1D beta Tensor with size matching the last dimension of t.
+ An offset to be added to the normalized tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>gamma</strong>: A 1D gamma Tensor with size matching the last dimension of t.
+ If &quot;scale_after_normalization&quot; is true, this tensor will be multiplied
+ with the normalized tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>result</strong></p></td></tr></table></div><div class="doc"><p>Batch normalization.</p><p>This op is deprecated. Prefer `tf.nn.batch_normalization`.</p></div></div><div class="top"><p class="src"><a name="v:batchMatrixDiag" class="def">batchMatrixDiag</a> <a href="src/TensorFlow-GenOps-Core.html#batchMatrixDiag" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>diagonal</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:unpack" class="def">unpack</a> <a href="src/TensorFlow-GenOps-Core.html#unpack" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>value</strong>: 1-D or higher, with <code>axis</code> dimension size equal to <code>num</code>.</p></td></tr><tr><td class="src">-&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t]</td><td class="doc"><p><strong>output</strong>: The list of tensors unpacked from <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>.</p></td></tr></table></div><div class="doc"><p>Unpacks a given dimension of a rank-<code>R</code> tensor into <code>num</code> rank-`(R-1)` tensors.</p><p>Unpacks <code>num</code> tensors from <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code> by chipping it along the <code>axis</code> dimension.
+ For example, given a tensor of shape `(A, B, C, D)`;</p><p>If `axis == 0` then the i'th tensor in <code>output</code> is the slice `value[i, :, :, :]`
+   and each tensor in <code>output</code> will have shape `(B, C, D)`. (Note that the
+   dimension unpacked along is gone, unlike <code><a href="TensorFlow-GenOps-Core.html#v:split">split</a></code>).</p><p>If `axis == 1` then the i'th tensor in <code>output</code> is the slice `value[:, i, :, :]`
+   and each tensor in <code>output</code> will have shape `(A, C, D)`.
+ Etc.</p><p>This is the opposite of <code><a href="TensorFlow-GenOps-Core.html#v:pack">pack</a></code>.</p></div></div><div class="top"><p class="src"><a name="v:sparseSplit" class="def">sparseSplit</a> <a href="src/TensorFlow-GenOps-Core.html#sparseSplit" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_split</strong>: The number of ways to split.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>split_dim</strong>: 0-D.  The dimension along which to split.  Must be in the range
+ `[0, rank(shape))`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>indices</strong>: 2-D tensor represents the indices of the sparse tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>values</strong>: 1-D tensor represents the values of the sparse tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>shape</strong>: 1-D. tensor represents the shape of the sparse tensor.
+ output indices: A list of 1-D tensors represents the indices of the output
+ sparse tensors.</p></td></tr><tr><td class="src">-&gt; ([<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>], [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t], [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>])</td><td class="doc"><p>(<strong>output_indices</strong>, <strong>output_values</strong>, <strong>output_shape</strong>)</p><ul><li><strong>output_indices</strong></li><li><strong>output_values</strong>: A list of 1-D tensors represents the values of the output sparse
+ tensors.</li><li><strong>output_shape</strong>: A list of 1-D tensors represents the shape of the output sparse
+ tensors.</li></ul></td></tr></table></div><div class="doc"><p>Split a <code>SparseTensor</code> into <code>num_split</code> tensors along one dimension.</p><p>If the `shape[split_dim]` is not an integer multiple of <code>num_split</code>. Slices
+ `[0 : shape[split_dim] % num_split]` gets one extra dimension.
+ For example, if `split_dim = 1` and `num_split = 2` and the input is</p><p>input_tensor = shape = [2, 7]
+     [    a   d e  ]
+     [b c          ]</p><p>Graphically the output tensors are:</p><p>output_tensor[0] = shape = [2, 4]
+     [    a  ]
+     [b c    ]</p><p>output_tensor[1] = shape = [2, 3]
+     [ d e  ]
+     [      ]</p></div></div><div class="top"><p class="src"><a name="v:mirrorPad" class="def">mirrorPad</a> <a href="src/TensorFlow-GenOps-Core.html#mirrorPad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tpaddings, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tpaddings)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: The input tensor to be padded.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tpaddings</td><td class="doc"><p><strong>paddings</strong>: A two-column matrix specifying the padding sizes. The number of
+ rows must be the same as the rank of <code>input</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The padded tensor.</p></td></tr></table></div><div class="doc"><p>Pads a tensor with mirrored values.</p><p>This operation pads a <code>input</code> with mirrored values according to the <code>paddings</code>
+ you specify. <code>paddings</code> is an integer tensor with shape `[n, 2]`, where n is
+ the rank of <code>input</code>. For each dimension D of <code>input</code>, `paddings[D, 0]` indicates
+ how many values to add before the contents of <code>input</code> in that dimension, and
+ `paddings[D, 1]` indicates how many values to add after the contents of <code>input</code>
+ in that dimension. Both `paddings[D, 0]` and `paddings[D, 1]` must be no greater
+ than `input.dim_size(D)` (or `input.dim_size(D) - 1`) if <code>copy_border</code> is true
+ (if false, respectively).</p><p>The padded size of each dimension D of the output is:</p><p>`paddings(D, 0) + input.dim_size(D) + paddings(D, 1)`</p><p>For example:</p><p>```prettyprint
+ # <code>t</code> is [[1, 2, 3], [4, 5, 6]].
+ # <code>paddings</code> is [[1, 1]], [2, 2]].
+ # <code>mode</code> is SYMMETRIC.
+ # rank of <code>t</code> is 2.
+ pad(t, paddings) ==&gt; [[2, 1, 1, 2, 3, 3, 2]
+                       [2, 1, 1, 2, 3, 3, 2]
+                       [5, 4, 4, 5, 6, 6, 5]
+                       [5, 4, 4, 5, 6, 6, 5]]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:batchMatrixDiagPart" class="def">batchMatrixDiagPart</a> <a href="src/TensorFlow-GenOps-Core.html#batchMatrixDiagPart" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>diagonal</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:fractionalMaxPoolGrad" class="def">fractionalMaxPoolGrad</a> <a href="src/TensorFlow-GenOps-Core.html#fractionalMaxPoolGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>orig_input</strong>: Original input for <code>fractional_max_pool</code></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>orig_output</strong>: Original output for <code>fractional_max_pool</code></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>out_backprop</strong>: 4-D with shape `[batch, height, width, channels]`.  Gradients
+ w.r.t. the output of <code>fractional_max_pool</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>row_pooling_sequence</strong>: row pooling sequence, form pooling region with
+ col_pooling_sequence.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>col_pooling_sequence</strong>: column pooling sequence, form pooling region with
+ row_pooling sequence.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 4-D.  Gradients w.r.t. the input of <code>fractional_max_pool</code>.</p></td></tr></table></div><div class="doc"><p>Computes gradient of the FractionalMaxPool function.</p></div></div><div class="top"><p class="src"><a name="v:matchingFiles" class="def">matchingFiles</a> <a href="src/TensorFlow-GenOps-Core.html#matchingFiles" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>pattern</strong>: A (scalar) shell wildcard pattern.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>filenames</strong>: A vector of matching filenames.</p></td></tr></table></div><div class="doc"><p>Returns the set of files matching a pattern.</p><p>Note that this routine only supports wildcard characters in the
+ basename portion of the pattern, not in the directory portion.</p></div></div><div class="top"><p class="src"><a name="v:tile" class="def">tile</a> <a href="src/TensorFlow-GenOps-Core.html#tile" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tmultiples, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tmultiples)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: 1-D or higher.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tmultiples</td><td class="doc"><p><strong>multiples</strong>: 1-D. Length must be the same as the number of dimensions in <code>input</code></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Constructs a tensor by tiling a given tensor.</p><p>This operation creates a new tensor by replicating <code>input</code> <code>multiples</code> times.
+ The output tensor's i'th dimension has `input.dims(i) * multiples[i]` elements,
+ and the values of <code>input</code> are replicated `multiples[i]` times along the <code>i</code>th
+ dimension. For example, tiling `[a b c d]` by `[2]` produces
+ `[a b c d a b c d]`.</p></div></div><div class="top"><p class="src"><a name="v:sparseSparseMinimum" class="def">sparseSparseMinimum</a> <a href="src/TensorFlow-GenOps-Core.html#sparseSparseMinimum" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>a_indices</strong>: 2-D.  `N x R` matrix with the indices of non-empty values in a
+ SparseTensor, in the canonical lexicographic ordering.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>a_values</strong>: 1-D.  <code>N</code> non-empty values corresponding to <code>a_indices</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>a_shape</strong>: 1-D.  Shape of the input SparseTensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>b_indices</strong>: counterpart to <code>a_indices</code> for the other operand.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>b_values</strong>: counterpart to <code>a_values</code> for the other operand; must be of the same dtype.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>b_shape</strong>: counterpart to <code>a_shape</code> for the other operand; the two shapes must be equal.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><td class="doc"><p>(<strong>output_indices</strong>, <strong>output_values</strong>)</p><ul><li><strong>output_indices</strong>: 2-D.  The indices of the output SparseTensor.</li><li><strong>output_values</strong>: 1-D.  The values of the output SparseTensor.</li></ul></td></tr></table></div><div class="doc"><p>Returns the element-wise min of two SparseTensors.</p><p>Assumes the two SparseTensors have the same shape, i.e., no broadcasting.</p></div></div><div class="top"><p class="src"><a name="v:allCandidateSampler" class="def">allCandidateSampler</a> <a href="src/TensorFlow-GenOps-Core.html#allCandidateSampler" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_sampled</strong>: Number of candidates to produce per batch.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_true</strong>: Number of true labels per context.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>unique</strong>: If unique is true, we sample with rejection, so that all sampled
+ candidates in a batch are unique. This requires some approximation to
+ estimate the post-rejection sampling probabilities.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>true_classes</strong>: A batch_size * num_true matrix, in which each row contains the
+ IDs of the num_true target_classes in the corresponding original label.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p>(<strong>sampled_candidates</strong>, <strong>true_expected_count</strong>, <strong>sampled_expected_count</strong>)</p><ul><li><strong>sampled_candidates</strong>: A vector of length num_sampled, in which each element is
+ the ID of a sampled candidate.</li><li><strong>true_expected_count</strong>: A batch_size * num_true matrix, representing
+ the number of times each candidate is expected to occur in a batch
+ of sampled candidates. If unique=true, then this is a probability.</li><li><strong>sampled_expected_count</strong>: A vector of length num_sampled, for each sampled
+ candidate representing the number of times the candidate is expected
+ to occur in a batch of sampled candidates.  If unique=true, then this is a
+ probability.</li></ul></td></tr></table></div><div class="doc"><p>Generates labels for candidate sampling with a learned unigram distribution.</p><p>See explanations of candidate sampling and the data formats at
+ go/candidate-sampling.</p><p>For each batch, this op picks a single set of sampled candidate labels.</p><p>The advantages of sampling candidates per-batch are simplicity and the
+ possibility of efficient dense matrix multiplication. The disadvantage is that
+ the sampled candidates must be chosen independently of the context and of the
+ true labels.</p></div></div><div class="top"><p class="src"><a name="v:refSwitch" class="def">refSwitch</a> <a href="src/TensorFlow-GenOps-Core.html#refSwitch" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong>: The ref tensor to be forwarded to the appropriate output.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>pred</strong>: A scalar that specifies which output port will receive data.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><td class="doc"><p>(<strong>output_false</strong>, <strong>output_true</strong>)</p><ul><li><strong>output_false</strong>: If <code><a href="../base-4.8.2.0/Prelude.html#v:pred">pred</a></code> is false, data will be forwarded to this output.</li><li><strong>output_true</strong>: If <code><a href="../base-4.8.2.0/Prelude.html#v:pred">pred</a></code> is true, data will be forwarded to this output.</li></ul></td></tr></table></div><div class="doc"><p>Forwards the ref tensor `data` to the output port determined by <code><a href="../base-4.8.2.0/Prelude.html#v:pred">pred</a></code>.</p><p>If <code><a href="../base-4.8.2.0/Prelude.html#v:pred">pred</a></code> is true, the `data` input is forwarded to <code>output_true</code>. Otherwise,
+ the data goes to <code>output_false</code>.</p><p>See also <code>Switch</code> and <code>Merge</code>.</p></div></div><div class="top"><p class="src"><a name="v:mergeSummary" class="def">mergeSummary</a> <a href="src/TensorFlow-GenOps-Core.html#mergeSummary" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>]</td><td class="doc"><p><strong>inputs</strong>: Can be of any shape.  Each must contain serialized <code>Summary</code> protocol
+ buffers.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>summary</strong>: Scalar. Serialized <code>Summary</code> protocol buffer.</p></td></tr></table></div><div class="doc"><p>Merges summaries.</p><p>This op creates a
+ <a href="https://www.tensorflow.org/code/tensorflow/core/framework/summary.proto">`Summary`</a>
+ protocol buffer that contains the union of all the values in the input
+ summaries.</p><p>When the Op is run, it reports an <code>InvalidArgument</code> error if multiple values
+ in the summaries to merge use the same tag.</p></div></div><div class="top"><p class="src"><a name="v:logicalNot" class="def">logicalNot</a> <a href="src/TensorFlow-GenOps-Core.html#logicalNot" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Returns the truth value of NOT x element-wise.</p></div></div><div class="top"><p class="src"><a name="v:lRNGrad" class="def">lRNGrad</a> <a href="src/TensorFlow-GenOps-Core.html#lRNGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input_grads</strong>: 4-D with shape `[batch, height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>input_image</strong>: 4-D with shape `[batch, height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>output_image</strong>: 4-D with shape `[batch, height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The gradients for LRN.</p></td></tr></table></div><div class="doc"><p>Gradients for Local Response Normalization.</p></div></div><div class="top"><p class="src"><a name="v:stringToNumber" class="def">stringToNumber</a> <a href="src/TensorFlow-GenOps-Core.html#stringToNumber" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_type, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` out_type)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>string_tensor</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_type</td><td class="doc"><p><strong>output</strong>: A Tensor of the same shape as the input <code>string_tensor</code>.</p></td></tr></table></div><div class="doc"><p>Converts each string in the input Tensor to the specified numeric type.</p><p>(Note that int32 overflow results in an error while float overflow
+ results in a rounded value.)</p></div></div><div class="top"><p class="src"><a name="v:sparseMatMul" class="def">sparseMatMul</a> <a href="src/TensorFlow-GenOps-Core.html#sparseMatMul" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> ta, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` ta, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tb, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` tb)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 ta</td><td class="doc"><p><strong>a</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tb</td><td class="doc"><p><strong>b</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>product</strong></p></td></tr></table></div><div class="doc"><p>Multiply matrix &quot;a&quot; by matrix &quot;b&quot;.</p><p>The inputs must be two-dimensional matrices and the inner dimension of &quot;a&quot; must
+ match the outer dimension of &quot;b&quot;. This op is optimized for the case where at
+ least one of &quot;a&quot; or &quot;b&quot; is sparse. The breakeven for using this versus a dense
+ matrix multiply on one platform was 30% zero values in the sparse matrix.</p></div></div><div class="top"><p class="src"><a name="v:merge" class="def">merge</a> <a href="src/TensorFlow-GenOps-Core.html#merge" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t]</td><td class="doc"><p><strong>inputs</strong>: The input tensors, exactly one of which will become available.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>)</td><td class="doc"><p>(<strong>output</strong>, <strong>value_index</strong>)</p><ul><li><strong>output</strong>: Will be set to the available input tensor.</li><li><strong>value_index</strong>: The index of the chosen input tensor in <code>inputs</code>.</li></ul></td></tr></table></div><div class="doc"><p>Forwards the value of an available tensor from <code>inputs</code> to <code>output</code>.</p><p><code>Merge</code> waits for at least one of the tensors in <code>inputs</code> to become available.
+ It is usually combined with <code>Switch</code> to implement branching.</p><p><code>Merge</code> forwards the first tensor for become available to <code>output</code>, and sets
+ <code>value_index</code> to its index in <code>inputs</code>.</p></div></div><div class="top"><p class="src"><a name="v:choleskyGrad" class="def">choleskyGrad</a> <a href="src/TensorFlow-GenOps-Core.html#choleskyGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>l</strong>: Output of batch Cholesky algorithm l = cholesky(A). Shape is `[..., M, M]`.
+ Algorithm depends only on lower triangular part of the innermost matrices of
+ this tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>grad</strong>: df/dl where f is some scalar function. Shape is `[..., M, M]`.
+ Algorithm depends only on lower triangular part of the innermost matrices of
+ this tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Symmetrized version of df/dA . Shape is `[..., M, M]`</p></td></tr></table></div><div class="doc"><p>Computes the reverse mode backpropagated gradient of the Cholesky algorithm.</p><p>For an explanation see &quot;Differentiation of the Cholesky algorithm&quot; by
+ Iain Murray <a href="http://arxiv.org/abs/1602.07527">http://arxiv.org/abs/1602.07527</a>.</p></div></div><div class="top"><p class="src"><a name="v:batchCholeskyGrad" class="def">batchCholeskyGrad</a> <a href="src/TensorFlow-GenOps-Core.html#batchCholeskyGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>l</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>grad</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:tensorArrayGather" class="def">tensorArrayGather</a> <a href="src/TensorFlow-GenOps-Core.html#tensorArrayGather" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a TensorArray.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>indices</strong>: The locations in the TensorArray from which to read tensor elements.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>flow_in</strong>: A float scalar that enforces proper chaining of operations.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</td><td class="doc"><p><strong>value</strong>: All of the elements in the TensorArray, concatenated along a new
+ axis (the new dimension 0).</p></td></tr></table></div><div class="doc"><p>Gather specific elements from the TensorArray into output <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>.</p><p>All elements selected by <code>indices</code> must have the same shape.</p></div></div><div class="top"><p class="src"><a name="v:resizeNearestNeighbor" class="def">resizeNearestNeighbor</a> <a href="src/TensorFlow-GenOps-Core.html#resizeNearestNeighbor" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>images</strong>: 4-D with shape `[batch, height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>size</strong>: = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
+ new size for the images.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>resized_images</strong>: 4-D with shape
+ `[batch, new_height, new_width, channels]`.</p></td></tr></table></div><div class="doc"><p>Resize <code>images</code> to <code><a href="TensorFlow-GenOps-Core.html#v:size">size</a></code> using nearest neighbor interpolation.</p></div></div><div class="top"><p class="src"><a name="v:negTrain" class="def">negTrain</a> <a href="src/TensorFlow-GenOps-Core.html#negTrain" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_negative_samples</strong>: Number of negative samples per example.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>w_in</strong>: input word embedding.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>w_out</strong>: output word embedding.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>examples</strong>: A vector of word ids.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>labels</strong>: A vector of word ids.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>lr</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Training via negative sampling.</p></div></div><div class="top"><p class="src"><a name="v:tensorArrayGrad" class="def">tensorArrayGrad</a> <a href="src/TensorFlow-GenOps-Core.html#tensorArrayGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to the forward TensorArray.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>flow_in</strong>: A float scalar that enforces proper chaining of operations.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>grad_handle</strong></p></td></tr></table></div><div class="doc"><p>Creates a TensorArray for storing the gradients of values in the given handle.</p><p>If the given TensorArray gradient already exists, returns a reference to it.</p><p>Locks the size of the original TensorArray by disabling its dynamic size flag.</p><ul><li>*A note about the input flow_in:**</li></ul><p>The handle flow_in forces the execution of the gradient lookup to occur
+ only after certain other operations have occurred.  For example, when
+ the forward TensorArray is dynamically sized, writes to this TensorArray
+ may resize the object.  The gradient TensorArray is statically sized based
+ on the size of the forward TensorArray when this operation executes.
+ Furthermore, the size of the forward TensorArray is frozen by this call.
+ As a result, the flow is used to ensure that the call to generate the gradient
+ TensorArray only happens after all writes are executed.</p><p>In the case of dynamically sized TensorArrays, gradient computation should
+ only be performed on read operations that have themselves been chained via
+ flow to occur only after all writes have executed. That way the final size
+ of the forward TensorArray is known when this operation is called.</p><ul><li>*A note about the source attribute:**</li></ul><p>TensorArray gradient calls use an accumulator TensorArray object.  If
+ multiple gradients are calculated and run in the same session, the multiple
+ gradient nodes may accidentally flow throuth the same accumulator TensorArray.
+ This double counts and generally breaks the TensorArray gradient flow.</p><p>The solution is to identify which gradient call this particular
+ TensorArray gradient is being called in.  This is performed by identifying
+ a unique string (e.g. &quot;gradients&quot;, &quot;gradients_1&quot;, ...) from the input
+ gradient Tensor's name.  This string is used as a suffix when creating
+ the TensorArray gradient object here (the attribute <code>source</code>).</p><p>The attribute <code>source</code> is added as a suffix to the forward TensorArray's
+ name when performing the creation / lookup, so that each separate gradient
+ calculation gets its own TensorArray accumulator.</p></div></div><div class="top"><p class="src"><a name="v:audioSummary" class="def">audioSummary</a> <a href="src/TensorFlow-GenOps-Core.html#audioSummary" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>sample_rate</strong>: The sample rate of the signal in hertz.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>tag</strong>: Scalar. Used to build the <code>tag</code> attribute of the summary values.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>tensor</strong>: 2-D of shape `[batch_size, frames]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>summary</strong>: Scalar. Serialized <code>Summary</code> protocol buffer.</p></td></tr></table></div><div class="doc"><p>Outputs a <code>Summary</code> protocol buffer with audio.</p><p>The summary has up to <code>max_outputs</code> summary values containing audio. The
+ audio is built from <code>tensor</code> which must be 3-D with shape `[batch_size,
+ frames, channels]` or 2-D with shape `[batch_size, frames]`. The values are
+ assumed to be in the range of `[-1.0, 1.0]` with a sample rate of <code>sample_rate</code>.</p><p>The <code>tag</code> argument is a scalar <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> of type <code>string</code>.  It is used to
+ build the <code>tag</code> of the summary values:</p><ul><li>If <code>max_outputs</code> is 1, the summary value tag is '*tag*/audio'.</li><li>If <code>max_outputs</code> is greater than 1, the summary value tags are
+    generated sequentially as '*tag*/audio/0', '*tag*/audio/1', etc.</li></ul></div></div><div class="top"><p class="src"><a name="v:noOp" class="def">noOp</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a> <a href="src/TensorFlow-GenOps-Core.html#noOp" class="link">Source</a></p><div class="doc"><p>Does nothing. Only useful as a placeholder for control edges.</p></div></div><div class="top"><p class="src"><a name="v:nextIteration" class="def">nextIteration</a> <a href="src/TensorFlow-GenOps-Core.html#nextIteration" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong>: The tensor to be made available to the next iteration.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The same tensor as `data`.</p></td></tr></table></div><div class="doc"><p>Makes its input available to the next iteration.</p></div></div><div class="top"><p class="src"><a name="v:softplusGrad" class="def">softplusGrad</a> <a href="src/TensorFlow-GenOps-Core.html#softplusGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>gradients</strong>: The backpropagated gradients to the corresponding softplus operation.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>features</strong>: The features passed as input to the corresponding softplus operation.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>backprops</strong>: The gradients: `gradients / (1 + exp(-features))`.</p></td></tr></table></div><div class="doc"><p>Computes softplus gradients for a softplus operation.</p></div></div><div class="top"><p class="src"><a name="v:svd" class="def">svd</a> <a href="src/TensorFlow-GenOps-Core.html#svd" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: A tensor of shape `[..., M, N]` whose inner-most 2 dimensions
+ form matrices of size `[M, N]`. Let <code>P</code> be the minimum of <code>M</code> and <code>N</code>.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><td class="doc"><p>(<strong>s</strong>, <strong>u</strong>, <strong>v</strong>)</p><ul><li><strong>s</strong>: Singular values. Shape is `[..., P]`.</li><li><strong>u</strong>: Left singular vectors. If <code>full_matrices</code> is <code><a href="../base-4.8.2.0/Data-Bool.html#v:False">False</a></code> then shape is
+ `[..., M, M]`; if <code>full_matrices</code> is <code><a href="../base-4.8.2.0/Data-Bool.html#v:True">True</a></code> then shape is
+ `[..., M, P]`. Undefined if <code>compute_uv</code> is <code><a href="../base-4.8.2.0/Data-Bool.html#v:False">False</a></code>.</li><li><strong>v</strong>: Left singular vectors. If <code>full_matrices</code> is <code><a href="../base-4.8.2.0/Data-Bool.html#v:False">False</a></code> then shape is
+ `[..., N, N]`. If <code>full_matrices</code> is <code><a href="../base-4.8.2.0/Data-Bool.html#v:True">True</a></code> then shape is `[..., N, P]`.
+ Undefined if <code>compute_uv</code> is false.</li></ul></td></tr></table></div><div class="doc"><p>Computes the singular value decompositions of one or more matrices.</p><p>Computes the SVD of each inner matrix in <code>input</code> such that
+ `input[..., :, :] = u[..., :, :] * diag(s[..., :, :]) * transpose(v[..., :, :])`</p><p>```prettyprint
+ # a is a tensor containing a batch of matrices.
+ # s is a tensor of singular values for each matrix.
+ # u is the tensor containing of left singular vectors for each matrix.
+ # v is the tensor containing of right singular vectors for each matrix.
+ s, u, v = svd(a)
+ s, _, _ = svd(a, compute_uv=False)
+ ```</p></div></div><div class="top"><p class="src"><a name="v:hSVToRGB" class="def">hSVToRGB</a> <a href="src/TensorFlow-GenOps-Core.html#hSVToRGB" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>images</strong>: 1-D or higher rank. HSV data to convert. Last dimension must be size 3.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: <code>images</code> converted to RGB.</p></td></tr></table></div><div class="doc"><p>Convert one or more images from HSV to RGB.</p><p>Outputs a tensor of the same shape as the <code>images</code> tensor, containing the RGB
+ value of the pixels. The output is only well defined if the value in <code>images</code>
+ are in `[0,1]`.</p><p>See <code>rgb_to_hsv</code> for a description of the HSV encoding.</p></div></div><div class="top"><p class="src"><a name="v:parameterizedTruncatedNormal" class="def">parameterizedTruncatedNormal</a> <a href="src/TensorFlow-GenOps-Core.html#parameterizedTruncatedNormal" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` dtype)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>shape</strong>: The shape of the output tensor. Batches are indexed by the 0th dimension.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 dtype</td><td class="doc"><p><strong>means</strong>: The mean parameter of each batch.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 dtype</td><td class="doc"><p><strong>stdevs</strong>: The standard deviation parameter of each batch. Must be greater than 0.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 dtype</td><td class="doc"><p><strong>minvals</strong>: The minimum cutoff. May be -infinity.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 dtype</td><td class="doc"><p><strong>maxvals</strong>: The maximum cutoff. May be +infinity, and must be more than the minval
+ for each batch.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</td><td class="doc"><p><strong>output</strong>: A matrix of shape num_batches x samples_per_batch, filled with random
+ truncated normal values using the parameters for each row.</p></td></tr></table></div><div class="doc"><p>Outputs random values from a normal distribution. The parameters may each be a</p><p>scalar which applies to the entire output, or a vector of length shape[0] which
+ stores the parameters for each batch.</p></div></div><div class="top"><p class="src"><a name="v:square" class="def">square</a> <a href="src/TensorFlow-GenOps-Core.html#square" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes square of x element-wise.</p><p>I.e., \(y = x * x = x^2\).</p></div></div><div class="top"><p class="src"><a name="v:elu" class="def">elu</a> <a href="src/TensorFlow-GenOps-Core.html#elu" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>features</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>activations</strong></p></td></tr></table></div><div class="doc"><p>Computes exponential linear: `exp(features) - 1` if &lt; 0, <code>features</code> otherwise.</p><p>See <a href="http://arxiv.org/abs/1511.07289">Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)</a></p></div></div><div class="top"><p class="src"><a name="v:lookupTableExport" class="def">lookupTableExport</a> <a href="src/TensorFlow-GenOps-Core.html#lookupTableExport" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tkeys, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tvalues)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>table_handle</strong>: Handle to the table.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tkeys, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tvalues)</td><td class="doc"><p>(<strong>keys</strong>, <strong>values</strong>)</p><ul><li><strong>keys</strong>: Vector of all keys present in the table.</li><li><strong>values</strong>: Tensor of all values in the table. Indexed in parallel with <code>keys</code>.</li></ul></td></tr></table></div><div class="doc"><p>Outputs all keys and values in the table.</p></div></div><div class="top"><p class="src"><a name="v:lookupTableSize" class="def">lookupTableSize</a> <a href="src/TensorFlow-GenOps-Core.html#lookupTableSize" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>table_handle</strong>: Handle to the table.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>size</strong>: Scalar that contains number of elements in the table.</p></td></tr></table></div><div class="doc"><p>Computes the number of elements in the given table.</p></div></div><div class="top"><p class="src"><a name="v:avgPoolGrad" class="def">avgPoolGrad</a> <a href="src/TensorFlow-GenOps-Core.html#avgPoolGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>orig_input_shape</strong>: 1-D.  Shape of the original input to <code>avg_pool</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>grad</strong>: 4-D with shape `[batch, height, width, channels]`.  Gradients w.r.t.
+ the output of <code>avg_pool</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 4-D.  Gradients w.r.t. the input of <code>avg_pool</code>.</p></td></tr></table></div><div class="doc"><p>Computes gradients of the average pooling function.</p></div></div><div class="top"><p class="src"><a name="v:computeAccidentalHits" class="def">computeAccidentalHits</a> <a href="src/TensorFlow-GenOps-Core.html#computeAccidentalHits" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_true</strong>: Number of true labels per context.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>true_classes</strong>: The true_classes output of UnpackSparseLabels.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>sampled_candidates</strong>: The sampled_candidates output of CandidateSampler.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p>(<strong>indices</strong>, <strong>ids</strong>, <strong>weights</strong>)</p><ul><li><strong>indices</strong>: A vector of indices corresponding to rows of true_candidates.</li><li><strong>ids</strong>: A vector of IDs of positions in sampled_candidates that match a true_label
+ for the row with the corresponding index in indices.</li><li><strong>weights</strong>: A vector of the same length as indices and ids, in which each element
+ is -FLOAT_MAX.</li></ul></td></tr></table></div><div class="doc"><p>Computes the ids of the positions in sampled_candidates that match true_labels.</p><p>When doing log-odds NCE, the result of this op should be passed through a
+ SparseToDense op, then added to the logits of the sampled candidates. This has
+ the effect of <code>removing</code> the sampled labels that match the true labels by
+ making the classifier sure that they are sampled labels.</p></div></div><div class="top"><p class="src"><a name="v:cTCLoss" class="def">cTCLoss</a> <a href="src/TensorFlow-GenOps-Core.html#cTCLoss" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>inputs</strong>: 3-D, shape: `(max_time x batch_size x num_classes)`, the logits.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>labels_indices</strong>: The indices of a `SparseTensor<a href="int32,">2</a>`.
+ `labels_indices(i, :) == [b, t]` means `labels_values(i)` stores the id for
+ `(batch b, time t)`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>labels_values</strong>: The values (labels) associated with the given batch and time.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>sequence_length</strong>: A vector containing sequence lengths (batch).</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p>(<strong>loss</strong>, <strong>gradient</strong>)</p><ul><li><strong>loss</strong>: A vector (batch) containing log-probabilities.</li><li><strong>gradient</strong>: The gradient of <code>loss</code>.  3-D, shape:
+ `(max_time x batch_size x num_classes)`.</li></ul></td></tr></table></div><div class="doc"><p>Calculates the CTC Loss (log probability) for each batch entry.  Also calculates</p><p>the gradient.  This class performs the softmax operation for you, so inputs
+ should be e.g. linear projections of outputs by an LSTM.</p></div></div><div class="top"><p class="src"><a name="v:avgPool3D" class="def">avgPool3D</a> <a href="src/TensorFlow-GenOps-Core.html#avgPool3D" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Shape `[batch, depth, rows, cols, channels]` tensor to pool over.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The average pooled output tensor.</p></td></tr></table></div><div class="doc"><p>Performs 3D average pooling on the input.</p></div></div><div class="top"><p class="src"><a name="v:inv" class="def">inv</a> <a href="src/TensorFlow-GenOps-Core.html#inv" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes the reciprocal of x element-wise.</p><p>I.e., \(y = 1 / x\).</p></div></div><div class="top"><p class="src"><a name="v:stackPop" class="def">stackPop</a> <a href="src/TensorFlow-GenOps-Core.html#stackPop" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> elem_type</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a stack.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> elem_type</td><td class="doc"><p><strong>elem</strong>: The tensor that is popped from the top of the stack.</p></td></tr></table></div><div class="doc"><p>Pop the element at the top of the stack.</p></div></div><div class="top"><p class="src"><a name="v:paddingFIFOQueue" class="def">paddingFIFOQueue</a> <a href="src/TensorFlow-GenOps-Core.html#paddingFIFOQueue" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to the queue.</p></td></tr></table></div><div class="doc"><p>A queue that produces elements in first-in first-out order.</p><p>Variable-size shapes are allowed by setting the corresponding shape dimensions
+ to 0 in the shape attr.  In this case DequeueMany will pad up to the maximum
+ size of any given element in the minibatch.  See below for details.</p></div></div><div class="top"><p class="src"><a name="v:batchSelfAdjointEigV2" class="def">batchSelfAdjointEigV2</a> <a href="src/TensorFlow-GenOps-Core.html#batchSelfAdjointEigV2" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><td class="doc"><p>(<strong>e</strong>, <strong>v</strong>)</p><ul><li><strong>e</strong></li><li><strong>v</strong></li></ul></td></tr></table></div></div><div class="top"><p class="src"><a name="v:batchMatrixTriangularSolve" class="def">batchMatrixTriangularSolve</a> <a href="src/TensorFlow-GenOps-Core.html#batchMatrixTriangularSolve" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>matrix</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>rhs</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:batchMatrixSolveLs" class="def">batchMatrixSolveLs</a> <a href="src/TensorFlow-GenOps-Core.html#batchMatrixSolveLs" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>matrix</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>rhs</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a></td><td class="doc"><p><strong>l2_regularizer</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:batchSvd" class="def">batchSvd</a> <a href="src/TensorFlow-GenOps-Core.html#batchSvd" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><td class="doc"><p>(<strong>s</strong>, <strong>u</strong>, <strong>v</strong>)</p><ul><li><strong>s</strong></li><li><strong>u</strong></li><li><strong>v</strong></li></ul></td></tr></table></div></div><div class="top"><p class="src"><a name="v:tensorSummary" class="def">tensorSummary</a> <a href="src/TensorFlow-GenOps-Core.html#tensorSummary" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>tensor</strong>: A tensor to serialize.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>summary</strong></p></td></tr></table></div><div class="doc"><p>Outputs a <code>Summary</code> protocol buffer with a tensor.</p></div></div><div class="top"><p class="src"><a name="v:sparseSoftmaxCrossEntropyWithLogits" class="def">sparseSoftmaxCrossEntropyWithLogits</a> <a href="src/TensorFlow-GenOps-Core.html#sparseSoftmaxCrossEntropyWithLogits" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tlabels, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tlabels)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>features</strong>: batch_size x num_classes matrix</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tlabels</td><td class="doc"><p><strong>labels</strong>: batch_size vector with values in [0, num_classes).
+ This is the label for the given minibatch entry.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><td class="doc"><p>(<strong>loss</strong>, <strong>backprop</strong>)</p><ul><li><strong>loss</strong>: Per example loss (batch_size vector).</li><li><strong>backprop</strong>: backpropagated gradients (batch_size x num_classes matrix).</li></ul></td></tr></table></div><div class="doc"><p>Computes softmax cross entropy cost and gradients to backpropagate.</p><p>Unlike <code>SoftmaxCrossEntropyWithLogits</code>, this operation does not accept
+ a matrix of label probabilities, but rather a single label per row
+ of features.  This label is considered to have probability 1.0 for the
+ given row.</p><p>Inputs are the logits, not probabilities.</p></div></div><div class="top"><p class="src"><a name="v:maxPoolWithArgmax" class="def">maxPoolWithArgmax</a> <a href="src/TensorFlow-GenOps-Core.html#maxPoolWithArgmax" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> targmax, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` targmax)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: 4-D with shape `[batch, height, width, channels]`.  Input to pool over.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> targmax)</td><td class="doc"><p>(<strong>output</strong>, <strong>argmax</strong>)</p><ul><li><strong>output</strong>: The max pooled output tensor.</li><li><strong>argmax</strong>: 4-D.  The flattened indices of the max values chosen for each output.</li></ul></td></tr></table></div><div class="doc"><p>Performs max pooling on the input and outputs both max values and indices.</p><p>The indices in <code>argmax</code> are flattened, so that a maximum value at position
+ `[b, y, x, c]` becomes flattened index
+ `((b * height + y) * width + x) * channels + c`.</p></div></div><div class="top"><p class="src"><a name="v:fFT" class="def">fFT</a> <a href="src/TensorFlow-GenOps-Core.html#fFT" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>input</strong>: A complex64 tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>output</strong>: A complex64 tensor of the same shape as <code>input</code>. The inner-most
+ dimension of <code>input</code> is replaced with its 1D Fourier Transform.</p></td></tr></table></div><div class="doc"><p>Compute the 1-dimensional discrete Fourier Transform over the inner-most</p><p>dimension of <code>input</code>.</p></div></div><div class="top"><p class="src"><a name="v:histogramSummary" class="def">histogramSummary</a> <a href="src/TensorFlow-GenOps-Core.html#histogramSummary" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>tag</strong>: Scalar.  Tag to use for the <code><a href="Summary.html#v:Value">Value</a></code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>values</strong>: Any shape. Values to use to build the histogram.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>summary</strong>: Scalar. Serialized <code>Summary</code> protocol buffer.</p></td></tr></table></div><div class="doc"><p>Outputs a <code>Summary</code> protocol buffer with a histogram.</p><p>The generated
+ <a href="https://www.tensorflow.org/code/tensorflow/core/framework/summary.proto">`Summary`</a>
+ has one summary value containing a histogram for <code>values</code>.</p><p>This op reports an <code>InvalidArgument</code> error if any value is not finite.</p></div></div><div class="top"><p class="src"><a name="v:pad" class="def">pad</a> <a href="src/TensorFlow-GenOps-Core.html#pad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tpaddings, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tpaddings)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tpaddings</td><td class="doc"><p><strong>paddings</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Pads a tensor with zeros.</p><p>This operation pads a <code>input</code> with zeros according to the <code>paddings</code> you
+ specify. <code>paddings</code> is an integer tensor with shape `[Dn, 2]`, where n is the
+ rank of <code>input</code>. For each dimension D of <code>input</code>, `paddings[D, 0]` indicates
+ how many zeros to add before the contents of <code>input</code> in that dimension, and
+ `paddings[D, 1]` indicates how many zeros to add after the contents of <code>input</code>
+ in that dimension.</p><p>The padded size of each dimension D of the output is:</p><p>`paddings(D, 0) + input.dim_size(D) + paddings(D, 1)`</p><p>For example:</p><p>```prettyprint
+ # <code>t</code> is [[1, 1], [2, 2]]
+ # <code>paddings</code> is [[1, 1], [2, 2]]
+ # rank of <code>t</code> is 2
+ pad(t, paddings) ==&gt; [[0, 0, 0, 0, 0, 0]
+                       [0, 0, 1, 1, 0, 0]
+                       [0, 0, 2, 2, 0, 0]
+                       [0, 0, 0, 0, 0, 0]]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:batchIFFT3D" class="def">batchIFFT3D</a> <a href="src/TensorFlow-GenOps-Core.html#batchIFFT3D" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:imageSummary" class="def">imageSummary</a> <a href="src/TensorFlow-GenOps-Core.html#imageSummary" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>tag</strong>: Scalar. Used to build the <code>tag</code> attribute of the summary values.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>tensor</strong>: 4-D of shape `[batch_size, height, width, channels]` where
+ <code>channels</code> is 1, 3, or 4.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>summary</strong>: Scalar. Serialized <code>Summary</code> protocol buffer.</p></td></tr></table></div><div class="doc"><p>Outputs a <code>Summary</code> protocol buffer with images.</p><p>The summary has up to <code>max_images</code> summary values containing images. The
+ images are built from <code>tensor</code> which must be 4-D with shape `[batch_size,
+ height, width, channels]` and where <code>channels</code> can be:</p><ul><li>1: <code>tensor</code> is interpreted as Grayscale.</li><li>3: <code>tensor</code> is interpreted as RGB.</li><li>4: <code>tensor</code> is interpreted as RGBA.</li></ul><p>The images have the same number of channels as the input tensor. For float
+ input, the values are normalized one image at a time to fit in the range
+ `[0, 255]`.  <code>uint8</code> values are unchanged.  The op uses two different
+ normalization algorithms:</p><ul><li>If the input values are all positive, they are rescaled so the largest one
+    is 255.</li><li>If any input value is negative, the values are shifted so input value 0.0
+    is at 127.  They are then rescaled so that either the smallest value is 0,
+    or the largest one is 255.</li></ul><p>The <code>tag</code> argument is a scalar <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> of type <code>string</code>.  It is used to
+ build the <code>tag</code> of the summary values:</p><ul><li>If <code>max_images</code> is 1, the summary value tag is '*tag*/image'.</li><li>If <code>max_images</code> is greater than 1, the summary value tags are
+    generated sequentially as '*tag*/image/0', '*tag*/image/1', etc.</li></ul><p>The <code>bad_color</code> argument is the color to use in the generated images for
+ non-finite input values.  It is a <code>unit8</code> 1-D tensor of length <code>channels</code>.
+ Each element must be in the range `[0, 255]` (It represents the value of a
+ pixel in the output image).  Non-finite values in the input tensor are
+ replaced by this tensor in the output image.  The default value is the color
+ red.</p></div></div><div class="top"><p class="src"><a name="v:segmentSum" class="def">segmentSum</a> <a href="src/TensorFlow-GenOps-Core.html#segmentSum" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices</td><td class="doc"><p><strong>segment_ids</strong>: A 1-D tensor whose rank is equal to the rank of `data`'s
+ first dimension.  Values should be sorted and can be repeated.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Has same shape as data, except for dimension 0 which
+ has size <code>k</code>, the number of segments.</p></td></tr></table></div><div class="doc"><p>Computes the sum along segments of a tensor.</p><p>Read <a href="../../api_docs/python/math_ops.md#segmentation">the section on Segmentation</a>
+ for an explanation of segments.</p><p>Computes a tensor such that
+ \(output_i = sum_j data_j\) where sum is over <code>j</code> such
+ that `segment_ids[j] == i`.</p><p><a href="div">style=&quot;width:70%; margin:auto; margin-bottom:10px; margin-top:20px;&quot;</a>
+ <a href="img">style=&quot;width:100%&quot; src=&quot;../../images/SegmentSum.png&quot; alt</a>
+ <a href="/div">/div</a></p></div></div><div class="top"><p class="src"><a name="v:encodeJpeg" class="def">encodeJpeg</a> <a href="src/TensorFlow-GenOps-Core.html#encodeJpeg" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a></td><td class="doc"><p><strong>image</strong>: 3-D with shape `[height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>contents</strong>: 0-D. JPEG-encoded image.</p></td></tr></table></div><div class="doc"><p>JPEG-encode an image.</p><p><code>image</code> is a 3-D uint8 Tensor of shape `[height, width, channels]`.</p><p>The attr <code>format</code> can be used to override the color format of the encoded
+ output.  Values can be:</p><ul><li>`''`: Use a default format based on the number of channels in the image.</li><li><code>grayscale</code>: Output a grayscale JPEG image.  The <code>channels</code> dimension
+     of <code>image</code> must be 1.</li><li><code>rgb</code>: Output an RGB JPEG image. The <code>channels</code> dimension
+     of <code>image</code> must be 3.</li></ul><p>If <code>format</code> is not specified or is the empty string, a default format is picked
+ in function of the number of channels in <code>image</code>:</p><ul><li>1: Output a grayscale image.</li><li>3: Output an RGB image.</li></ul></div></div><div class="top"><p class="src"><a name="v:batchNormWithGlobalNormalizationGrad" class="def">batchNormWithGlobalNormalizationGrad</a> <a href="src/TensorFlow-GenOps-Core.html#batchNormWithGlobalNormalizationGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>scale_after_normalization</strong>: A bool indicating whether the resulted tensor
+ needs to be multiplied with gamma.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>variance_epsilon</strong>: A small float number to avoid dividing by 0.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>t</strong>: A 4D input Tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>m</strong>: A 1D mean Tensor with size matching the last dimension of t.
+ This is the first output from tf.nn.moments,
+ or a saved moving average thereof.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>v</strong>: A 1D variance Tensor with size matching the last dimension of t.
+ This is the second output from tf.nn.moments,
+ or a saved moving average thereof.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>gamma</strong>: A 1D gamma Tensor with size matching the last dimension of t.
+ If &quot;scale_after_normalization&quot; is true, this Tensor will be multiplied
+ with the normalized Tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>backprop</strong>: 4D backprop Tensor.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><td class="doc"><p>(<strong>dx</strong>, <strong>dm</strong>, <strong>dv</strong>, <strong>db</strong>, <strong>dg</strong>)</p><ul><li><strong>dx</strong>: 4D backprop tensor for input.</li><li><strong>dm</strong>: 1D backprop tensor for mean.</li><li><strong>dv</strong>: 1D backprop tensor for variance.</li><li><strong>db</strong>: 1D backprop tensor for beta.</li><li><strong>dg</strong>: 1D backprop tensor for gamma.</li></ul></td></tr></table></div><div class="doc"><p>Gradients for batch normalization.</p><p>This op is deprecated. See `tf.nn.batch_normalization`.</p></div></div><div class="top"><p class="src"><a name="v:biasAddV1" class="def">biasAddV1</a> <a href="src/TensorFlow-GenOps-Core.html#biasAddV1" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>value</strong>: Any number of dimensions.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>bias</strong>: 1-D with size the last dimension of <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Broadcasted sum of <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code> and <code>bias</code>.</p></td></tr></table></div><div class="doc"><p>Adds <code>bias</code> to <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>.</p><p>This is a deprecated version of BiasAdd and will be soon removed.</p><p>This is a special case of `tf.add` where <code>bias</code> is restricted to be 1-D.
+ Broadcasting is supported, so <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code> may have any number of dimensions.</p></div></div><div class="top"><p class="src"><a name="v:invertPermutation" class="def">invertPermutation</a> <a href="src/TensorFlow-GenOps-Core.html#invertPermutation" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong>: 1-D.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong>: 1-D.</p></td></tr></table></div><div class="doc"><p>Computes the inverse permutation of a tensor.</p><p>This operation computes the inverse of an index permutation. It takes a 1-D
+ integer tensor <code>x</code>, which represents the indices of a zero-based array, and
+ swaps each value with its index position. In other words, for an output tensor
+ <code>y</code> and an input tensor <code>x</code>, this operation computes the following:</p><p>`y[x[i]] = i for i in [0, 1, ..., len(x) - 1]`</p><p>The values must include 0. There can be no duplicate values or negative values.</p><p>For example:</p><p>```prettyprint
+ # tensor <code>x</code> is [3, 4, 0, 2, 1]
+ invert_permutation(x) ==&gt; [2, 4, 3, 0, 1]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:mirrorPadGrad" class="def">mirrorPadGrad</a> <a href="src/TensorFlow-GenOps-Core.html#mirrorPadGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tpaddings, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tpaddings)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: The input tensor to be folded.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tpaddings</td><td class="doc"><p><strong>paddings</strong>: A two-column matrix specifying the padding sizes. The number of
+ rows must be the same as the rank of <code>input</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The folded tensor.</p></td></tr></table></div><div class="doc"><p>Gradient op for <code>MirrorPad</code> op. This op folds a mirror-padded tensor.</p><p>This operation folds the padded areas of <code>input</code> by <code>MirrorPad</code> according to the
+ <code>paddings</code> you specify. <code>paddings</code> must be the same as <code>paddings</code> argument
+ given to the corresponding <code>MirrorPad</code> op.</p><p>The folded size of each dimension D of the output is:</p><p>`input.dim_size(D) - paddings(D, 0) - paddings(D, 1)`</p><p>For example:</p><p>```prettyprint
+ # <code>t</code> is [[1, 2, 3], [4, 5, 6], [7, 8, 9]].
+ # <code>paddings</code> is [[0, 1]], [0, 1]].
+ # <code>mode</code> is SYMMETRIC.
+ # rank of <code>t</code> is 2.
+ pad(t, paddings) ==&gt; [[ 1,  5]
+                       [11, 28]]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:reverse" class="def">reverse</a> <a href="src/TensorFlow-GenOps-Core.html#reverse" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>tensor</strong>: Up to 8-D.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>dims</strong>: 1-D. The dimensions to reverse.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The same shape as <code>tensor</code>.</p></td></tr></table></div><div class="doc"><p>Reverses specific dimensions of a tensor.</p><p>Given a <code>tensor</code>, and a <code>bool</code> tensor <code>dims</code> representing the dimensions
+ of <code>tensor</code>, this operation reverses each dimension i of <code>tensor</code> where
+ `dims[i]` is <code><a href="../base-4.8.2.0/Data-Bool.html#v:True">True</a></code>.</p><p><code>tensor</code> can have up to 8 dimensions. The number of dimensions
+ of <code>tensor</code> must equal the number of elements in <code>dims</code>. In other words:</p><p>`rank(tensor) = size(dims)`</p><p>For example:</p><p>```prettyprint
+ # tensor <code>t</code> is [[[[ 0,  1,  2,  3],
+ #                  [ 4,  5,  6,  7],
+ #                  [ 8,  9, 10, 11]],
+ #                 [[12, 13, 14, 15],
+ #                  [16, 17, 18, 19],
+ #                  [20, 21, 22, 23]]]]
+ # tensor <code>t</code> shape is [1, 2, 3, 4]</p><p># <code>dims</code> is [False, False, False, True]
+ reverse(t, dims) ==&gt; [[[[ 3,  2,  1,  0],
+                         [ 7,  6,  5,  4],
+                         [ 11, 10, 9, 8]],
+                        [[15, 14, 13, 12],
+                         [19, 18, 17, 16],
+                         [23, 22, 21, 20]]]]</p><p># <code>dims</code> is [False, True, False, False]
+ reverse(t, dims) ==&gt; [[[[12, 13, 14, 15],
+                         [16, 17, 18, 19],
+                         [20, 21, 22, 23]
+                        [[ 0,  1,  2,  3],
+                         [ 4,  5,  6,  7],
+                         [ 8,  9, 10, 11]]]]</p><p># <code>dims</code> is [False, False, True, False]
+ reverse(t, dims) ==&gt; [[[[8, 9, 10, 11],
+                         [4, 5, 6, 7],
+                         [0, 1, 2, 3]]
+                        [[20, 21, 22, 23],
+                         [16, 17, 18, 19],
+                         [12, 13, 14, 15]]]]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:conv2D" class="def">conv2D</a> <a href="src/TensorFlow-GenOps-Core.html#conv2D" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>filter</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Computes a 2-D convolution given 4-D <code>input</code> and <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code> tensors.</p><p>Given an input tensor of shape `[batch, in_height, in_width, in_channels]`
+ and a filter / kernel tensor of shape
+ `[filter_height, filter_width, in_channels, out_channels]`, this op
+ performs the following:</p><ol><li>Flattens the filter to a 2-D matrix with shape
+    `[filter_height * filter_width * in_channels, output_channels]`.</li><li>Extracts image patches from the input tensor to form a *virtual*
+    tensor of shape `[batch, out_height, out_width,
+    filter_height * filter_width * in_channels]`.</li><li>For each patch, right-multiplies the filter matrix and the image patch
+    vector.</li></ol><p>In detail, with the default NHWC format,</p><p>output[b, i, j, k] =
+         sum_{di, dj, q} input[b, strides[1] * i + di, strides[2] * j + dj, q] *
+                         filter[di, dj, q, k]</p><p>Must have `strides[0] = strides[3] = 1`.  For the most common case of the same
+ horizontal and vertices strides, `strides = [1, stride, stride, 1]`.</p></div></div><div class="top"><p class="src"><a name="v:conv2DBackpropInput" class="def">conv2DBackpropInput</a> <a href="src/TensorFlow-GenOps-Core.html#conv2DBackpropInput" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>input_sizes</strong>: An integer vector representing the shape of <code>input</code>,
+ where <code>input</code> is a 4-D `[batch, height, width, channels]` tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>filter</strong>: 4-D with shape
+ `[filter_height, filter_width, in_channels, out_channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>out_backprop</strong>: 4-D with shape `[batch, out_height, out_width, out_channels]`.
+ Gradients w.r.t. the output of the convolution.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 4-D with shape `[batch, in_height, in_width, in_channels]`.  Gradient
+ w.r.t. the input of the convolution.</p></td></tr></table></div><div class="doc"><p>Computes the gradients of convolution with respect to the input.</p></div></div><div class="top"><p class="src"><a name="v:readerSerializeState" class="def">readerSerializeState</a> <a href="src/TensorFlow-GenOps-Core.html#readerSerializeState" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>reader_handle</strong>: Handle to a Reader.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>state</strong></p></td></tr></table></div><div class="doc"><p>Produce a string tensor that encodes the state of a Reader.</p><p>Not all Readers support being serialized, so this can produce an
+ Unimplemented error.</p></div></div><div class="top"><p class="src"><a name="v:temporaryVariable" class="def">temporaryVariable</a> <a href="src/TensorFlow-GenOps-Core.html#temporaryVariable" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</td><td class="doc"><p><strong>ref</strong>: A reference to the variable tensor.</p></td></tr></table></div><div class="doc"><p>Returns a tensor that may be mutated, but only persists within a single step.</p><p>This is an experimental op for internal use only and it is possible to use this
+ op in unsafe ways.  DO NOT USE unless you fully understand the risks.</p><p>It is the caller's responsibility to ensure that <code>ref</code> is eventually passed to a
+ matching <code>DestroyTemporaryVariable</code> op after all other uses have completed.</p><p>Outputs a ref to the tensor state so it may be read or modified.</p><p>E.g.
+       var = state_ops._temporary_variable([1, 2], types.float_)
+       var_name = var.op.name
+       var = state_ops.assign(var, [[4.0, 5.0]])
+       var = state_ops.assign_add(var, [[6.0, 7.0]])
+       final = state_ops._destroy_temporary_variable(var, var_name=var_name)</p></div></div><div class="top"><p class="src"><a name="v:cropAndResize" class="def">cropAndResize</a> <a href="src/TensorFlow-GenOps-Core.html#cropAndResize" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>image</strong>: A 4-D tensor of shape `[batch, image_height, image_width, depth]`.
+ Both <code>image_height</code> and <code>image_width</code> need to be positive.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>boxes</strong>: A 2-D tensor of shape `[num_boxes, 4]`. The <code>i</code>-th row of the tensor
+ specifies the coordinates of a box in the `box_ind[i]` image and is specified
+ in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of
+ <code>y</code> is mapped to the image coordinate at `y * (image_height - 1)`, so as the
+ `[0, 1]` interval of normalized image height is mapped to
+ `[0, image_height - 1] in image height coordinates. We do allow y1 &gt; y2, in
+ which case the sampled crop is an up-down flipped version of the original
+ image. The width dimension is treated similarly. Normalized coordinates
+ outside the `[0, 1]` range are allowed, in which case we use
+ <code>extrapolation_value</code> to extrapolate the input image values.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>box_ind</strong>: A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.
+ The value of `box_ind[i]` specifies the image that the <code>i</code>-th box refers to.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>crop_size</strong>: A 1-D tensor of 2 elements, `size = [crop_height, crop_width]`. All
+ cropped image patches are resized to this size. The aspect ratio of the image
+ content is not preserved. Both <code>crop_height</code> and <code>crop_width</code> need to be
+ positive.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>crops</strong>: A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`.</p></td></tr></table></div><div class="doc"><p>Extracts crops from the input image tensor and bilinearly resizes them (possibly</p><p>with aspect ratio change) to a common output size specified by <code>crop_size</code>. This
+ is more general than the <code>crop_to_bounding_box</code> op which extracts a fixed size
+ slice from the input image and does not allow resizing or aspect ratio change.</p><p>Returns a tensor with <code>crops</code> from the input <code>image</code> at positions defined at the
+ bounding box locations in <code>boxes</code>. The cropped boxes are all resized (with
+ bilinear interpolation) to a fixed `size = [crop_height, crop_width]`. The
+ result is a 4-D tensor `[num_boxes, crop_height, crop_width, depth]`.</p></div></div><div class="top"><p class="src"><a name="v:maxPoolGrad" class="def">maxPoolGrad</a> <a href="src/TensorFlow-GenOps-Core.html#maxPoolGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>orig_input</strong>: The original input tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>orig_output</strong>: The original output tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>grad</strong>: 4-D.  Gradients w.r.t. the output of <code>max_pool</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Gradients w.r.t. the input to <code>max_pool</code>.</p></td></tr></table></div><div class="doc"><p>Computes gradients of the maxpooling function.</p></div></div><div class="top"><p class="src"><a name="v:fusedResizeAndPadConv2D" class="def">fusedResizeAndPadConv2D</a> <a href="src/TensorFlow-GenOps-Core.html#fusedResizeAndPadConv2D" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: 4-D with shape `[batch, in_height, in_width, in_channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>size</strong>: A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
+ new size for the images.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>paddings</strong>: A two-column matrix specifying the padding sizes. The number of
+ rows must be the same as the rank of <code>input</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>filter</strong>: 4-D with shape
+ `[filter_height, filter_width, in_channels, out_channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Performs a resize and padding as a preprocess during a convolution.</p><p>It's often possible to do spatial transformations more efficiently as part of
+ the packing stage of a convolution, so this op allows for an optimized
+ implementation where these stages are fused together. This prevents the need to
+ write out the intermediate results as whole tensors, reducing memory pressure,
+ and we can get some latency gains by merging the transformation calculations.
+ The data_format attribute for Conv2D isn't supported by this op, and defaults to
+ <code>NHWC</code> order.
+ Internally this op uses a single per-graph scratch buffer, which means that it
+ will block if multiple versions are being run in parallel. This is because this
+ operator is primarily an optimization to minimize memory usage.</p></div></div><div class="top"><p class="src"><a name="v:randomUniform" class="def">randomUniform</a> <a href="src/TensorFlow-GenOps-Core.html#randomUniform" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` dtype)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>shape</strong>: The shape of the output tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</td><td class="doc"><p><strong>output</strong>: A tensor of the specified shape filled with uniform random values.</p></td></tr></table></div><div class="doc"><p>Outputs random values from a uniform distribution.</p><p>The generated values follow a uniform distribution in the range `[0, 1)`. The
+ lower bound 0 is included in the range, while the upper bound 1 is excluded.</p></div></div><div class="top"><p class="src"><a name="v:depthwiseConv2dNative" class="def">depthwiseConv2dNative</a> <a href="src/TensorFlow-GenOps-Core.html#depthwiseConv2dNative" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>filter</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Computes a 2-D depthwise convolution given 4-D <code>input</code> and <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code> tensors.</p><p>Given an input tensor of shape `[batch, in_height, in_width, in_channels]`
+ and a filter / kernel tensor of shape
+ `[filter_height, filter_width, in_channels, channel_multiplier]`, containing
+ <code>in_channels</code> convolutional filters of depth 1, <code>depthwise_conv2d</code> applies
+ a different filter to each input channel (expanding from 1 channel to
+ <code>channel_multiplier</code> channels for each), then concatenates the results
+ together. Thus, the output has `in_channels * channel_multiplier` channels.</p><p>for k in 0..in_channels-1
+   for q in 0..channel_multiplier-1
+     output[b, i, j, k * channel_multiplier + q] =
+       sum_{di, dj} input[b, strides[1] * i + di, strides[2] * j + dj, k] *
+                         filter[di, dj, k, q]</p><p>Must have `strides[0] = strides[3] = 1`.  For the most common case of the same
+ horizontal and vertices strides, `strides = [1, stride, stride, 1]`.</p></div></div><div class="top"><p class="src"><a name="v:sparseApplyAdadelta" class="def">sparseApplyAdadelta</a> <a href="src/TensorFlow-GenOps-Core.html#sparseApplyAdadelta" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>accum</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>accum_update</strong>: : Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>lr</strong>: Learning rate. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>rho</strong>: Decay factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t</td><td class="doc"><p><strong>epsilon</strong>: Constant factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t</td><td class="doc"><p><strong>grad</strong>: The gradient.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v8 tindices</td><td class="doc"><p><strong>indices</strong>: A vector of indices into the first dimension of var and accum.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>var: Should be from a Variable().</p></div></div><div class="top"><p class="src"><a name="v:depthwiseConv2dNativeBackpropFilter" class="def">depthwiseConv2dNativeBackpropFilter</a> <a href="src/TensorFlow-GenOps-Core.html#depthwiseConv2dNativeBackpropFilter" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: 4-D with shape `[batch, in_height, in_width, in_channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>filter_sizes</strong>: An integer vector representing the tensor shape of <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code>,
+ where <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code> is a 4-D
+ `[filter_height, filter_width, in_channels, depthwise_multiplier]` tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>out_backprop</strong>: 4-D with shape `[batch, out_height, out_width, out_channels]`.
+ Gradients w.r.t. the output of the convolution.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 4-D with shape
+ `[filter_height, filter_width, in_channels, out_channels]`.  Gradient w.r.t.
+ the <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code> input of the convolution.</p></td></tr></table></div><div class="doc"><p>Computes the gradients of depthwise convolution with respect to the filter.</p></div></div><div class="top"><p class="src"><a name="v:conv3D" class="def">conv3D</a> <a href="src/TensorFlow-GenOps-Core.html#conv3D" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Shape `[batch, in_depth, in_height, in_width, in_channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>filter</strong>: Shape `[filter_depth, filter_height, filter_width, in_channels,
+ out_channels]`. <code>in_channels</code> must match between <code>input</code> and <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Computes a 3-D convolution given 5-D <code>input</code> and <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code> tensors.</p><p>In signal processing, cross-correlation is a measure of similarity of
+ two waveforms as a function of a time-lag applied to one of them. This
+ is also known as a sliding dot product or sliding inner-product.</p><p>Our Conv3D implements a form of cross-correlation.</p></div></div><div class="top"><p class="src"><a name="v:greaterEqual" class="def">greaterEqual</a> <a href="src/TensorFlow-GenOps-Core.html#greaterEqual" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns the truth value of (x &gt;= y) element-wise.</p><ul><li>NOTE*: <code>GreaterEqual</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:sparseDenseCwiseAdd" class="def">sparseDenseCwiseAdd</a> <a href="src/TensorFlow-GenOps-Core.html#sparseDenseCwiseAdd" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>sp_indices</strong>: 2-D.  `N x R` matrix with the indices of non-empty values in a
+ SparseTensor, possibly not in canonical ordering.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>sp_values</strong>: 1-D.  <code>N</code> non-empty values corresponding to <code>sp_indices</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>sp_shape</strong>: 1-D.  Shape of the input SparseTensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>dense</strong>: <code>R</code>-D.  The dense Tensor operand.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 1-D.  The <code>N</code> values that are operated on.</p></td></tr></table></div><div class="doc"><p>Adds up a SparseTensor and a dense Tensor, using these special rules:</p><ol><li>Broadcasts the dense side to have the same shape as the sparse side, if
+     eligible;</li><li>Then, only the dense values pointed to by the indices of the SparseTensor
+     participate in the cwise addition.</li></ol><p>By these rules, the result is a logical SparseTensor with exactly the same
+ indices and shape, but possibly with different non-zero values.  The output of
+ this Op is the resultant non-zero values.</p></div></div><div class="top"><p class="src"><a name="v:conv3DBackpropFilter" class="def">conv3DBackpropFilter</a> <a href="src/TensorFlow-GenOps-Core.html#conv3DBackpropFilter" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Shape `[batch, depth, rows, cols, in_channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>filter</strong>: Shape `[depth, rows, cols, in_channels, out_channels]`.
+ <code>in_channels</code> must match between <code>input</code> and <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>out_backprop</strong>: Backprop signal of shape `[batch, out_depth, out_rows, out_cols,
+ out_channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Computes the gradients of 3-D convolution with respect to the filter.</p></div></div><div class="top"><p class="src"><a name="v:conv3DBackpropInputV2" class="def">conv3DBackpropInputV2</a> <a href="src/TensorFlow-GenOps-Core.html#conv3DBackpropInputV2" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>input_sizes</strong>: An integer vector representing the tensor shape of <code>input</code>,
+ where <code>input</code> is a 5-D
+ `[batch, depth, rows, cols, in_channels]` tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>filter</strong>: Shape `[depth, rows, cols, in_channels, out_channels]`.
+ <code>in_channels</code> must match between <code>input</code> and <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>out_backprop</strong>: Backprop signal of shape `[batch, out_depth, out_rows, out_cols,
+ out_channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Computes the gradients of 3-D convolution with respect to the input.</p></div></div><div class="top"><p class="src"><a name="v:mod" class="def">mod</a> <a href="src/TensorFlow-GenOps-Core.html#mod" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns element-wise remainder of division.</p><ul><li>NOTE*: <code>Mod</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:refMerge" class="def">refMerge</a> <a href="src/TensorFlow-GenOps-Core.html#refMerge" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t]</td><td class="doc"><p><strong>inputs</strong>: The input tensors, exactly one of which will become available.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>)</td><td class="doc"><p>(<strong>output</strong>, <strong>value_index</strong>)</p><ul><li><strong>output</strong>: Will be set to the available input tensor.</li><li><strong>value_index</strong>: The index of the chosen input tensor in <code>inputs</code>.</li></ul></td></tr></table></div><div class="doc"><p>Forwards the value of an available tensor from <code>inputs</code> to <code>output</code>.</p><p><code>Merge</code> waits for at least one of the tensors in <code>inputs</code> to become available.
+ It is usually combined with <code>Switch</code> to implement branching.</p><p><code>Merge</code> forwards the first tensor for become available to <code>output</code>, and sets
+ <code>value_index</code> to its index in <code>inputs</code>.</p></div></div><div class="top"><p class="src"><a name="v:conv3DBackpropFilterV2" class="def">conv3DBackpropFilterV2</a> <a href="src/TensorFlow-GenOps-Core.html#conv3DBackpropFilterV2" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Shape `[batch, depth, rows, cols, in_channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>filter_sizes</strong>: An integer vector representing the tensor shape of <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code>,
+ where <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code> is a 5-D
+ `[filter_depth, filter_height, filter_width, in_channels, out_channels]`
+ tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>out_backprop</strong>: Backprop signal of shape `[batch, out_depth, out_rows, out_cols,
+ out_channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Computes the gradients of 3-D convolution with respect to the filter.</p></div></div><div class="top"><p class="src"><a name="v:serializeManySparse" class="def">serializeManySparse</a> <a href="src/TensorFlow-GenOps-Core.html#serializeManySparse" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>sparse_indices</strong>: 2-D.  The <code>indices</code> of the minibatch <code>SparseTensor</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>sparse_values</strong>: 1-D.  The <code>values</code> of the minibatch <code>SparseTensor</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>sparse_shape</strong>: 1-D.  The <code><a href="TensorFlow-GenOps-Core.html#v:shape">shape</a></code> of the minibatch <code>SparseTensor</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>serialized_sparse</strong></p></td></tr></table></div><div class="doc"><p>Serialize an <code>N</code>-minibatch <code>SparseTensor</code> into an `[N, 3]` string <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code>.</p><p>The <code>SparseTensor</code> must have rank <code>R</code> greater than 1, and the first dimension
+ is treated as the minibatch dimension.  Elements of the <code>SparseTensor</code>
+ must be sorted in increasing order of this first dimension.  The serialized
+ <code>SparseTensor</code> objects going into each row of <code>serialized_sparse</code> will have
+ rank `R-1`.</p><p>The minibatch size <code>N</code> is extracted from `sparse_shape[0]`.</p></div></div><div class="top"><p class="src"><a name="v:avgPool3DGrad" class="def">avgPool3DGrad</a> <a href="src/TensorFlow-GenOps-Core.html#avgPool3DGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>orig_input_shape</strong>: The original input dimensions.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>grad</strong>: Output backprop of shape `[batch, depth, rows, cols, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The backprop for input.</p></td></tr></table></div><div class="doc"><p>Computes gradients of average pooling function.</p></div></div><div class="top"><p class="src"><a name="v:maxPool3DGrad" class="def">maxPool3DGrad</a> <a href="src/TensorFlow-GenOps-Core.html#maxPool3DGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>orig_input</strong>: The original input tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>orig_output</strong>: The original output tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>grad</strong>: Output backprop of shape `[batch, depth, rows, cols, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Computes gradients of max pooling function.</p></div></div><div class="top"><p class="src"><a name="v:sparseReduceSum" class="def">sparseReduceSum</a> <a href="src/TensorFlow-GenOps-Core.html#sparseReduceSum" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>input_indices</strong>: 2-D.  `N x R` matrix with the indices of non-empty values in a
+ SparseTensor, possibly not in canonical ordering.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>input_values</strong>: 1-D.  <code>N</code> non-empty values corresponding to <code>input_indices</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>input_shape</strong>: 1-D.  Shape of the input SparseTensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>reduction_axes</strong>: 1-D.  Length-<code>K</code> vector containing the reduction axes.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: `R-K`-D.  The reduced Tensor.</p></td></tr></table></div><div class="doc"><p>Computes the sum of elements across dimensions of a SparseTensor.</p><p>This Op takes a SparseTensor and is the sparse counterpart to
+ `tf.reduce_sum()`.  In particular, this Op also returns a dense <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code>
+ instead of a sparse one.</p><p>Reduces <code>sp_input</code> along the dimensions given in <code>reduction_axes</code>.  Unless
+ <code>keep_dims</code> is true, the rank of the tensor is reduced by 1 for each entry in
+ <code>reduction_axes</code>. If <code>keep_dims</code> is true, the reduced dimensions are retained
+ with length 1.</p><p>If <code>reduction_axes</code> has no entries, all dimensions are reduced, and a tensor
+ with a single element is returned.  Additionally, the axes can be negative,
+ which are interpreted according to the indexing rules in Python.</p></div></div><div class="top"><p class="src"><a name="v:relu" class="def">relu</a> <a href="src/TensorFlow-GenOps-Core.html#relu" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>features</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>activations</strong></p></td></tr></table></div><div class="doc"><p>Computes rectified linear: `max(features, 0)`.</p></div></div><div class="top"><p class="src"><a name="v:l2Loss" class="def">l2Loss</a> <a href="src/TensorFlow-GenOps-Core.html#l2Loss" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>t</strong>: Typically 2-D, but may have any dimensions.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 0-D.</p></td></tr></table></div><div class="doc"><p>L2 Loss.</p><p>Computes half the L2 norm of a tensor without the <code><a href="../base-4.8.2.0/Prelude.html#v:sqrt">sqrt</a></code>:</p><p>output = sum(t ** 2) / 2</p></div></div><div class="top"><p class="src"><a name="v:readerRestoreState" class="def">readerRestoreState</a> <a href="src/TensorFlow-GenOps-Core.html#readerRestoreState" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>reader_handle</strong>: Handle to a Reader.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>state</strong>: Result of a ReaderSerializeState of a Reader with type
+ matching reader_handle.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Restore a reader to a previously saved state.</p><p>Not all Readers support being restored, so this can produce an
+ Unimplemented error.</p></div></div><div class="top"><p class="src"><a name="v:shape" class="def">shape</a> <a href="src/TensorFlow-GenOps-Core.html#shape" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_type, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` out_type)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_type</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Returns the shape of a tensor.</p><p>This operation returns a 1-D integer tensor representing the shape of <code>input</code>.</p><p>For example:</p><p>```prettyprint
+ # <code>t</code> is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
+ shape(t) ==&gt; [2, 2, 3]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:softmaxCrossEntropyWithLogits" class="def">softmaxCrossEntropyWithLogits</a> <a href="src/TensorFlow-GenOps-Core.html#softmaxCrossEntropyWithLogits" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>features</strong>: batch_size x num_classes matrix</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>labels</strong>: batch_size x num_classes matrix
+ The caller must ensure that each batch of labels represents a valid
+ probability distribution.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><td class="doc"><p>(<strong>loss</strong>, <strong>backprop</strong>)</p><ul><li><strong>loss</strong>: Per example loss (batch_size vector).</li><li><strong>backprop</strong>: backpropagated gradients (batch_size x num_classes matrix).</li></ul></td></tr></table></div><div class="doc"><p>Computes softmax cross entropy cost and gradients to backpropagate.</p><p>Inputs are the logits, not probabilities.</p></div></div><div class="top"><p class="src"><a name="v:maxPool" class="def">maxPool</a> <a href="src/TensorFlow-GenOps-Core.html#maxPool" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: 4-D input to pool over.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The max pooled output tensor.</p></td></tr></table></div><div class="doc"><p>Performs max pooling on the input.</p></div></div><div class="top"><p class="src"><a name="v:dilation2DBackpropInput" class="def">dilation2DBackpropInput</a> <a href="src/TensorFlow-GenOps-Core.html#dilation2DBackpropInput" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: 4-D with shape `[batch, in_height, in_width, depth]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>filter</strong>: 3-D with shape `[filter_height, filter_width, depth]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>out_backprop</strong>: 4-D with shape `[batch, out_height, out_width, depth]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>in_backprop</strong>: 4-D with shape `[batch, in_height, in_width, depth]`.</p></td></tr></table></div><div class="doc"><p>Computes the gradient of morphological 2-D dilation with respect to the input.</p></div></div><div class="top"><p class="src"><a name="v:equal" class="def">equal</a> <a href="src/TensorFlow-GenOps-Core.html#equal" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a>, <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns the truth value of (x == y) element-wise.</p><ul><li>NOTE*: <code>Equal</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:dilation2DBackpropFilter" class="def">dilation2DBackpropFilter</a> <a href="src/TensorFlow-GenOps-Core.html#dilation2DBackpropFilter" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: 4-D with shape `[batch, in_height, in_width, depth]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>filter</strong>: 3-D with shape `[filter_height, filter_width, depth]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>out_backprop</strong>: 4-D with shape `[batch, out_height, out_width, depth]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>filter_backprop</strong>: 3-D with shape `[filter_height, filter_width, depth]`.</p></td></tr></table></div><div class="doc"><p>Computes the gradient of morphological 2-D dilation with respect to the filter.</p></div></div><div class="top"><p class="src"><a name="v:reluGrad" class="def">reluGrad</a> <a href="src/TensorFlow-GenOps-Core.html#reluGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>gradients</strong>: The backpropagated gradients to the corresponding Relu operation.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>features</strong>: The features passed as input to the corresponding Relu operation, OR
+ the outputs of that operation (both work equivalently).</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>backprops</strong>: `gradients * (features &gt; 0)`.</p></td></tr></table></div><div class="doc"><p>Computes rectified linear gradients for a Relu operation.</p></div></div><div class="top"><p class="src"><a name="v:relu6" class="def">relu6</a> <a href="src/TensorFlow-GenOps-Core.html#relu6" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>features</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>activations</strong></p></td></tr></table></div><div class="doc"><p>Computes rectified linear 6: `min(max(features, 0), 6)`.</p></div></div><div class="top"><p class="src"><a name="v:resizeBicubic" class="def">resizeBicubic</a> <a href="src/TensorFlow-GenOps-Core.html#resizeBicubic" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>images</strong>: 4-D with shape `[batch, height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>size</strong>: = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
+ new size for the images.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>resized_images</strong>: 4-D with shape
+ `[batch, new_height, new_width, channels]`.</p></td></tr></table></div><div class="doc"><p>Resize <code>images</code> to <code><a href="TensorFlow-GenOps-Core.html#v:size">size</a></code> using bicubic interpolation.</p><p>Input images can be of different types but output images are always float.</p></div></div><div class="top"><p class="src"><a name="v:relu6Grad" class="def">relu6Grad</a> <a href="src/TensorFlow-GenOps-Core.html#relu6Grad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>gradients</strong>: The backpropagated gradients to the corresponding Relu6 operation.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>features</strong>: The features passed as input to the corresponding Relu6 operation.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>backprops</strong>: The gradients:
+ `gradients * features * (features &gt; 0) * (features &lt; 6)`.</p></td></tr></table></div><div class="doc"><p>Computes rectified linear 6 gradients for a Relu6 operation.</p></div></div><div class="top"><p class="src"><a name="v:sparseTensorDenseMatMul" class="def">sparseTensorDenseMatMul</a> <a href="src/TensorFlow-GenOps-Core.html#sparseTensorDenseMatMul" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>a_indices</strong>: 2-D.  The <code>indices</code> of the <code>SparseTensor</code>, size `[nnz, 2]` Matrix.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>a_values</strong>: 1-D.  The <code>values</code> of the <code>SparseTensor</code>, size `[nnz]` Vector.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>a_shape</strong>: 1-D.  The <code><a href="TensorFlow-GenOps-Core.html#v:shape">shape</a></code> of the <code>SparseTensor</code>, size `[2]` Vector.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>b</strong>: 2-D.  A dense Matrix.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>product</strong></p></td></tr></table></div><div class="doc"><p>Multiply SparseTensor (of rank 2) <a href="A.html">A</a> by dense matrix <a href="B.html">B</a>.</p><p>No validity checking is performed on the indices of A.  However, the following
+ input format is recommended for optimal behavior:</p><p>if adjoint_a == false:
+   A should be sorted in lexicographically increasing order.  Use SparseReorder
+   if you're not sure.
+ if adjoint_a == true:
+   A should be sorted in order of increasing dimension 1 (i.e., &quot;column major&quot;
+   order instead of &quot;row major&quot; order).</p></div></div><div class="top"><p class="src"><a name="v:softplus" class="def">softplus</a> <a href="src/TensorFlow-GenOps-Core.html#softplus" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>features</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>activations</strong></p></td></tr></table></div><div class="doc"><p>Computes softplus: `log(exp(features) + 1)`.</p></div></div><div class="top"><p class="src"><a name="v:batchMatMul" class="def">batchMatMul</a> <a href="src/TensorFlow-GenOps-Core.html#batchMatMul" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong>: 3-D or higher with shape `[..., r_x, c_x]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong>: 3-D or higher with shape `[..., r_y, c_y]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 3-D or higher with shape `[..., r_o, c_o]`</p></td></tr></table></div><div class="doc"><p>Multiplies slices of two tensors in batches.</p><p>Multiplies all slices of <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> <code>x</code> and <code>y</code> (each slice can be
+ viewed as an element of a batch), and arranges the individual results
+ in a single output tensor of the same batch size. Each of the
+ individual slices can optionally be adjointed (to adjoint a matrix
+ means to transpose and conjugate it) before multiplication by setting
+ the <code>adj_x</code> or <code>adj_y</code> flag to <code><a href="../base-4.8.2.0/Data-Bool.html#v:True">True</a></code>, which are by default <code><a href="../base-4.8.2.0/Data-Bool.html#v:False">False</a></code>.</p><p>The input tensors <code>x</code> and <code>y</code> are 3-D or higher with shape `[..., r_x, c_x]`
+ and `[..., r_y, c_y]`.</p><p>The output tensor is 3-D or higher with shape `[..., r_o, c_o]`, where:</p><p>r_o = c_x if adj_x else r_x
+     c_o = r_y if adj_y else c_y</p><p>It is computed as:</p><p>output[..., :, :] = matrix(x[..., :, :]) * matrix(y[..., :, :])</p></div></div><div class="top"><p class="src"><a name="v:softsignGrad" class="def">softsignGrad</a> <a href="src/TensorFlow-GenOps-Core.html#softsignGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>gradients</strong>: The backpropagated gradients to the corresponding softsign operation.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>features</strong>: The features passed as input to the corresponding softsign operation.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>backprops</strong>: The gradients: `gradients / (1 + abs(-features)) ** 2`.</p></td></tr></table></div><div class="doc"><p>Computes softsign gradients for a softsign operation.</p></div></div><div class="top"><p class="src"><a name="v:lessEqual" class="def">lessEqual</a> <a href="src/TensorFlow-GenOps-Core.html#lessEqual" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns the truth value of (x &lt;= y) element-wise.</p><ul><li>NOTE*: <code>LessEqual</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:logSoftmax" class="def">logSoftmax</a> <a href="src/TensorFlow-GenOps-Core.html#logSoftmax" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>logits</strong>: 2-D with shape `[batch_size, num_classes]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>logsoftmax</strong>: Same shape as <code>logits</code>.</p></td></tr></table></div><div class="doc"><p>Computes log softmax activations.</p><p>For each batch <code>i</code> and class <code>j</code> we have</p><p>logsoftmax[i, j] = logits[i, j] - log(sum(exp(logits[i])))</p></div></div><div class="top"><p class="src"><a name="v:inTopK" class="def">inTopK</a> <a href="src/TensorFlow-GenOps-Core.html#inTopK" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>k</strong>: Number of top elements to look at for computing precision.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>predictions</strong>: A <code>batch_size</code> x <code>classes</code> tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>targets</strong>: A <code>batch_size</code> vector of class ids.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>precision</strong>: Computed Precision at <code>k</code> as a `bool Tensor`.</p></td></tr></table></div><div class="doc"><p>Says whether the targets are in the top <code>K</code> predictions.</p><p>This outputs a <code>batch_size</code> bool array, an entry `out[i]` is <code>true</code> if the
+ prediction for the target class is among the top <code>k</code> predictions among
+ all predictions for example <code>i</code>. Note that the behavior of <code>InTopK</code> differs
+ from the <code>TopK</code> op in its handling of ties; if multiple classes have the
+ same prediction value and straddle the top-<code>k</code> boundary, all of those
+ classes are considered to be in the top <code>k</code>.</p><p>More formally, let</p><p>\(predictions_i\) be the predictions for all classes for example <code>i</code>,
+   \(targets_i\) be the target class for example <code>i</code>,
+   \(out_i\) be the output for example <code>i</code>,</p><p>$$out_i = predictions_{i, targets_i} in TopKIncludingTies(predictions_i)$$</p></div></div><div class="top"><p class="src"><a name="v:matrixDiag" class="def">matrixDiag</a> <a href="src/TensorFlow-GenOps-Core.html#matrixDiag" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>diagonal</strong>: Rank <code>k</code>, where `k &gt;= 1`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Rank `k+1`, with `output.shape = diagonal.shape + [diagonal.shape[-1]]`.</p></td></tr></table></div><div class="doc"><p>Returns a batched diagonal tensor with a given batched diagonal values.</p><p>Given a <code>diagonal</code>, this operation returns a tensor with the <code>diagonal</code> and
+ everything else padded with zeros. The diagonal is computed as follows:</p><p>Assume <code>diagonal</code> has <code>k</code> dimensions `[I, J, K, ..., N]`, then the output is a
+ tensor of rank `k+1` with dimensions [I, J, K, ..., N, N]` where:</p><p>`output[i, j, k, ..., m, n] = 1{m=n} * diagonal[i, j, k, ..., n]`.</p><p>For example:</p><p>```prettyprint
+ # <code>diagonal</code> is [[1, 2, 3, 4], [5, 6, 7, 8]]</p><p>and diagonal.shape = (2, 4)</p><p>tf.matrix_diag(diagonal) ==&gt; [[[1, 0, 0, 0]
+                                      [0, 2, 0, 0]
+                                      [0, 0, 3, 0]
+                                      [0, 0, 0, 4]],
+                                     [[5, 0, 0, 0]
+                                      [0, 6, 0, 0]
+                                      [0, 0, 7, 0]
+                                      [0, 0, 0, 8]]]</p><p>which has shape (2, 4, 4)
+ ```</p></div></div><div class="top"><p class="src"><a name="v:maxPool3D" class="def">maxPool3D</a> <a href="src/TensorFlow-GenOps-Core.html#maxPool3D" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Shape `[batch, depth, rows, cols, channels]` tensor to pool over.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The max pooled output tensor.</p></td></tr></table></div><div class="doc"><p>Performs 3D max pooling on the input.</p></div></div><div class="top"><p class="src"><a name="v:topK" class="def">topK</a> <a href="src/TensorFlow-GenOps-Core.html#topK" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>k</strong>: Number of top elements to look for along the last dimension (along each
+ row for matrices).</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: 1-D or higher with last dimension at least <code>k</code>.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>)</td><td class="doc"><p>(<strong>values</strong>, <strong>indices</strong>)</p><ul><li><strong>values</strong>: The <code>k</code> largest elements along each last dimensional slice.</li><li><strong>indices</strong>: The indices of <code>values</code> within the last dimension of <code>input</code>.</li></ul></td></tr></table></div><div class="doc"><p>Finds values and indices of the <code>k</code> largest elements for the last dimension.</p><p>If the input is a vector (rank-1), finds the <code>k</code> largest entries in the vector
+ and outputs their values and indices as vectors.  Thus `values[j]` is the
+ <code>j</code>-th largest entry in <code>input</code>, and its index is `indices[j]`.</p><p>For matrices (resp. higher rank input), computes the top <code>k</code> entries in each
+ row (resp. vector along the last dimension).  Thus,</p><p>values.shape = indices.shape = input.shape[:-1] + [k]</p><p>If two elements are equal, the lower-index element appears first.</p><p>If <code>k</code> varies dynamically, use <code>TopKV2</code> below.</p></div></div><div class="top"><p class="src"><a name="v:topKV2" class="def">topKV2</a> <a href="src/TensorFlow-GenOps-Core.html#topKV2" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: 1-D or higher with last dimension at least <code>k</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>k</strong>: 0-D.  Number of top elements to look for along the last dimension (along each
+ row for matrices).</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>)</td><td class="doc"><p>(<strong>values</strong>, <strong>indices</strong>)</p><ul><li><strong>values</strong>: The <code>k</code> largest elements along each last dimensional slice.</li><li><strong>indices</strong>: The indices of <code>values</code> within the last dimension of <code>input</code>.</li></ul></td></tr></table></div><div class="doc"><p>Finds values and indices of the <code>k</code> largest elements for the last dimension.</p><p>If the input is a vector (rank-1), finds the <code>k</code> largest entries in the vector
+ and outputs their values and indices as vectors.  Thus `values[j]` is the
+ <code>j</code>-th largest entry in <code>input</code>, and its index is `indices[j]`.</p><p>For matrices (resp. higher rank input), computes the top <code>k</code> entries in each
+ row (resp. vector along the last dimension).  Thus,</p><p>values.shape = indices.shape = input.shape[:-1] + [k]</p><p>If two elements are equal, the lower-index element appears first.</p><p>This is the same as <code>TopK</code>, but takes <code>k</code> as in input rather than an attr.</p></div></div><div class="top"><p class="src"><a name="v:fractionalMaxPool" class="def">fractionalMaxPool</a> <a href="src/TensorFlow-GenOps-Core.html#fractionalMaxPool" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>value</strong>: 4-D with shape `[batch, height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</td><td class="doc"><p>(<strong>output</strong>, <strong>row_pooling_sequence</strong>, <strong>col_pooling_sequence</strong>)</p><ul><li><strong>output</strong>: output tensor after fractional max pooling.</li><li><strong>row_pooling_sequence</strong>: row pooling sequence, needed to calculate gradient.</li><li><strong>col_pooling_sequence</strong>: column pooling sequence, needed to calculate gradient.</li></ul></td></tr></table></div><div class="doc"><p>Performs fractional max pooling on the input.</p><p>Fractional max pooling is slightly different than regular max pooling.  In
+ regular max pooling, you downsize an input set by taking the maximum value of
+ smaller N x N subsections of the set (often 2x2), and try to reduce the set by
+ a factor of N, where N is an integer.  Fractional max pooling, as you might
+ expect from the word &quot;fractional&quot;, means that the overall reduction ratio N
+ does not have to be an integer.</p><p>The sizes of the pooling regions are generated randomly but are fairly uniform.
+ For example, let's look at the height dimension, and the constraints on the
+ list of rows that will be pool boundaries.</p><p>First we define the following:</p><ol><li>input_row_length : the number of rows from the input set</li><li>output_row_length : which will be smaller than the input</li><li>alpha = input_row_length / output_row_length : our reduction ratio</li><li>K = floor(alpha)</li><li>row_pooling_sequence : this is the result list of pool boundary rows</li></ol><p>Then, row_pooling_sequence should satisfy:</p><ol><li>a[0] = 0 : the first value of the sequence is 0</li><li>a[end] = input_row_length : the last value of the sequence is the size</li><li>K &lt;= (a[i+1] - a[i]) &lt;= K+1 : all intervals are K or K+1 size</li><li>length(row_pooling_sequence) = output_row_length+1</li></ol><p>For more details on fractional max pooling, see this paper:
+ <a href="http://arxiv.org/abs/1412.6071">Benjamin Graham, Fractional Max-Pooling</a></p></div></div><div class="top"><p class="src"><a name="v:matrixBandPart" class="def">matrixBandPart</a> <a href="src/TensorFlow-GenOps-Core.html#matrixBandPart" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Rank <code>k</code> tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_lower</strong>: 0-D tensor. Number of subdiagonals to keep. If negative, keep entire
+ lower triangle.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_upper</strong>: 0-D tensor. Number of superdiagonals to keep. If negative, keep
+ entire upper triangle.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>band</strong>: Rank <code>k</code> tensor of the same shape as input. The extracted banded tensor.</p></td></tr></table></div><div class="doc"><p>Copy a tensor setting everything outside a central band in each innermost matrix</p><p>to zero.</p><p>The <code>band</code> part is computed as follows:
+ Assume <code>input</code> has <code>k</code> dimensions `[I, J, K, ..., M, N]`, then the output is a
+ tensor with the same shape where</p><p>`band[i, j, k, ..., m, n] = in_band(m, n) * input[i, j, k, ..., m, n]`.</p><p>The indicator function 'in_band(m, n)` is one if
+ `(num_lower &lt; 0 || (m-n) &lt;= num_lower)) &amp;&amp;
+ (num_upper &lt; 0 || (n-m) &lt;= num_upper)`, and zero otherwise.</p><p>For example:</p><p>```prettyprint
+ # if <code>input</code> is [[ 0,  1,  2, 3]
+                  [-1,  0,  1, 2]
+                  [-2, -1,  0, 1]
+                  [-3, -2, -1, 0]],</p><p>tf.matrix_band_part(input, 1, -1) ==&gt; [[ 0,  1,  2, 3]
+                                              [-1,  0,  1, 2]
+                                              [ 0, -1,  0, 1]
+                                              [ 0,  0, -1, 0]],</p><p>tf.matrix_band_part(input, 2, 1) ==&gt; [[ 0,  1,  0, 0]
+                                             [-1,  0,  1, 0]
+                                             [-2, -1,  0, 1]
+                                             [ 0, -2, -1, 0]]
+ ```</p><p>Useful special cases:</p><p>```prettyprint
+  tf.matrix_band_part(input, 0, -1) ==&gt; Upper triangular part.
+  tf.matrix_band_part(input, -1, 0) ==&gt; Lower triangular part.
+  tf.matrix_band_part(input, 0, 0) ==&gt; Diagonal.
+ ```</p></div></div><div class="top"><p class="src"><a name="v:decodeRaw" class="def">decodeRaw</a> <a href="src/TensorFlow-GenOps-Core.html#decodeRaw" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_type, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` out_type)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>bytes</strong>: All the elements must have the same length.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_type</td><td class="doc"><p><strong>output</strong>: A Tensor with one more dimension than the input <code>bytes</code>.  The
+ added dimension will have size equal to the length of the elements
+ of <code>bytes</code> divided by the number of bytes to represent <code>out_type</code>.</p></td></tr></table></div><div class="doc"><p>Reinterpret the bytes of a string as a vector of numbers.</p></div></div><div class="top"><p class="src"><a name="v:decodeJSONExample" class="def">decodeJSONExample</a> <a href="src/TensorFlow-GenOps-Core.html#decodeJSONExample" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>json_examples</strong>: Each string is a JSON object serialized according to the JSON
+ mapping of the Example proto.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>binary_examples</strong>: Each string is a binary Example protocol buffer corresponding
+ to the respective element of <code>json_examples</code>.</p></td></tr></table></div><div class="doc"><p>Convert JSON-encoded Example records to binary protocol buffer strings.</p><p>This op translates a tensor containing Example records, encoded using
+ the <a href="https://developers.google.com/protocol-buffers/docs/proto3#json">standard JSON
+ mapping</a>,
+ into a tensor containing the same records encoded as binary protocol
+ buffers. The resulting tensor can then be fed to any of the other
+ Example-parsing ops.</p></div></div><div class="top"><p class="src"><a name="v:truncatedNormal" class="def">truncatedNormal</a> <a href="src/TensorFlow-GenOps-Core.html#truncatedNormal" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` dtype)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>shape</strong>: The shape of the output tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</td><td class="doc"><p><strong>output</strong>: A tensor of the specified shape filled with random truncated normal
+ values.</p></td></tr></table></div><div class="doc"><p>Outputs random values from a truncated normal distribution.</p><p>The generated values follow a normal distribution with mean 0 and standard
+ deviation 1, except that values whose magnitude is more than 2 standard
+ deviations from the mean are dropped and re-picked.</p></div></div><div class="top"><p class="src"><a name="v:randomShuffle" class="def">randomShuffle</a> <a href="src/TensorFlow-GenOps-Core.html#randomShuffle" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>value</strong>: The tensor to be shuffled.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: A tensor of same shape and type as <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>, shuffled along its first
+ dimension.</p></td></tr></table></div><div class="doc"><p>Randomly shuffles a tensor along its first dimension.</p><p>The tensor is shuffled along dimension 0, such that each `value[j]` is mapped
+   to one and only one `output[i]`. For example, a mapping that might occur for a
+   3x2 tensor is:</p><p>```prettyprint
+ [[1, 2],       [[5, 6],
+  [3, 4],  ==&gt;   [1, 2],
+  [5, 6]]        [3, 4]]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:multinomial" class="def">multinomial</a> <a href="src/TensorFlow-GenOps-Core.html#multinomial" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>logits</strong>: 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
+ represents the unnormalized log probabilities for all classes.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>num_samples</strong>: 0-D.  Number of independent samples to draw for each row slice.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>output</strong>: 2-D Tensor with shape `[batch_size, num_samples]`.  Each slice `[i, :]`
+ contains the drawn class labels with range `[0, num_classes)`.</p></td></tr></table></div><div class="doc"><p>Draws samples from a multinomial distribution.</p></div></div><div class="top"><p class="src"><a name="v:randomGamma" class="def">randomGamma</a> <a href="src/TensorFlow-GenOps-Core.html#randomGamma" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> s, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` s, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 s</td><td class="doc"><p><strong>shape</strong>: 1-D integer tensor. Shape of independent samples to draw from each
+ distribution described by the shape parameters given in alpha.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>alpha</strong>: A tensor in which each scalar is a &quot;shape&quot; parameter describing the
+ associated gamma distribution.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: A tensor with shape `shape + shape(alpha)`. Each slice
+ `[:, ..., :, i0, i1, ...iN]` contains the samples drawn for
+ `alpha[i0, i1, ...iN]`. The dtype of the output matches the dtype of alpha.</p></td></tr></table></div><div class="doc"><p>Outputs random values from the Gamma distribution(s) described by alpha.</p><p>This op uses the algorithm by Marsaglia et al. to acquire samples via
+ transformation-rejection from pairs of uniform and normal random variables.
+ See <a href="http://dl.acm.org/citation.cfm?id=358414">http://dl.acm.org/citation.cfm?id=358414</a></p></div></div><div class="top"><p class="src"><a name="v:addN" class="def">addN</a> <a href="src/TensorFlow-GenOps-Core.html#addN" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t]</td><td class="doc"><p><strong>inputs</strong>: Must all be the same size and shape.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>sum</strong></p></td></tr></table></div><div class="doc"><p>Add all input tensors element wise.</p></div></div><div class="top"><p class="src"><a name="v:max" class="def">max</a> <a href="src/TensorFlow-GenOps-Core.html#max" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: The tensor to reduce.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>reduction_indices</strong>: The dimensions to reduce.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The reduced tensor.</p></td></tr></table></div><div class="doc"><p>Computes the maximum of elements across dimensions of a tensor.</p><p>Reduces <code>input</code> along the dimensions given in <code>reduction_indices</code>. Unless
+ <code>keep_dims</code> is true, the rank of the tensor is reduced by 1 for each entry in
+ <code>reduction_indices</code>. If <code>keep_dims</code> is true, the reduced dimensions are
+ retained with length 1.</p></div></div><div class="top"><p class="src"><a name="v:_Retval" class="def">_Retval</a> <a href="src/TensorFlow-GenOps-Core.html#_Retval" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>index</strong>: This return value is the index-th return value of the function.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: The return value.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>A graph node which represents a return value of a function.</p></div></div><div class="top"><p class="src"><a name="v:destroyTemporaryVariable" class="def">destroyTemporaryVariable</a> <a href="src/TensorFlow-GenOps-Core.html#destroyTemporaryVariable" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>ref</strong>: A reference to the temporary variable tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>value</strong></p></td></tr></table></div><div class="doc"><p>Destroys the temporary variable and returns its final value.</p><p>Sets output to the value of the Tensor pointed to by <code>ref</code>, then destroys
+ the temporary variable called <code>var_name</code>.
+ All other uses of <code>ref</code> *must* have executed before this op.
+ This is typically achieved by chaining the ref through each assign op, or by
+ using control dependencies.</p><p>Outputs the final value of the tensor pointed to by <code>ref</code>.</p></div></div><div class="top"><p class="src"><a name="v:cast" class="def">cast</a> <a href="src/TensorFlow-GenOps-Core.html#cast" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dstT, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> srcT)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 srcT</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dstT</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Cast x of type SrcT to y of DstT.</p></div></div><div class="top"><p class="src"><a name="v:countUpTo" class="def">countUpTo</a> <a href="src/TensorFlow-GenOps-Core.html#countUpTo" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>limit</strong>: If incrementing ref would bring it above limit, instead generates an
+ <code>OutOfRange</code> error.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>ref</strong>: Should be from a scalar <code>Variable</code> node.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: A copy of the input before increment. If nothing else modifies the
+ input, the values produced will all be distinct.</p></td></tr></table></div><div class="doc"><p>Increments <code>ref</code> until it reaches <code>limit</code>.</p><p>This operation outputs &quot;ref&quot; after the update is done.  This makes it
+ easier to chain operations that need to use the updated value.</p></div></div><div class="top"><p class="src"><a name="v:abs" class="def">abs</a> <a href="src/TensorFlow-GenOps-Core.html#abs" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes the absolute value of a tensor.</p><p>Given a tensor <code>x</code>, this operation returns a tensor containing the absolute
+ value of each element in <code>x</code>. For example, if x is an input element and y is
+ an output element, this operation computes \(y = |x|\).</p></div></div><div class="top"><p class="src"><a name="v:neg" class="def">neg</a> <a href="src/TensorFlow-GenOps-Core.html#neg" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes numerical negative value element-wise.</p><p>I.e., \(y = -x\).</p></div></div><div class="top"><p class="src"><a name="v:sparseSparseMaximum" class="def">sparseSparseMaximum</a> <a href="src/TensorFlow-GenOps-Core.html#sparseSparseMaximum" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>a_indices</strong>: 2-D.  `N x R` matrix with the indices of non-empty values in a
+ SparseTensor, in the canonical lexicographic ordering.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>a_values</strong>: 1-D.  <code>N</code> non-empty values corresponding to <code>a_indices</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>a_shape</strong>: 1-D.  Shape of the input SparseTensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>b_indices</strong>: counterpart to <code>a_indices</code> for the other operand.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>b_values</strong>: counterpart to <code>a_values</code> for the other operand; must be of the same dtype.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>b_shape</strong>: counterpart to <code>a_shape</code> for the other operand; the two shapes must be equal.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><td class="doc"><p>(<strong>output_indices</strong>, <strong>output_values</strong>)</p><ul><li><strong>output_indices</strong>: 2-D.  The indices of the output SparseTensor.</li><li><strong>output_values</strong>: 1-D.  The values of the output SparseTensor.</li></ul></td></tr></table></div><div class="doc"><p>Returns the element-wise max of two SparseTensors.</p><p>Assumes the two SparseTensors have the same shape, i.e., no broadcasting.</p></div></div><div class="top"><p class="src"><a name="v:invGrad" class="def">invGrad</a> <a href="src/TensorFlow-GenOps-Core.html#invGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Computes the gradient for the inverse of <code>x</code> wrt its input.</p><p>Specifically, `grad = -dy * y*y`, where `y = 1/x`, and <code>dy</code>
+ is the corresponding input gradient.</p></div></div><div class="top"><p class="src"><a name="v:sqrt" class="def">sqrt</a> <a href="src/TensorFlow-GenOps-Core.html#sqrt" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes square root of x element-wise.</p><p>I.e., \(y = sqrt{x} = x^{1/2}\).</p></div></div><div class="top"><p class="src"><a name="v:matrixInverse" class="def">matrixInverse</a> <a href="src/TensorFlow-GenOps-Core.html#matrixInverse" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Shape is `[..., M, M]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Shape is `[..., M, M]`.</p></td></tr></table></div><div class="doc"><p>Computes the inverse of one or more square invertible matrices or their</p><p>adjoints (conjugate transposes).</p><p>The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+ form square matrices. The output is a tensor of the same shape as the input
+ containing the inverse for all input submatrices `[..., :, :]`.</p><p>The op uses LU decomposition with partial pivoting to compute the inverses.</p><p>If a matrix is not invertible there is no guarantee what the op does. It
+ may detect the condition and raise an exception or it may simply return a
+ garbage result.</p></div></div><div class="top"><p class="src"><a name="v:sqrtGrad" class="def">sqrtGrad</a> <a href="src/TensorFlow-GenOps-Core.html#sqrtGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Computes the gradient for the sqrt of <code>x</code> wrt its input.</p><p>Specifically, `grad = dy * 0.5 / y`, where `y = sqrt(x)`, and <code>dy</code>
+ is the corresponding input gradient.</p></div></div><div class="top"><p class="src"><a name="v:expandDims" class="def">expandDims</a> <a href="src/TensorFlow-GenOps-Core.html#expandDims" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tdim, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tdim)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tdim</td><td class="doc"><p><strong>dim</strong>: 0-D (scalar). Specifies the dimension index at which to
+ expand the shape of <code>input</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Contains the same data as <code>input</code>, but its shape has an additional
+ dimension of size 1 added.</p></td></tr></table></div><div class="doc"><p>Inserts a dimension of 1 into a tensor's shape.</p><p>Given a tensor <code>input</code>, this operation inserts a dimension of 1 at the
+ dimension index <code>dim</code> of <code>input</code>'s shape. The dimension index <code>dim</code> starts at
+ zero; if you specify a negative number for <code>dim</code> it is counted backward from
+ the end.</p><p>This operation is useful if you want to add a batch dimension to a single
+ element. For example, if you have a single image of shape `[height, width,
+ channels]`, you can make it a batch of 1 image with `expand_dims(image, 0)`,
+ which will make the shape `[1, height, width, channels]`.</p><p>Other examples:</p><p>```prettyprint
+ # <code>t</code> is a tensor of shape [2]
+ shape(expand_dims(t, 0)) ==&gt; [1, 2]
+ shape(expand_dims(t, 1)) ==&gt; [2, 1]
+ shape(expand_dims(t, -1)) ==&gt; [2, 1]</p><p># <code>t2</code> is a tensor of shape [2, 3, 5]
+ shape(expand_dims(t2, 0)) ==&gt; [1, 2, 3, 5]
+ shape(expand_dims(t2, 2)) ==&gt; [2, 3, 1, 5]
+ shape(expand_dims(t2, 3)) ==&gt; [2, 3, 5, 1]
+ ```</p><p>This operation requires that:</p><p>`-1-input.dims() &lt;= dim &lt;= input.dims()`</p><p>This operation is related to `squeeze()`, which removes dimensions of
+ size 1.</p></div></div><div class="top"><p class="src"><a name="v:all" class="def">all</a> <a href="src/TensorFlow-GenOps-Core.html#all" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>input</strong>: The tensor to reduce.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>reduction_indices</strong>: The dimensions to reduce.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>output</strong>: The reduced tensor.</p></td></tr></table></div><div class="doc"><p>Computes the &quot;logical and&quot; of elements across dimensions of a tensor.</p><p>Reduces <code>input</code> along the dimensions given in <code>reduction_indices</code>. Unless
+ <code>keep_dims</code> is true, the rank of the tensor is reduced by 1 for each entry in
+ <code>reduction_indices</code>. If <code>keep_dims</code> is true, the reduced dimensions are
+ retained with length 1.</p></div></div><div class="top"><p class="src"><a name="v:cTCBeamSearchDecoder" class="def">cTCBeamSearchDecoder</a> <a href="src/TensorFlow-GenOps-Core.html#cTCBeamSearchDecoder" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>beam_width</strong>: A scalar &gt;= 0 (beam search beam width).</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>top_paths</strong>: A scalar &gt;= 0, &lt;= beam_width (controls output size).</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>inputs</strong>: 3-D, shape: `(max_time x batch_size x num_classes)`, the logits.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>sequence_length</strong>: A vector containing sequence lengths, size `(batch)`.</p></td></tr><tr><td class="src">-&gt; ([<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>], [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>], [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>], <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p>(<strong>decoded_indices</strong>, <strong>decoded_values</strong>, <strong>decoded_shape</strong>, <strong>log_probability</strong>)</p><ul><li><strong>decoded_indices</strong>: A list (length: top_paths) of indices matrices.  Matrix j,
+ size `(total_decoded_outputs[j] x 2)`, has indices of a
+ `SparseTensor<a href="int64,">2</a>`.  The rows store: [batch, time].</li><li><strong>decoded_values</strong>: A list (length: top_paths) of values vectors.  Vector j,
+ size `(length total_decoded_outputs[j])`, has the values of a
+ `SparseTensor<a href="int64,">2</a>`.  The vector stores the decoded classes for beam j.</li><li><strong>decoded_shape</strong>: A list (length: top_paths) of shape vector.  Vector j,
+ size `(2)`, stores the shape of the decoded `SparseTensor[j]`.
+ Its values are: `[batch_size, max_decoded_length[j]]`.</li><li><strong>log_probability</strong>: A matrix, shaped: `(batch_size x top_paths)`.  The
+ sequence log-probabilities.</li></ul></td></tr></table></div><div class="doc"><p>Performs beam search decoding on the logits given in input.</p><p>A note about the attribute merge_repeated: For the beam search decoder,
+ this means that if consecutive entries in a beam are the same, only
+ the first of these is emitted.  That is, when the top path is &quot;A B B B B&quot;,
+ &quot;A B&quot; is returned if merge_repeated = True but &quot;A B B B B&quot; is
+ returned if merge_repeated = False.</p></div></div><div class="top"><p class="src"><a name="v:rsqrt" class="def">rsqrt</a> <a href="src/TensorFlow-GenOps-Core.html#rsqrt" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes reciprocal of square root of x element-wise.</p><p>I.e., \(y = 1 / sqrt{x}\).</p></div></div><div class="top"><p class="src"><a name="v:tanhGrad" class="def">tanhGrad</a> <a href="src/TensorFlow-GenOps-Core.html#tanhGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Computes the gradient for the tanh of <code>x</code> wrt its input.</p><p>Specifically, `grad = dy * (1 - y*y)`, where `y = tanh(x)`, and <code>dy</code>
+ is the corresponding input gradient.</p></div></div><div class="top"><p class="src"><a name="v:sin" class="def">sin</a> <a href="src/TensorFlow-GenOps-Core.html#sin" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes sin of x element-wise.</p></div></div><div class="top"><p class="src"><a name="v:matrixDeterminant" class="def">matrixDeterminant</a> <a href="src/TensorFlow-GenOps-Core.html#matrixDeterminant" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Shape is `[..., M, M]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Shape is `[...]`.</p></td></tr></table></div><div class="doc"><p>Computes the determinant of one ore more square matrices.</p><p>The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+ form square matrices. The output is a tensor containing the determinants
+ for all input submatrices `[..., :, :]`.</p></div></div><div class="top"><p class="src"><a name="v:cos" class="def">cos</a> <a href="src/TensorFlow-GenOps-Core.html#cos" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes cos of x element-wise.</p></div></div><div class="top"><p class="src"><a name="v:batchToSpace" class="def">batchToSpace</a> <a href="src/TensorFlow-GenOps-Core.html#batchToSpace" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>block_size</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: 4-D tensor with shape
+ `[batch*block_size*block_size, height_pad<em>block_size, width_pad</em>block_size,
+   depth]`. Note that the batch size of the input tensor must be divisible by
+ `block_size * block_size`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>crops</strong>: 2-D tensor of non-negative integers with shape `[2, 2]`. It specifies
+ how many elements to crop from the intermediate result across the spatial
+ dimensions as follows:</p><p>crops = [[crop_top, crop_bottom], [crop_left, crop_right]]</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 4-D with shape `[batch, height, width, depth]`, where:</p><p>height = height_pad - crop_top - crop_bottom
+       width = width_pad - crop_left - crop_right</p><p>The attr <code>block_size</code> must be greater than one. It indicates the block size.</p><p>Some examples:</p><ol><li>For the following input of shape `[4, 1, 1, 1]` and block_size of 2:</li></ol><p>```prettyprint
+ [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
+ ```</p><p>The output tensor has shape `[1, 2, 2, 1]` and value:</p><p>```prettyprint
+ x = [[[[1], [2]], [[3], [4]]]]
+ ```</p><ol><li>For the following input of shape `[4, 1, 1, 3]` and block_size of 2:</li></ol><p>```prettyprint
+ [[[1, 2, 3]], [[4, 5, 6]], [[7, 8, 9]], [[10, 11, 12]]]
+ ```</p><p>The output tensor has shape `[1, 2, 2, 3]` and value:</p><p>```prettyprint
+ x = [[[[1, 2, 3], [4, 5, 6]],
+       [[7, 8, 9], [10, 11, 12]]]]
+ ```</p><ol><li>For the following input of shape `[4, 2, 2, 1]` and block_size of 2:</li></ol><p>```prettyprint
+ x = [[[[1], [3]], [[5], [7]]],
+      [[[2], [4]], [[10], [12]]],
+      [[[5], [7]], [[13], [15]]],
+      [[[6], [8]], [[14], [16]]]]
+ ```</p><p>The output tensor has shape `[1, 4, 4, 1]` and value:</p><p>```prettyprint
+ x = [[[1],   [2],  [3],  [4]],
+      [[5],   [6],  [7],  [8]],
+      [[9],  [10], [11],  [12]],
+      [[13], [14], [15],  [16]]]
+ ```</p><ol><li>For the following input of shape `[8, 1, 2, 1]` and block_size of 2:</li></ol><p>```prettyprint
+ x = [[[[1], [3]]], [[[9], [11]]], [[[2], [4]]], [[[10], [12]]],
+      [[[5], [7]]], [[[13], [15]]], [[[6], [8]]], [[[14], [16]]]]
+ ```</p><p>The output tensor has shape `[2, 2, 4, 1]` and value:</p><p>```prettyprint
+ x = [[[[1], [3]], [[5], [7]]],
+      [[[2], [4]], [[10], [12]]],
+      [[[5], [7]], [[13], [15]]],
+      [[[6], [8]], [[14], [16]]]]
+ ```</p></td></tr></table></div><div class="doc"><p>BatchToSpace for 4-D tensors of type T.</p><p>This is a legacy version of the more general BatchToSpaceND.</p><p>Rearranges (permutes) data from batch into blocks of spatial data, followed by
+ cropping. This is the reverse transformation of SpaceToBatch. More specifically,
+ this op outputs a copy of the input tensor where values from the <code>batch</code>
+ dimension are moved in spatial blocks to the <code>height</code> and <code>width</code> dimensions,
+ followed by cropping along the <code>height</code> and <code>width</code> dimensions.</p></div></div><div class="top"><p class="src"><a name="v:sparseToDense" class="def">sparseToDense</a> <a href="src/TensorFlow-GenOps-Core.html#sparseToDense" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tindices</td><td class="doc"><p><strong>sparse_indices</strong>: 0-D, 1-D, or 2-D.  `sparse_indices[i]` contains the complete
+ index where `sparse_values[i]` will be placed.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices</td><td class="doc"><p><strong>output_shape</strong>: 1-D.  Shape of the dense output tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>sparse_values</strong>: 1-D.  Values corresponding to each row of <code>sparse_indices</code>,
+ or a scalar value to be used for all sparse indices.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>default_value</strong>: Scalar value to set for indices not specified in
+ <code>sparse_indices</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>dense</strong>: Dense output tensor of shape <code>output_shape</code>.</p></td></tr></table></div><div class="doc"><p>Converts a sparse representation into a dense tensor.</p><p>Builds an array <code>dense</code> with shape <code>output_shape</code> such that</p><p>```prettyprint
+ # If sparse_indices is scalar
+ dense[i] = (i == sparse_indices ? sparse_values : default_value)</p><p># If sparse_indices is a vector, then for each i
+ dense[sparse_indices[i]] = sparse_values[i]</p><p># If sparse_indices is an n by d matrix, then for each i in [0, n)
+ dense[sparse_indices[i][0], ..., sparse_indices[i][d-1]] = sparse_values[i]
+ ```</p><p>All other values in <code>dense</code> are set to <code>default_value</code>.  If <code>sparse_values</code> is a
+ scalar, all sparse indices are set to this single value.</p><p>Indices should be sorted in lexicographic order, and indices must not
+ contain any repeats. If <code>validate_indices</code> is true, these properties
+ are checked during execution.</p></div></div><div class="top"><p class="src"><a name="v:asin" class="def">asin</a> <a href="src/TensorFlow-GenOps-Core.html#asin" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes asin of x element-wise.</p></div></div><div class="top"><p class="src"><a name="v:argMin" class="def">argMin</a> <a href="src/TensorFlow-GenOps-Core.html#argMin" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>dimension</strong>: int32, 0 &lt;= dimension &lt; rank(input).  Describes which dimension
+ of the input Tensor to reduce across. For vectors, use dimension = 0.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Returns the index with the smallest value across dimensions of a tensor.</p></div></div><div class="top"><p class="src"><a name="v:isInf" class="def">isInf</a> <a href="src/TensorFlow-GenOps-Core.html#isInf" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Returns which elements of x are Inf.</p></div></div><div class="top"><p class="src"><a name="v:sign" class="def">sign</a> <a href="src/TensorFlow-GenOps-Core.html#sign" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Returns an element-wise indication of the sign of a number.</p><p>`y = sign(x) = -1` if `x <a href="0`;">0 if `x == 0`; 1 if `x</a> 0`.</p><p>For complex numbers, `y = sign(x) = x / |x|` if `x != 0`, otherwise `y = 0`.</p></div></div><div class="top"><p class="src"><a name="v:add" class="def">add</a> <a href="src/TensorFlow-GenOps-Core.html#add" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns x + y element-wise.</p><ul><li>NOTE*: <code>Add</code> supports broadcasting. <code>AddN</code> does not. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:sparseApplyFtrl" class="def">sparseApplyFtrl</a> <a href="src/TensorFlow-GenOps-Core.html#sparseApplyFtrl" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>var</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>accum</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>linear</strong>: Should be from a Variable().</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>grad</strong>: The gradient.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 tindices</td><td class="doc"><p><strong>indices</strong>: A vector of indices into the first dimension of var and accum.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 t</td><td class="doc"><p><strong>lr</strong>: Scaling factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 t</td><td class="doc"><p><strong>l1</strong>: L1 regularization. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v8 t</td><td class="doc"><p><strong>l2</strong>: L2 regularization. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v9 t</td><td class="doc"><p><strong>lr_power</strong>: Scaling factor. Must be a scalar.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong>: Same as &quot;var&quot;.</p></td></tr></table></div><div class="doc"><p>Update relevant entries in '*var' according to the Ftrl-proximal scheme.</p><p>That is for rows we have grad for, we update var, accum and linear as follows:
+ accum_new = accum + grad * grad
+ linear += grad + (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
+ quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
+ var = (sign(linear) * l1 - linear) / quadratic if |linear| &gt; l1 else 0.0
+ accum = accum_new</p></div></div><div class="top"><p class="src"><a name="v:sub" class="def">sub</a> <a href="src/TensorFlow-GenOps-Core.html#sub" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns x - y element-wise.</p><ul><li>NOTE*: <code>Sub</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:batchFFT3D" class="def">batchFFT3D</a> <a href="src/TensorFlow-GenOps-Core.html#batchFFT3D" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:sparseReduceSumSparse" class="def">sparseReduceSumSparse</a> <a href="src/TensorFlow-GenOps-Core.html#sparseReduceSumSparse" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>input_indices</strong>: 2-D.  `N x R` matrix with the indices of non-empty values in a
+ SparseTensor, possibly not in canonical ordering.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>input_values</strong>: 1-D.  <code>N</code> non-empty values corresponding to <code>input_indices</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>input_shape</strong>: 1-D.  Shape of the input SparseTensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>reduction_axes</strong>: 1-D.  Length-<code>K</code> vector containing the reduction axes.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</td><td class="doc"><p>(<strong>output_indices</strong>, <strong>output_values</strong>, <strong>output_shape</strong>)</p><ul><li><strong>output_indices</strong></li><li><strong>output_values</strong></li><li><strong>output_shape</strong></li></ul></td></tr></table></div><div class="doc"><p>Computes the sum of elements across dimensions of a SparseTensor.</p><p>This Op takes a SparseTensor and is the sparse counterpart to
+ `tf.reduce_sum()`.  In contrast to SparseReduceSum, this Op returns a
+ SparseTensor.</p><p>Reduces <code>sp_input</code> along the dimensions given in <code>reduction_axes</code>.  Unless
+ <code>keep_dims</code> is true, the rank of the tensor is reduced by 1 for each entry in
+ <code>reduction_axes</code>. If <code>keep_dims</code> is true, the reduced dimensions are retained
+ with length 1.</p><p>If <code>reduction_axes</code> has no entries, all dimensions are reduced, and a tensor
+ with a single element is returned.  Additionally, the axes can be negative,
+ which are interpreted according to the indexing rules in Python.</p></div></div><div class="top"><p class="src"><a name="v:biasAdd" class="def">biasAdd</a> <a href="src/TensorFlow-GenOps-Core.html#biasAdd" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>value</strong>: Any number of dimensions.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>bias</strong>: 1-D with size the last dimension of <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Broadcasted sum of <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code> and <code>bias</code>.</p></td></tr></table></div><div class="doc"><p>Adds <code>bias</code> to <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>.</p><p>This is a special case of `tf.add` where <code>bias</code> is restricted to be 1-D.
+ Broadcasting is supported, so <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code> may have any number of dimensions.</p></div></div><div class="top"><p class="src"><a name="v:mul" class="def">mul</a> <a href="src/TensorFlow-GenOps-Core.html#mul" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns x * y element-wise.</p><ul><li>NOTE*: <code>Mul</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:div" class="def">div</a> <a href="src/TensorFlow-GenOps-Core.html#div" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns x / y element-wise.</p><ul><li>NOTE*: <code>Div</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:loopCond" class="def">loopCond</a> <a href="src/TensorFlow-GenOps-Core.html#loopCond" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>input</strong>: A boolean scalar, representing the branch predicate of the Switch op.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>output</strong>: The same tensor as <code>input</code>.</p></td></tr></table></div><div class="doc"><p>Forwards the input to the output.</p><p>This operator represents the loop termination condition used by the
+ &quot;pivot&quot; switches of a loop.</p></div></div><div class="top"><p class="src"><a name="v:squaredDifference" class="def">squaredDifference</a> <a href="src/TensorFlow-GenOps-Core.html#squaredDifference" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns (x - y)(x - y) element-wise.</p><ul><li>NOTE*: <code>SquaredDifference</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:maximum" class="def">maximum</a> <a href="src/TensorFlow-GenOps-Core.html#maximum" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns the max of x and y (i.e. x &gt; y ? x : y) element-wise.</p><ul><li>NOTE*: <code>Maximum</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:logUniformCandidateSampler" class="def">logUniformCandidateSampler</a> <a href="src/TensorFlow-GenOps-Core.html#logUniformCandidateSampler" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_sampled</strong>: Number of candidates to randomly sample per batch.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_true</strong>: Number of true labels per context.</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>range_max</strong>: The sampler will sample integers from the interval [0, range_max).</p></td></tr><tr><td class="src">-&gt; <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>unique</strong>: If unique is true, we sample with rejection, so that all sampled
+ candidates in a batch are unique. This requires some approximation to
+ estimate the post-rejection sampling probabilities.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>true_classes</strong>: A batch_size * num_true matrix, in which each row contains the
+ IDs of the num_true target_classes in the corresponding original label.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p>(<strong>sampled_candidates</strong>, <strong>true_expected_count</strong>, <strong>sampled_expected_count</strong>)</p><ul><li><strong>sampled_candidates</strong>: A vector of length num_sampled, in which each element is
+ the ID of a sampled candidate.</li><li><strong>true_expected_count</strong>: A batch_size * num_true matrix, representing
+ the number of times each candidate is expected to occur in a batch
+ of sampled candidates. If unique=true, then this is a probability.</li><li><strong>sampled_expected_count</strong>: A vector of length num_sampled, for each sampled
+ candidate representing the number of times the candidate is expected
+ to occur in a batch of sampled candidates.  If unique=true, then this is a
+ probability.</li></ul></td></tr></table></div><div class="doc"><p>Generates labels for candidate sampling with a log-uniform distribution.</p><p>See explanations of candidate sampling and the data formats at
+ go/candidate-sampling.</p><p>For each batch, this op picks a single set of sampled candidate labels.</p><p>The advantages of sampling candidates per-batch are simplicity and the
+ possibility of efficient dense matrix multiplication. The disadvantage is that
+ the sampled candidates must be chosen independently of the context and of the
+ true labels.</p></div></div><div class="top"><p class="src"><a name="v:less" class="def">less</a> <a href="src/TensorFlow-GenOps-Core.html#less" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns the truth value of (x &lt; y) element-wise.</p><ul><li>NOTE*: <code>Less</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:pow" class="def">pow</a> <a href="src/TensorFlow-GenOps-Core.html#pow" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Computes the power of one value to another.</p><p>Given a tensor <code>x</code> and a tensor <code>y</code>, this operation computes \(x^y\) for
+ corresponding elements in <code>x</code> and <code>y</code>. For example:</p><p>```
+ # tensor <code>x</code> is [[2, 2]], [3, 3]]
+ # tensor <code>y</code> is [[8, 16], [2, 3]]
+ tf.pow(x, y) ==&gt; [[256, 65536], [9, 27]]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:igammac" class="def">igammac</a> <a href="src/TensorFlow-GenOps-Core.html#igammac" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>a</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Compute the upper regularized incomplete Gamma function `Q(a, x)`.</p><p>The upper regularized incomplete Gamma function is defined as:</p><p>```
+ Q(a, x) = Gamma(a, x) / Gamma(a) = 1 - P(a, x)
+ ```
+ where
+ ```
+ Gamma(a, x) = int_{x}^{infty} t^{a-1} exp(-t) dt
+ ```
+ is the upper incomplete Gama function.</p><p>Note, above `P(a, x)` (<code>Igamma</code>) is the lower regularized complete
+ Gamma function.</p></div></div><div class="top"><p class="src"><a name="v:igamma" class="def">igamma</a> <a href="src/TensorFlow-GenOps-Core.html#igamma" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>a</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Compute the lower regularized incomplete Gamma function `Q(a, x)`.</p><p>The lower regularized incomplete Gamma function is defined as:</p><p>```
+ P(a, x) = gamma(a, x) / Gamma(a) = 1 - Q(a, x)
+ ```
+ where
+ ```
+ gamma(a, x) = int_{0}^{x} t^{a-1} exp(-t) dt
+ ```
+ is the lower incomplete Gamma function.</p><p>Note, above `Q(a, x)` (<code>Igammac</code>) is the upper regularized complete
+ Gamma function.</p></div></div><div class="top"><p class="src"><a name="v:zeta" class="def">zeta</a> <a href="src/TensorFlow-GenOps-Core.html#zeta" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>q</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Compute the Hurwitz zeta function \(zeta(x, q)\).</p><p>The Hurwitz zeta function is defined as:</p><p>```
+ zeta(x, q) = sum_{n=0}^{infty} (q + n)^{-x}
+ ```</p></div></div><div class="top"><p class="src"><a name="v:imag" class="def">imag</a> <a href="src/TensorFlow-GenOps-Core.html#imag" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tout, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` tout)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tout</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Returns the imaginary part of a complex number.</p><p>Given a tensor <code>input</code> of complex numbers, this operation returns a tensor of
+ type <code>float</code> that is the imaginary part of each element in <code>input</code>. All
+ elements in <code>input</code> must be complex numbers of the form \(a + bj\), where *a*
+ is the real part and *b* is the imaginary part returned by this operation.</p><p>For example:</p><p>```
+ # tensor <code>input</code> is [-2.25 + 4.75j, 3.25 + 5.75j]
+ tf.imag(input) ==&gt; [4.75, 5.75]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:complex" class="def">complex</a> <a href="src/TensorFlow-GenOps-Core.html#complex" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tout, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` tout)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>real</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>imag</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tout</td><td class="doc"><p><strong>out</strong></p></td></tr></table></div><div class="doc"><p>Converts two real numbers to a complex number.</p><p>Given a tensor <code><a href="TensorFlow-GenOps-Core.html#v:real">real</a></code> representing the real part of a complex number, and a
+ tensor <code><a href="TensorFlow-GenOps-Core.html#v:imag">imag</a></code> representing the imaginary part of a complex number, this
+ operation returns complex numbers elementwise of the form \(a + bj\), where
+ *a* represents the <code><a href="TensorFlow-GenOps-Core.html#v:real">real</a></code> part and *b* represents the <code><a href="TensorFlow-GenOps-Core.html#v:imag">imag</a></code> part.</p><p>The input tensors <code><a href="TensorFlow-GenOps-Core.html#v:real">real</a></code> and <code><a href="TensorFlow-GenOps-Core.html#v:imag">imag</a></code> must have the same shape.</p><p>For example:</p><p>```
+ # tensor <code><a href="TensorFlow-GenOps-Core.html#v:real">real</a></code> is [2.25, 3.25]
+ # tensor <code><a href="TensorFlow-GenOps-Core.html#v:imag">imag</a></code> is [4.75, 5.75]
+ tf.complex(real, imag) ==&gt; [[2.25 + 4.75j], [3.25 + 5.75j]]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:notEqual" class="def">notEqual</a> <a href="src/TensorFlow-GenOps-Core.html#notEqual" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a>, <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns the truth value of (x != y) element-wise.</p><ul><li>NOTE*: <code>NotEqual</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:complexAbs" class="def">complexAbs</a> <a href="src/TensorFlow-GenOps-Core.html#complexAbs" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tout, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` tout)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tout</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes the complex absolute value of a tensor.</p><p>Given a tensor <code>x</code> of complex numbers, this operation returns a tensor of type
+ <code>float</code> or <code>double</code> that is the absolute value of each element in <code>x</code>. All
+ elements in <code>x</code> must be complex numbers of the form \(a + bj\). The absolute
+ value is computed as \( sqrt{a^2 + b^2}\).</p><p>For example:</p><p>```
+ # tensor <code>x</code> is [[-2.25 + 4.75j], [-3.25 + 5.75j]]
+ tf.complex_abs(x) ==&gt; [5.25594902, 6.60492229]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:logicalAnd" class="def">logicalAnd</a> <a href="src/TensorFlow-GenOps-Core.html#logicalAnd" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns the truth value of x AND y element-wise.</p><ul><li>NOTE*: <code>LogicalAnd</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:batchFFT" class="def">batchFFT</a> <a href="src/TensorFlow-GenOps-Core.html#batchFFT" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:select" class="def">select</a> <a href="src/TensorFlow-GenOps-Core.html#select" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>condition</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>t</strong>: = A <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> which may have the same shape as <code>condition</code>.
+ If <code>condition</code> is rank 1, <code>t</code> may have higher rank,
+ but its first dimension must match the size of <code>condition</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>e</strong>: = A <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> with the same type and shape as <code>t</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: = A <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> with the same type and shape as <code>t</code> and <code>e</code>.</p></td></tr></table></div><div class="doc"><p>Selects elements from <code>t</code> or <code>e</code>, depending on <code>condition</code>.</p><p>The <code>t</code>, and <code>e</code> tensors must all have the same shape,
+ and the output will also have that shape.  The <code>condition</code> tensor
+ must be a scalar if <code>t</code> and <code>e</code> are scalars.  If <code>t</code> and <code>e</code> are vectors
+ or higher rank, then <code>condition</code> must be either a vector with size
+ matching the first dimension of <code>t</code>, or must have the same shape as <code>t</code>.</p><p>The <code>condition</code> tensor acts as a mask that chooses, based on the value at each
+ element, whether the corresponding element / row in the output should be
+ taken from <code>t</code> (if true) or <code>e</code> (if false).</p><p>If <code>condition</code> is a vector and <code>t</code> and <code>e</code> are higher rank matrices, then
+ it chooses which row (outer dimension) to copy from <code>t</code> and <code>e</code>.
+ If <code>condition</code> has the same shape as <code>t</code> and <code>e</code>, then it chooses which
+ element to copy from <code>t</code> and <code>e</code>.</p><p>For example:</p><p>```prettyprint
+ # <code>condition</code> tensor is [[True,  False]
+ #                        [False, True]]
+ # <code>t</code> is [[1, 2],
+ #         [3, 4]]
+ # <code>e</code> is [[5, 6],
+ #         [7, 8]]
+ select(condition, t, e) ==&gt; [[1, 6],
+                              [7, 4]]</p><p># <code>condition</code> tensor is [True, False]
+ # <code>t</code> is [[1, 2],
+ #         [3, 4]]
+ # <code>e</code> is [[5, 6],
+ #         [7, 8]]
+ select(condition, t, e) ==&gt; [[1, 2],
+                              [7, 8]]</p><p>```</p></div></div><div class="top"><p class="src"><a name="v:matMul" class="def">matMul</a> <a href="src/TensorFlow-GenOps-Core.html#matMul" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>a</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>b</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>product</strong></p></td></tr></table></div><div class="doc"><p>Multiply the matrix &quot;a&quot; by the matrix &quot;b&quot;.</p><p>The inputs must be two-dimensional matrices and the inner dimension of
+ &quot;a&quot; (after being transposed if transpose_a is true) must match the
+ outer dimension of &quot;b&quot; (after being transposed if transposed_b is
+ true).</p><ul><li>Note*: The default kernel implementation for MatMul on GPUs uses
+ cublas.</li></ul></div></div><div class="top"><p class="src"><a name="v:digamma" class="def">digamma</a> <a href="src/TensorFlow-GenOps-Core.html#digamma" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes Psi, the derivative of Lgamma (the log of the absolute value of</p><p>`Gamma(x)`), element-wise.</p></div></div><div class="top"><p class="src"><a name="v:conv2DBackpropFilter" class="def">conv2DBackpropFilter</a> <a href="src/TensorFlow-GenOps-Core.html#conv2DBackpropFilter" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: 4-D with shape `[batch, in_height, in_width, in_channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>filter_sizes</strong>: An integer vector representing the tensor shape of <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code>,
+ where <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code> is a 4-D
+ `[filter_height, filter_width, in_channels, out_channels]` tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>out_backprop</strong>: 4-D with shape `[batch, out_height, out_width, out_channels]`.
+ Gradients w.r.t. the output of the convolution.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 4-D with shape
+ `[filter_height, filter_width, in_channels, out_channels]`.  Gradient w.r.t.
+ the <code><a href="../base-4.8.2.0/GHC-OldList.html#v:filter">filter</a></code> input of the convolution.</p></td></tr></table></div><div class="doc"><p>Computes the gradients of convolution with respect to the filter.</p></div></div><div class="top"><p class="src"><a name="v:min" class="def">min</a> <a href="src/TensorFlow-GenOps-Core.html#min" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: The tensor to reduce.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>reduction_indices</strong>: The dimensions to reduce.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The reduced tensor.</p></td></tr></table></div><div class="doc"><p>Computes the minimum of elements across dimensions of a tensor.</p><p>Reduces <code>input</code> along the dimensions given in <code>reduction_indices</code>. Unless
+ <code>keep_dims</code> is true, the rank of the tensor is reduced by 1 for each entry in
+ <code>reduction_indices</code>. If <code>keep_dims</code> is true, the reduced dimensions are
+ retained with length 1.</p></div></div><div class="top"><p class="src"><a name="v:isFinite" class="def">isFinite</a> <a href="src/TensorFlow-GenOps-Core.html#isFinite" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Returns which elements of x are finite.</p></div></div><div class="top"><p class="src"><a name="v:argMax" class="def">argMax</a> <a href="src/TensorFlow-GenOps-Core.html#argMax" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>dimension</strong>: int32, 0 &lt;= dimension &lt; rank(input).  Describes which dimension
+ of the input Tensor to reduce across. For vectors, use dimension = 0.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Returns the index with the largest value across dimensions of a tensor.</p></div></div><div class="top"><p class="src"><a name="v:segmentMean" class="def">segmentMean</a> <a href="src/TensorFlow-GenOps-Core.html#segmentMean" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices</td><td class="doc"><p><strong>segment_ids</strong>: A 1-D tensor whose rank is equal to the rank of `data`'s
+ first dimension.  Values should be sorted and can be repeated.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Has same shape as data, except for dimension 0 which
+ has size <code>k</code>, the number of segments.</p></td></tr></table></div><div class="doc"><p>Computes the mean along segments of a tensor.</p><p>Read <a href="../../api_docs/python/math_ops.md#segmentation">the section on
+ Segmentation</a> for an explanation
+ of segments.</p><p>Computes a tensor such that
+ \(output_i = frac{sum_j data_j}{N}\) where <code><a href="TensorFlow-GenOps-Core.html#v:mean">mean</a></code> is
+ over <code>j</code> such that `segment_ids[j] == i` and <code>N</code> is the total number of
+ values summed.</p><p><a href="div">style=&quot;width:70%; margin:auto; margin-bottom:10px; margin-top:20px;&quot;</a>
+ <a href="img">style=&quot;width:100%&quot; src=&quot;../../images/SegmentMean.png&quot; alt</a>
+ <a href="/div">/div</a></p></div></div><div class="top"><p class="src"><a name="v:cumprod" class="def">cumprod</a> <a href="src/TensorFlow-GenOps-Core.html#cumprod" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>axis</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong></p></td></tr></table></div><div class="doc"><p>Compute the cumulative product of the tensor <code>x</code> along <code>axis</code>.</p><p>By default, this op performs an inclusive cumprod, which means that the first
+ element of the input is identical to the first element of the output:
+ ```prettyprint
+ tf.cumprod([a, b, c]) ==&gt; [a, a * b, a * b * c]
+ ```</p><p>By setting the <code>exclusive</code> kwarg to <code><a href="../base-4.8.2.0/Data-Bool.html#v:True">True</a></code>, an exclusive cumprod is
+ performed instead:
+ ```prettyprint
+ tf.cumprod([a, b, c], exclusive=True) ==&gt; [0, a, a * b]
+ ```</p><p>By setting the <code><a href="../base-4.8.2.0/GHC-OldList.html#v:reverse">reverse</a></code> kwarg to <code><a href="../base-4.8.2.0/Data-Bool.html#v:True">True</a></code>, the cumprod is performed in the
+ opposite direction:
+ ```prettyprint
+ tf.cumprod([a, b, c], reverse=True) ==&gt; [a * b * c, b * c, c]
+ ```
+ This is more efficient than using separate `tf.reverse` ops.</p><p>The <code><a href="../base-4.8.2.0/GHC-OldList.html#v:reverse">reverse</a></code> and <code>exclusive</code> kwargs can also be combined:
+ ```prettyprint
+ tf.cumprod([a, b, c], exclusive=True, reverse=True) ==&gt; [b * c, c, 0]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:segmentMin" class="def">segmentMin</a> <a href="src/TensorFlow-GenOps-Core.html#segmentMin" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices</td><td class="doc"><p><strong>segment_ids</strong>: A 1-D tensor whose rank is equal to the rank of `data`'s
+ first dimension.  Values should be sorted and can be repeated.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Has same shape as data, except for dimension 0 which
+ has size <code>k</code>, the number of segments.</p></td></tr></table></div><div class="doc"><p>Computes the minimum along segments of a tensor.</p><p>Read <a href="../../api_docs/python/math_ops.md#segmentation">the section on
+ Segmentation</a> for an explanation
+ of segments.</p><p>Computes a tensor such that
+ \(output_i = min_j(data_j)\) where <code><a href="../base-4.8.2.0/Data-Ord.html#v:min">min</a></code> is over <code>j</code> such
+ that `segment_ids[j] == i`.</p><p><a href="div">style=&quot;width:70%; margin:auto; margin-bottom:10px; margin-top:20px;&quot;</a>
+ <a href="img">style=&quot;width:100%&quot; src=&quot;../../images/SegmentMin.png&quot; alt</a>
+ <a href="/div">/div</a></p></div></div><div class="top"><p class="src"><a name="v:unsortedSegmentSum" class="def">unsortedSegmentSum</a> <a href="src/TensorFlow-GenOps-Core.html#unsortedSegmentSum" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices</td><td class="doc"><p><strong>segment_ids</strong>: A tensor whose shape is a prefix of `data.shape`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>num_segments</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Has same shape as data, except for the first `segment_ids.rank`
+ dimensions, which are replaced with a single dimension which has size
+ <code>num_segments</code>.</p></td></tr></table></div><div class="doc"><p>Computes the sum along segments of a tensor.</p><p>Read <a href="../../api_docs/python/math_ops.md#segmentation">the section on
+ Segmentation</a> for an explanation
+ of segments.</p><p>Computes a tensor such that
+ `(output[i] = sum_{j...} data[j...]` where the sum is over tuples `j...` such
+ that `segment_ids[j...] == i`.  Unlike <code>SegmentSum</code>, <code>segment_ids</code>
+ need not be sorted and need not cover all values in the full
+ range of valid values.</p><p>If the sum is empty for a given segment ID <code>i</code>, `output[i] = 0`.</p><p><code>num_segments</code> should equal the number of distinct segment IDs.</p><p><a href="div">style=&quot;width:70%; margin:auto; margin-bottom:10px; margin-top:20px;&quot;</a>
+ <a href="img">style=&quot;width:100%&quot; src=&quot;../../images/UnsortedSegmentSum.png&quot; alt</a>
+ <a href="/div">/div</a></p></div></div><div class="top"><p class="src"><a name="v:tFRecordReader" class="def">tFRecordReader</a> <a href="src/TensorFlow-GenOps-Core.html#tFRecordReader" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>reader_handle</strong>: The handle to reference the Reader.</p></td></tr></table></div><div class="doc"><p>A Reader that outputs the records from a TensorFlow Records file.</p></div></div><div class="top"><p class="src"><a name="v:sparseSegmentSum" class="def">sparseSegmentSum</a> <a href="src/TensorFlow-GenOps-Core.html#sparseSegmentSum" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>indices</strong>: A 1-D tensor. Has same rank as <code>segment_ids</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>segment_ids</strong>: A 1-D tensor. Values should be sorted and can be repeated.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Has same shape as data, except for dimension 0 which
+ has size <code>k</code>, the number of segments.</p></td></tr></table></div><div class="doc"><p>Computes the sum along sparse segments of a tensor.</p><p>Read <a href="../../api_docs/python/math_ops.md#segmentation">the section on
+ Segmentation</a> for an explanation
+ of segments.</p><p>Like <code>SegmentSum</code>, but <code>segment_ids</code> can have rank less than `data`'s first
+ dimension, selecting a subset of dimension 0, specified by <code>indices</code>.</p><p>For example:</p><p>```prettyprint
+ c = tf.constant([[1,2,3,4], [-1,-2,-3,-4], [5,6,7,8]])</p><p># Select two rows, one segment.
+ tf.sparse_segment_sum(c, tf.constant([0, 1]), tf.constant([0, 0]))
+   ==&gt; [[0 0 0 0]]</p><p># Select two rows, two segment.
+ tf.sparse_segment_sum(c, tf.constant([0, 1]), tf.constant([0, 1]))
+   ==&gt; [[ 1  2  3  4]
+        [-1 -2 -3 -4]]</p><p># Select all rows, two segments.
+ tf.sparse_segment_sum(c, tf.constant([0, 1, 2]), tf.constant([0, 0, 1]))
+   ==&gt; [[0 0 0 0]
+        [5 6 7 8]]</p><p># Which is equivalent to:
+ tf.segment_sum(c, tf.constant([0, 0, 1]))
+ ```</p></div></div><div class="top"><p class="src"><a name="v:sparseSegmentSqrtN" class="def">sparseSegmentSqrtN</a> <a href="src/TensorFlow-GenOps-Core.html#sparseSegmentSqrtN" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>indices</strong>: A 1-D tensor. Has same rank as <code>segment_ids</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>segment_ids</strong>: A 1-D tensor. Values should be sorted and can be repeated.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Has same shape as data, except for dimension 0 which
+ has size <code>k</code>, the number of segments.</p></td></tr></table></div><div class="doc"><p>Computes the sum along sparse segments of a tensor divided by the sqrt of N.</p><p>N is the size of the segment being reduced.</p><p>Read <a href="../../api_docs/python/math_ops.md#segmentation">the section on
+ Segmentation</a> for an explanation
+ of segments.</p></div></div><div class="top"><p class="src"><a name="v:copyHost" class="def">copyHost</a> <a href="src/TensorFlow-GenOps-Core.html#copyHost" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: Input tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Output tensor, deep-copied from input.</p></td></tr></table></div><div class="doc"><p>Copy Host Op.</p><p>Performs CPU-to-CPU deep-copying of tensor.</p><p>Unlike the Copy Op, this op has HostMemory constraint on its input or output.</p></div></div><div class="top"><p class="src"><a name="v:variable" class="def">variable</a> <a href="src/TensorFlow-GenOps-Core.html#variable" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dtype</td><td class="doc"><p><strong>ref</strong>: A reference to the variable tensor.</p></td></tr></table></div><div class="doc"><p>Holds state in the form of a tensor that persists across steps.</p><p>Outputs a ref to the tensor state so it may be read or modified.
+ TODO(zhifengc/mrry): Adds a pointer to a more detail document
+ about sharing states in tensorflow.</p></div></div><div class="top"><p class="src"><a name="v:sparseSegmentSqrtNGrad" class="def">sparseSegmentSqrtNGrad</a> <a href="src/TensorFlow-GenOps-Core.html#sparseSegmentSqrtNGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>grad</strong>: gradient propagated to the SparseSegmentSqrtN op.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>indices</strong>: indices passed to the corresponding SparseSegmentSqrtN op.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>segment_ids</strong>: segment_ids passed to the corresponding SparseSegmentSqrtN op.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>output_dim0</strong>: dimension 0 of &quot;data&quot; passed to SparseSegmentSqrtN op.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Computes gradients for SparseSegmentSqrtN.</p><p>Returns tensor &quot;output&quot; with same shape as grad, except for dimension 0 whose
+ value is output_dim0.</p></div></div><div class="top"><p class="src"><a name="v:range" class="def">range</a> <a href="src/TensorFlow-GenOps-Core.html#range" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tidx</td><td class="doc"><p><strong>start</strong>: 0-D (scalar). First entry in the sequence.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>limit</strong>: 0-D (scalar). Upper limit of sequence, exclusive.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tidx</td><td class="doc"><p><strong>delta</strong>: 0-D (scalar). Optional. Default is 1. Number that increments <code>start</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tidx</td><td class="doc"><p><strong>output</strong>: 1-D.</p></td></tr></table></div><div class="doc"><p>Creates a sequence of integers.</p><p>This operation creates a sequence of integers that begins at <code>start</code> and
+ extends by increments of <code>delta</code> up to but not including <code>limit</code>.</p><p>For example:</p><p>```
+ # <code>start</code> is 3
+ # <code>limit</code> is 18
+ # <code>delta</code> is 3
+ tf.range(start, limit, delta) ==&gt; [3, 6, 9, 12, 15]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:any" class="def">any</a> <a href="src/TensorFlow-GenOps-Core.html#any" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>input</strong>: The tensor to reduce.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>reduction_indices</strong>: The dimensions to reduce.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></td><td class="doc"><p><strong>output</strong>: The reduced tensor.</p></td></tr></table></div><div class="doc"><p>Computes the &quot;logical or&quot; of elements across dimensions of a tensor.</p><p>Reduces <code>input</code> along the dimensions given in <code>reduction_indices</code>. Unless
+ <code>keep_dims</code> is true, the rank of the tensor is reduced by 1 for each entry in
+ <code>reduction_indices</code>. If <code>keep_dims</code> is true, the reduced dimensions are
+ retained with length 1.</p></div></div><div class="top"><p class="src"><a name="v:linSpace" class="def">linSpace</a> <a href="src/TensorFlow-GenOps-Core.html#linSpace" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>start</strong>: First entry in the range.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>stop</strong>: Last entry in the range.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tidx</td><td class="doc"><p><strong>num</strong>: Number of values to generate.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 1-D. The generated values.</p></td></tr></table></div><div class="doc"><p>Generates values in an interval.</p><p>A sequence of <code>num</code> evenly-spaced values are generated beginning at <code>start</code>.
+ If `num &gt; 1`, the values in the sequence increase by `stop - start / num - 1`,
+ so that the last one is exactly <code>stop</code>.</p><p>For example:</p><p>```
+ tf.linspace(10.0, 12.0, 3, name=&quot;linspace&quot;) =&gt; [ 10.0  11.0  12.0]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:resizeArea" class="def">resizeArea</a> <a href="src/TensorFlow-GenOps-Core.html#resizeArea" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>images</strong>: 4-D with shape `[batch, height, width, channels]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>size</strong>: = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
+ new size for the images.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></td><td class="doc"><p><strong>resized_images</strong>: 4-D with shape
+ `[batch, new_height, new_width, channels]`.</p></td></tr></table></div><div class="doc"><p>Resize <code>images</code> to <code><a href="TensorFlow-GenOps-Core.html#v:size">size</a></code> using area interpolation.</p><p>Input images can be of different types but output images are always float.</p></div></div><div class="top"><p class="src"><a name="v:real" class="def">real</a> <a href="src/TensorFlow-GenOps-Core.html#real" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tout, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` tout)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tout</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Returns the real part of a complex number.</p><p>Given a tensor <code>input</code> of complex numbers, this operation returns a tensor of
+ type <code>float</code> that is the real part of each element in <code>input</code>. All elements in
+ <code>input</code> must be complex numbers of the form \(a + bj\), where *a* is the real
+  part returned by this operation and *b* is the imaginary part.</p><p>For example:</p><p>```
+ # tensor <code>input</code> is [-2.25 + 4.75j, 3.25 + 5.75j]
+ tf.real(input) ==&gt; [-2.25, 3.25]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:iFFT" class="def">iFFT</a> <a href="src/TensorFlow-GenOps-Core.html#iFFT" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>input</strong>: A complex64 tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>output</strong>: A complex64 tensor of the same shape as <code>input</code>. The inner-most
+ dimension of <code>input</code> is replaced with its inverse 1D Fourier Transform.</p></td></tr></table></div><div class="doc"><p>Compute the inverse 1-dimensional discrete Fourier Transform over the inner-most</p><p>dimension of <code>input</code>.</p></div></div><div class="top"><p class="src"><a name="v:iFFT3D" class="def">iFFT3D</a> <a href="src/TensorFlow-GenOps-Core.html#iFFT3D" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>input</strong>: A complex64 tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>output</strong>: A complex64 tensor of the same shape as <code>input</code>. The inner-most 3
+ dimensions of <code>input</code> are replaced with their inverse 3D Fourier Transform.</p></td></tr></table></div><div class="doc"><p>Compute the inverse 3-dimensional discrete Fourier Transform over the inner-most</p><p>3 dimensions of <code>input</code>.</p></div></div><div class="top"><p class="src"><a name="v:cross" class="def">cross</a> <a href="src/TensorFlow-GenOps-Core.html#cross" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>a</strong>: A tensor containing 3-element vectors.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>b</strong>: Another tensor, of same type and shape as <code>a</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>product</strong>: Pairwise cross product of the vectors in <code>a</code> and <code>b</code>.</p></td></tr></table></div><div class="doc"><p>Compute the pairwise cross product.</p><p><code>a</code> and <code>b</code> must be the same shape; they can either be simple 3-element vectors,
+ or any shape where the innermost dimension is 3. In the latter case, each pair
+ of corresponding 3-element vectors is cross-multiplied independently.</p></div></div><div class="top"><p class="src"><a name="v:cumsum" class="def">cumsum</a> <a href="src/TensorFlow-GenOps-Core.html#cumsum" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>axis</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>out</strong></p></td></tr></table></div><div class="doc"><p>Compute the cumulative sum of the tensor <code>x</code> along <code>axis</code>.</p><p>By default, this op performs an inclusive cumsum, which means that the first
+ element of the input is identical to the first element of the output:
+ ```prettyprint
+ tf.cumsum([a, b, c]) ==&gt; [a, a + b, a + b + c]
+ ```</p><p>By setting the <code>exclusive</code> kwarg to <code><a href="../base-4.8.2.0/Data-Bool.html#v:True">True</a></code>, an exclusive cumsum is
+ performed instead:
+ ```prettyprint
+ tf.cumsum([a, b, c], exclusive=True) ==&gt; [0, a, a + b]
+ ```</p><p>By setting the <code><a href="../base-4.8.2.0/GHC-OldList.html#v:reverse">reverse</a></code> kwarg to <code><a href="../base-4.8.2.0/Data-Bool.html#v:True">True</a></code>, the cumsum is performed in the
+ opposite direction:
+ ```prettyprint
+ tf.cumsum([a, b, c], reverse=True) ==&gt; [a + b + c, b + c, c]
+ ```
+ This is more efficient than using separate `tf.reverse` ops.</p><p>The <code><a href="../base-4.8.2.0/GHC-OldList.html#v:reverse">reverse</a></code> and <code>exclusive</code> kwargs can also be combined:
+ ```prettyprint
+ tf.cumsum([a, b, c], exclusive=True, reverse=True) ==&gt; [b + c, c, 0]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:batchIFFT" class="def">batchIFFT</a> <a href="src/TensorFlow-GenOps-Core.html#batchIFFT" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:erf" class="def">erf</a> <a href="src/TensorFlow-GenOps-Core.html#erf" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes the Gauss error function of <code>x</code> element-wise.</p></div></div><div class="top"><p class="src"><a name="v:barrierInsertMany" class="def">barrierInsertMany</a> <a href="src/TensorFlow-GenOps-Core.html#barrierInsertMany" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>component_index</strong>: The component of the barrier elements that is being assigned.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>handle</strong>: The handle to a barrier.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>keys</strong>: A one-dimensional tensor of keys, with length n.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>values</strong>: An any-dimensional tensor of values, which are associated with the
+ respective keys. The 0th dimension must have length n.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>For each key, assigns the respective value to the specified component.</p><p>If a key is not found in the barrier, this operation will create a new
+ incomplete element. If a key is found in the barrier, and the element
+ already has a value at component_index, this operation will fail with
+ INVALID_ARGUMENT, and leave the barrier in an undefined state.</p></div></div><div class="top"><p class="src"><a name="v:floor" class="def">floor</a> <a href="src/TensorFlow-GenOps-Core.html#floor" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Returns element-wise largest integer not greater than x.</p></div></div><div class="top"><p class="src"><a name="v:batchFFT2D" class="def">batchFFT2D</a> <a href="src/TensorFlow-GenOps-Core.html#batchFFT2D" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:sparseAddGrad" class="def">sparseAddGrad</a> <a href="src/TensorFlow-GenOps-Core.html#sparseAddGrad" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>backprop_val_grad</strong>: 1-D with shape `[nnz(sum)]`.  The gradient with respect to
+ the non-empty values of the sum.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>a_indices</strong>: 2-D.  The <code>indices</code> of the <code>SparseTensor</code> A, size `[nnz(A), ndims]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>b_indices</strong>: 2-D.  The <code>indices</code> of the <code>SparseTensor</code> B, size `[nnz(B), ndims]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>sum_indices</strong>: 2-D.  The <code>indices</code> of the sum <code>SparseTensor</code>, size
+ `[nnz(sum), ndims]`.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><td class="doc"><p>(<strong>a_val_grad</strong>, <strong>b_val_grad</strong>)</p><ul><li><strong>a_val_grad</strong>: 1-D with shape `[nnz(A)]`. The gradient with respect to the
+ non-empty values of A.</li><li><strong>b_val_grad</strong>: 1-D with shape `[nnz(B)]`. The gradient with respect to the
+ non-empty values of B.</li></ul></td></tr></table></div><div class="doc"><p>The gradient operator for the SparseAdd op.</p><p>The SparseAdd op calculates A + B, where A, B, and the sum are all represented
+ as <code>SparseTensor</code> objects.  This op takes in the upstream gradient w.r.t.
+ non-empty values of the sum, and outputs the gradients w.r.t. the non-empty
+ values of A and B.</p></div></div><div class="top"><p class="src"><a name="v:sparseAdd" class="def">sparseAdd</a> <a href="src/TensorFlow-GenOps-Core.html#sparseAdd" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> treal, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` treal)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>a_indices</strong>: 2-D.  The <code>indices</code> of the first <code>SparseTensor</code>, size `[nnz, ndims]` Matrix.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>a_values</strong>: 1-D.  The <code>values</code> of the first <code>SparseTensor</code>, size `[nnz]` Vector.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>a_shape</strong>: 1-D.  The <code><a href="TensorFlow-GenOps-Core.html#v:shape">shape</a></code> of the first <code>SparseTensor</code>, size `[ndims]` Vector.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>b_indices</strong>: 2-D.  The <code>indices</code> of the second <code>SparseTensor</code>, size `[nnz, ndims]` Matrix.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v5 t</td><td class="doc"><p><strong>b_values</strong>: 1-D.  The <code>values</code> of the second <code>SparseTensor</code>, size `[nnz]` Vector.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v6 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>b_shape</strong>: 1-D.  The <code><a href="TensorFlow-GenOps-Core.html#v:shape">shape</a></code> of the second <code>SparseTensor</code>, size `[ndims]` Vector.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v7 treal</td><td class="doc"><p><strong>thresh</strong>: 0-D.  The magnitude threshold that determines if an output value/index
+ pair takes space.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</td><td class="doc"><p>(<strong>sum_indices</strong>, <strong>sum_values</strong>, <strong>sum_shape</strong>)</p><ul><li><strong>sum_indices</strong></li><li><strong>sum_values</strong></li><li><strong>sum_shape</strong></li></ul></td></tr></table></div><div class="doc"><p>Adds two <code>SparseTensor</code> objects to produce another <code>SparseTensor</code>.</p><p>The input <code>SparseTensor</code> objects' indices are assumed ordered in standard
+ lexicographic order.  If this is not the case, before this step run
+ <code>SparseReorder</code> to restore index ordering.</p><p>By default, if two values sum to zero at some index, the output <code>SparseTensor</code>
+ would still include that particular location in its index, storing a zero in the
+ corresponding value slot.  To override this, callers can specify <code>thresh</code>,
+ indicating that if the sum has a magnitude strictly smaller than <code>thresh</code>, its
+ corresponding value and index would then not be included.  In particular,
+ `thresh == 0` (default) means everything is kept and actual thresholding happens
+ only for a positive value.</p><p>In the following shapes, <code>nnz</code> is the count after taking <code>thresh</code> into account.</p></div></div><div class="top"><p class="src"><a name="v:batchCholesky" class="def">batchCholesky</a> <a href="src/TensorFlow-GenOps-Core.html#batchCholesky" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div></div><div class="top"><p class="src"><a name="v:dynamicPartition" class="def">dynamicPartition</a> <a href="src/TensorFlow-GenOps-Core.html#dynamicPartition" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>num_partitions</strong>: The number of partitions to output.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>partitions</strong>: Any shape.  Indices in the range `[0, num_partitions)`.</p></td></tr><tr><td class="src">-&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t]</td><td class="doc"><p><strong>outputs</strong></p></td></tr></table></div><div class="doc"><p>Partitions `data` into <code>num_partitions</code> tensors using indices from <code>partitions</code>.</p><p>For each index tuple <code>js</code> of size `partitions.ndim`, the slice `data[js, ...]`
+ becomes part of `outputs[partitions[js]]`.  The slices with `partitions[js] = i`
+ are placed in `outputs[i]` in lexicographic order of <code>js</code>, and the first
+ dimension of `outputs[i]` is the number of entries in <code>partitions</code> equal to <code>i</code>.
+ In detail,</p><p>outputs[i].shape = [sum(partitions == i)] + data.shape[partitions.ndim:]</p><p>outputs[i] = pack([data[js, ...] for js if partitions[js] == i])</p><p>`data.shape` must start with `partitions.shape`.</p><p>For example:</p><p># Scalar partitions
+     partitions = 1
+     num_partitions = 2
+     data = [10, 20]
+     outputs[0] = []  # Empty with shape [0, 2]
+     outputs[1] = [[10, 20]]</p><p># Vector partitions
+     partitions = [0, 0, 1, 1, 0]
+     num_partitions = 2
+     data = [10, 20, 30, 40, 50]
+     outputs[0] = [10, 20, 50]
+     outputs[1] = [30, 40]</p><p><a href="div">style=&quot;width:70%; margin:auto; margin-bottom:10px; margin-top:20px;&quot;</a>
+ <a href="img">style=&quot;width:100%&quot; src=&quot;../../images/DynamicPartition.png&quot; alt</a>
+ <a href="/div">/div</a></p></div></div><div class="top"><p class="src"><a name="v:serializeSparse" class="def">serializeSparse</a> <a href="src/TensorFlow-GenOps-Core.html#serializeSparse" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>sparse_indices</strong>: 2-D.  The <code>indices</code> of the <code>SparseTensor</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>sparse_values</strong>: 1-D.  The <code>values</code> of the <code>SparseTensor</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>sparse_shape</strong>: 1-D.  The <code><a href="TensorFlow-GenOps-Core.html#v:shape">shape</a></code> of the <code>SparseTensor</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p><strong>serialized_sparse</strong></p></td></tr></table></div><div class="doc"><p>Serialize a <code>SparseTensor</code> into a string 3-vector (1-D <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code>) object.</p></div></div><div class="top"><p class="src"><a name="v:sparseConcat" class="def">sparseConcat</a> <a href="src/TensorFlow-GenOps-Core.html#sparseConcat" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>concat_dim</strong>: Dimension to concatenate along. Must be in range [-rank, rank),
+ where rank is the number of dimensions in each input <code>SparseTensor</code>.</p></td></tr><tr><td class="src">-&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]</td><td class="doc"><p><strong>indices</strong>: 2-D.  Indices of each input <code>SparseTensor</code>.</p></td></tr><tr><td class="src">-&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t]</td><td class="doc"><p><strong>values</strong>: 1-D.  Non-empty values of each <code>SparseTensor</code>.</p></td></tr><tr><td class="src">-&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]</td><td class="doc"><p><strong>shapes</strong>: 1-D.  Shapes of each <code>SparseTensor</code>.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</td><td class="doc"><p>(<strong>output_indices</strong>, <strong>output_values</strong>, <strong>output_shape</strong>)</p><ul><li><strong>output_indices</strong>: 2-D.  Indices of the concatenated <code>SparseTensor</code>.</li><li><strong>output_values</strong>: 1-D.  Non-empty values of the concatenated <code>SparseTensor</code>.</li><li><strong>output_shape</strong>: 1-D.  Shape of the concatenated <code>SparseTensor</code>.</li></ul></td></tr></table></div><div class="doc"><p>Concatenates a list of <code>SparseTensor</code> along the specified dimension.</p><p>Concatenation is with respect to the dense versions of these sparse tensors.
+ It is assumed that each input is a <code>SparseTensor</code> whose elements are ordered
+ along increasing dimension number.</p><p>All inputs' shapes must match, except for the concat dimension.  The
+ <code>indices</code>, <code>values</code>, and <code>shapes</code> lists must have the same length.</p><p>The output shape is identical to the inputs', except along the concat
+ dimension, where it is the sum of the inputs' sizes along that dimension.</p><p>The output elements will be resorted to preserve the sort order along
+ increasing dimension number.</p><p>This op runs in `O(M log M)` time, where <code>M</code> is the total number of non-empty
+ values across all inputs. This is due to the need for an internal sort in
+ order to concatenate efficiently across an arbitrary dimension.</p><p>For example, if `concat_dim = 1` and the inputs are</p><p>sp_inputs[0]: shape = [2, 3]
+     [0, 2]: &quot;a&quot;
+     [1, 0]: &quot;b&quot;
+     [1, 1]: &quot;c&quot;</p><p>sp_inputs[1]: shape = [2, 4]
+     [0, 1]: &quot;d&quot;
+     [0, 2]: &quot;e&quot;</p><p>then the output will be</p><p>shape = [2, 7]
+     [0, 2]: &quot;a&quot;
+     [0, 4]: &quot;d&quot;
+     [0, 5]: &quot;e&quot;
+     [1, 0]: &quot;b&quot;
+     [1, 1]: &quot;c&quot;</p><p>Graphically this is equivalent to doing</p><dl><dt>    a</dt><dd>concat [  d e  ] = [    a   d e  ]</dd><dt>b c  </dt><dd>[       ]   [b c          ]</dd></dl></div></div><div class="top"><p class="src"><a name="v:segmentProd" class="def">segmentProd</a> <a href="src/TensorFlow-GenOps-Core.html#segmentProd" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>data</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices</td><td class="doc"><p><strong>segment_ids</strong>: A 1-D tensor whose rank is equal to the rank of `data`'s
+ first dimension.  Values should be sorted and can be repeated.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: Has same shape as data, except for dimension 0 which
+ has size <code>k</code>, the number of segments.</p></td></tr></table></div><div class="doc"><p>Computes the product along segments of a tensor.</p><p>Read <a href="../../api_docs/python/math_ops.md#segmentation">the section on
+ Segmentation</a> for an explanation
+ of segments.</p><p>Computes a tensor such that
+ \(output_i = prod_j data_j\) where the product is over <code>j</code> such
+ that `segment_ids[j] == i`.</p><p><a href="div">style=&quot;width:70%; margin:auto; margin-bottom:10px; margin-top:20px;&quot;</a>
+ <a href="img">style=&quot;width:100%&quot; src=&quot;../../images/SegmentProd.png&quot; alt</a>
+ <a href="/div">/div</a></p></div></div><div class="top"><p class="src"><a name="v:sparseReshape" class="def">sparseReshape</a> <a href="src/TensorFlow-GenOps-Core.html#sparseReshape" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>input_indices</strong>: 2-D.  `N x R_in` matrix with the indices of non-empty values in a
+ SparseTensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>input_shape</strong>: 1-D.  <code>R_in</code> vector with the input SparseTensor's dense shape.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>new_shape</strong>: 1-D.  <code>R_out</code> vector with the requested new dense shape.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</td><td class="doc"><p>(<strong>output_indices</strong>, <strong>output_shape</strong>)</p><ul><li><strong>output_indices</strong>: 2-D.  `N x R_out` matrix with the updated indices of non-empty
+ values in the output SparseTensor.</li><li><strong>output_shape</strong>: 1-D.  <code>R_out</code> vector with the full dense shape of the output
+ SparseTensor.  This is the same as <code>new_shape</code> but with any -1 dimensions
+ filled in.</li></ul></td></tr></table></div><div class="doc"><p>Reshapes a SparseTensor to represent values in a new dense shape.</p><p>This operation has the same semantics as reshape on the represented dense
+ tensor.  The <code>input_indices</code> are recomputed based on the requested <code>new_shape</code>.</p><p>If one component of <code>new_shape</code> is the special value -1, the size of that
+ dimension is computed so that the total dense size remains constant.  At
+ most one component of <code>new_shape</code> can be -1.  The number of dense elements
+ implied by <code>new_shape</code> must be the same as the number of dense elements
+ originally implied by <code>input_shape</code>.</p><p>Reshaping does not affect the order of values in the SparseTensor.</p><p>If the input tensor has rank <code>R_in</code> and <code>N</code> non-empty values, and <code>new_shape</code>
+ has length <code>R_out</code>, then <code>input_indices</code> has shape `[N, R_in]`,
+ <code>input_shape</code> has length <code>R_in</code>, <code>output_indices</code> has shape `[N, R_out]`, and
+ <code>output_shape</code> has length <code>R_out</code>.</p></div></div><div class="top"><p class="src"><a name="v:sparseDenseCwiseMul" class="def">sparseDenseCwiseMul</a> <a href="src/TensorFlow-GenOps-Core.html#sparseDenseCwiseMul" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>sp_indices</strong>: 2-D.  `N x R` matrix with the indices of non-empty values in a
+ SparseTensor, possibly not in canonical ordering.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>sp_values</strong>: 1-D.  <code>N</code> non-empty values corresponding to <code>sp_indices</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>sp_shape</strong>: 1-D.  Shape of the input SparseTensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>dense</strong>: <code>R</code>-D.  The dense Tensor operand.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 1-D.  The <code>N</code> values that are operated on.</p></td></tr></table></div><div class="doc"><p>Component-wise multiplies a SparseTensor by a dense Tensor.</p><p>The output locations corresponding to the implicitly zero elements in the sparse
+ tensor will be zero (i.e., will not take up storage space), regardless of the
+ contents of the dense tensor (even if it's +/-INF and that INF*0 == NaN).</p><ul><li>Limitation*: this Op only broadcasts the dense side to the sparse side, but not
+ the other direction.</li></ul></div></div><div class="top"><p class="src"><a name="v:sparseDenseCwiseDiv" class="def">sparseDenseCwiseDiv</a> <a href="src/TensorFlow-GenOps-Core.html#sparseDenseCwiseDiv" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a>, <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>, <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>, <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]` t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>sp_indices</strong>: 2-D.  `N x R` matrix with the indices of non-empty values in a
+ SparseTensor, possibly not in canonical ordering.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>sp_values</strong>: 1-D.  <code>N</code> non-empty values corresponding to <code>sp_indices</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>sp_shape</strong>: 1-D.  Shape of the input SparseTensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>dense</strong>: <code>R</code>-D.  The dense Tensor operand.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: 1-D.  The <code>N</code> values that are operated on.</p></td></tr></table></div><div class="doc"><p>Component-wise divides a SparseTensor by a dense Tensor.</p><ul><li>Limitation*: this Op only broadcasts the dense side to the sparse side, but not
+ the other direction.</li></ul></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-95.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-95.html
new file mode 100644
index 0000000..0e1566a
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-95.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - _)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - _</p><table><tr><td class="src">_Arg</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:_Arg">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_HostCast</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:_HostCast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_HostRecv</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:_HostRecv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_HostSend</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:_HostSend">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_Recv</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:_Recv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_Retval</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:_Retval">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_Send</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:_Send">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-A.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-A.html
new file mode 100644
index 0000000..10ea155
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-A.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - A)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - A</p><table><tr><td class="src">abort</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:abort">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">abs</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:abs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">acos</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:acos">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">add</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:add">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">addN</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:addN">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">adjustContrast</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:adjustContrast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">adjustContrastv2</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:adjustContrastv2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">all</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:all">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">allCandidateSampler</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:allCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">any</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:any">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyAdadelta</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyAdadelta">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyAdagrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyAdagrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyAdagradDA</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyAdagradDA">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyAdam</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyAdam">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyFtrl</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyFtrl">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyGradientDescent</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyGradientDescent">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyMomentum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyMomentum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyProximalAdagrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyProximalAdagrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyProximalGradientDescent</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyProximalGradientDescent">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyRMSProp</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyRMSProp">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">argMax</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:argMax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">argMin</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:argMin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">asin</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:asin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">assign</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:assign">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">assignAdd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:assignAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">assignSub</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:assignSub">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">asString</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:asString">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">atan</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:atan">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">audioSummary</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:audioSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">avgPool</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:avgPool">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">avgPool3D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:avgPool3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">avgPool3DGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:avgPool3DGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">avgPoolGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:avgPoolGrad">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-All.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-All.html
new file mode 100644
index 0000000..4a2a07b
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-All.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="index"><p class="caption">Index</p><table><tr><td class="src">abort</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:abort">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">abs</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:abs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">acos</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:acos">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">add</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:add">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">addN</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:addN">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">adjustContrast</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:adjustContrast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">adjustContrastv2</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:adjustContrastv2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">all</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:all">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">allCandidateSampler</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:allCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">any</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:any">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyAdadelta</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyAdadelta">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyAdagrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyAdagrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyAdagradDA</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyAdagradDA">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyAdam</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyAdam">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyFtrl</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyFtrl">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyGradientDescent</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyGradientDescent">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyMomentum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyMomentum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyProximalAdagrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyProximalAdagrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyProximalGradientDescent</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyProximalGradientDescent">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">applyRMSProp</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:applyRMSProp">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">argMax</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:argMax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">argMin</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:argMin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">asin</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:asin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">assign</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:assign">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">assignAdd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:assignAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">assignSub</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:assignSub">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">asString</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:asString">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">atan</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:atan">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">audioSummary</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:audioSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">avgPool</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:avgPool">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">avgPool3D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:avgPool3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">avgPool3DGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:avgPool3DGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">avgPoolGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:avgPoolGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrier</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:barrier">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrierClose</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:barrierClose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrierIncompleteSize</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:barrierIncompleteSize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrierInsertMany</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:barrierInsertMany">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrierReadySize</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:barrierReadySize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchCholesky</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchCholesky">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchCholeskyGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchCholeskyGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchFFT</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchFFT">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchFFT2D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchFFT2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchFFT3D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchFFT3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchIFFT</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchIFFT">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchIFFT2D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchIFFT2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchIFFT3D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchIFFT3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatMul</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixBandPart</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixBandPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixDeterminant</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixDeterminant">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixDiag</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixDiag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixDiagPart</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixDiagPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixInverse</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixInverse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixSetDiag</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixSetDiag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixSolve</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixSolve">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixSolveLs</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixSolveLs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixTriangularSolve</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixTriangularSolve">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchNormWithGlobalNormalization</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchNormWithGlobalNormalization">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchNormWithGlobalNormalizationGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchNormWithGlobalNormalizationGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchSelfAdjointEig</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchSelfAdjointEig">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchSelfAdjointEigV2</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchSelfAdjointEigV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchSvd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchSvd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchToSpace</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchToSpace">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchToSpaceND</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchToSpaceND">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">betainc</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:betainc">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">biasAdd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:biasAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">biasAddGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:biasAddGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">biasAddV1</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:biasAddV1">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">bitcast</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:bitcast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">broadcastGradientArgs</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:broadcastGradientArgs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cast</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">ceil</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:ceil">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">checkNumerics</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:checkNumerics">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cholesky</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cholesky">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">choleskyGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:choleskyGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">complex</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:complex">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">complexAbs</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:complexAbs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">computeAccidentalHits</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:computeAccidentalHits">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">concat</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:concat">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">concatOffset</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:concatOffset">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conj</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conj">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">const</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:const">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">controlTrigger</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:controlTrigger">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv2D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conv2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv2DBackpropFilter</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conv2DBackpropFilter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv2DBackpropInput</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conv2DBackpropInput">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conv3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3DBackpropFilter</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conv3DBackpropFilter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3DBackpropFilterV2</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conv3DBackpropFilterV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3DBackpropInput</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conv3DBackpropInput">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3DBackpropInputV2</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conv3DBackpropInputV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">copy</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:copy">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">copyHost</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:copyHost">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cos</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cos">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">countUpTo</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:countUpTo">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cropAndResize</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cropAndResize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cropAndResizeGradBoxes</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cropAndResizeGradBoxes">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cropAndResizeGradImage</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cropAndResizeGradImage">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cross</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cross">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cTCBeamSearchDecoder</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cTCBeamSearchDecoder">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cTCGreedyDecoder</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cTCGreedyDecoder">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cTCLoss</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cTCLoss">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cumprod</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cumprod">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cumsum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cumsum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">debugIdentity</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:debugIdentity">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">debugNanCount</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:debugNanCount">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeBase64</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:decodeBase64">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeGif</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:decodeGif">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeJpeg</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:decodeJpeg">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeJSONExample</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:decodeJSONExample">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodePng</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:decodePng">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeRaw</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:decodeRaw">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">deleteSessionTensor</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:deleteSessionTensor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">depthToSpace</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:depthToSpace">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">depthwiseConv2dNative</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:depthwiseConv2dNative">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">depthwiseConv2dNativeBackpropFilter</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:depthwiseConv2dNativeBackpropFilter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">depthwiseConv2dNativeBackpropInput</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:depthwiseConv2dNativeBackpropInput">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">deserializeManySparse</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:deserializeManySparse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">destroyTemporaryVariable</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:destroyTemporaryVariable">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">diag</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:diag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">diagPart</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:diagPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">digamma</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:digamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dilation2D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:dilation2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dilation2DBackpropFilter</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:dilation2DBackpropFilter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dilation2DBackpropInput</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:dilation2DBackpropInput">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">div</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:div">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">drawBoundingBoxes</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:drawBoundingBoxes">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dynamicPartition</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:dynamicPartition">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dynamicStitch</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:dynamicStitch">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">editDistance</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:editDistance">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">elu</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:elu">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">eluGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:eluGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">encodeBase64</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:encodeBase64">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">encodeJpeg</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:encodeJpeg">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">encodePng</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:encodePng">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">enter</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:enter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">equal</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:equal">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">erf</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:erf">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">erfc</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:erfc">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">exit</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:exit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">exp</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:exp">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">expandDims</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:expandDims">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">extractGlimpse</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:extractGlimpse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">extractImagePatches</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:extractImagePatches">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fact</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fact">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fFT</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fFT">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fFT2D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fFT2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fFT3D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fFT3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fIFOQueue</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fIFOQueue">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fill</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fill">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fixedLengthRecordReader</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fixedLengthRecordReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fixedUnigramCandidateSampler</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fixedUnigramCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">floor</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:floor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fractionalAvgPool</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fractionalAvgPool">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fractionalAvgPoolGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fractionalAvgPoolGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fractionalMaxPool</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fractionalMaxPool">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fractionalMaxPoolGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fractionalMaxPoolGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fusedResizeAndPadConv2D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fusedResizeAndPadConv2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">gather</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:gather">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">gatherNd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:gatherNd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">getSessionHandle</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:getSessionHandle">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">getSessionTensor</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:getSessionTensor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">greater</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:greater">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">greaterEqual</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:greaterEqual">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">histogramSummary</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:histogramSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">hSVToRGB</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:hSVToRGB">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">identity</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:identity">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">identityReader</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:identityReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">iFFT</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:iFFT">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">iFFT2D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:iFFT2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">iFFT3D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:iFFT3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">igamma</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:igamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">igammac</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:igammac">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">imag</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:imag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">imageSummary</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:imageSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">immutableConst</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:immutableConst">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">initializeTable</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:initializeTable">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">initializeTableFromTextFile</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:initializeTableFromTextFile">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">inTopK</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:inTopK">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">inv</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:inv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">invertPermutation</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:invertPermutation">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">invGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:invGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">isFinite</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:isFinite">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">isInf</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:isInf">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">isNan</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:isNan">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">isVariableInitialized</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:isVariableInitialized">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">l2Loss</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:l2Loss">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">learnedUnigramCandidateSampler</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:learnedUnigramCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">less</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:less">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lessEqual</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lessEqual">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lgamma</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lgamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">linSpace</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:linSpace">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">listDiff</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:listDiff">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">log</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:log">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logicalAnd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:logicalAnd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logicalNot</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:logicalNot">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logicalOr</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:logicalOr">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logSoftmax</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:logSoftmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logUniformCandidateSampler</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:logUniformCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableExport</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lookupTableExport">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableFind</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lookupTableFind">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableImport</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lookupTableImport">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableInsert</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lookupTableInsert">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableSize</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lookupTableSize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">loopCond</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:loopCond">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lRN</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lRN">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lRNGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lRNGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matchingFiles</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matchingFiles">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matMul</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixBandPart</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixBandPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixDeterminant</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixDeterminant">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixDiag</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixDiag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixDiagPart</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixDiagPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixInverse</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixInverse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixSetDiag</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixSetDiag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixSolve</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixSolve">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixSolveLs</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixSolveLs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixTriangularSolve</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixTriangularSolve">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">max</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:max">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maximum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:maximum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPool</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:maxPool">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPool3D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:maxPool3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPool3DGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:maxPool3DGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPoolGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:maxPoolGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPoolGradWithArgmax</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:maxPoolGradWithArgmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPoolWithArgmax</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:maxPoolWithArgmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mean</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:mean">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">merge</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:merge">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mergeSummary</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:mergeSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">min</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:min">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">minimum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:minimum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mirrorPad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:mirrorPad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mirrorPadGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:mirrorPadGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mod</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:mod">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mul</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:mul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">multinomial</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:multinomial">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">neg</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:neg">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">negTrain</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:negTrain">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">nextIteration</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:nextIteration">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">nonMaxSuppression</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:nonMaxSuppression">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">noOp</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:noOp">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">notEqual</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:notEqual">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">oneHot</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:oneHot">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">pack</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:pack">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">pad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:pad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">paddingFIFOQueue</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:paddingFIFOQueue">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">parameterizedTruncatedNormal</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:parameterizedTruncatedNormal">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">parseTensor</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:parseTensor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">placeholder</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:placeholder">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">placeholderWithDefault</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:placeholderWithDefault">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">polygamma</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:polygamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">pow</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:pow">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">priorityQueue</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:priorityQueue">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">prod</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:prod">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">quantizeAndDequantize</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:quantizeAndDequantize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">queueClose</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:queueClose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">queueSize</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:queueSize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomCrop</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:randomCrop">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomGamma</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:randomGamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomShuffle</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:randomShuffle">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomShuffleQueue</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:randomShuffleQueue">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomStandardNormal</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:randomStandardNormal">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomUniform</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:randomUniform">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomUniformInt</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:randomUniformInt">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">range</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:range">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">rank</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:rank">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerNumRecordsProduced</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:readerNumRecordsProduced">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerNumWorkUnitsCompleted</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:readerNumWorkUnitsCompleted">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerRead</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:readerRead">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerReadUpTo</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:readerReadUpTo">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerReset</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:readerReset">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerRestoreState</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:readerRestoreState">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerSerializeState</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:readerSerializeState">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readFile</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:readFile">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">real</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:real">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reduceJoin</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:reduceJoin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refEnter</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:refEnter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refExit</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:refExit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refIdentity</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:refIdentity">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refMerge</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:refMerge">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refNextIteration</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:refNextIteration">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refSelect</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:refSelect">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refSwitch</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:refSwitch">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">relu</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:relu">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">relu6</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:relu6">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">relu6Grad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:relu6Grad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reluGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:reluGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reshape</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:reshape">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeArea</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:resizeArea">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeBicubic</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:resizeBicubic">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeBilinear</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:resizeBilinear">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeBilinearGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:resizeBilinearGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeNearestNeighbor</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:resizeNearestNeighbor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeNearestNeighborGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:resizeNearestNeighborGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">restore</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:restore">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">restoreSlice</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:restoreSlice">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reverse</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:reverse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reverseSequence</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:reverseSequence">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">rGBToHSV</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:rGBToHSV">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">rsqrt</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:rsqrt">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">rsqrtGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:rsqrtGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sampleDistortedBoundingBox</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sampleDistortedBoundingBox">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scalarSummary</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:scalarSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterAdd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:scatterAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterDiv</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:scatterDiv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterMul</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:scatterMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterSub</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:scatterSub">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterUpdate</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:scatterUpdate">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentMax</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:segmentMax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentMean</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:segmentMean">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentMin</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:segmentMin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentProd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:segmentProd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentSum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:segmentSum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">select</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:select">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">selfAdjointEig</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:selfAdjointEig">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">selfAdjointEigV2</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:selfAdjointEigV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">serializeManySparse</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:serializeManySparse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">serializeSparse</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:serializeSparse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">shape</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:shape">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">shapeN</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:shapeN">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">shardedFilename</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:shardedFilename">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">shardedFilespec</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:shardedFilespec">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sigmoid</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sigmoid">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sigmoidGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sigmoidGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sign</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sign">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sin</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">size</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:size">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">slice</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:slice">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softmax</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:softmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softmaxCrossEntropyWithLogits</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:softmaxCrossEntropyWithLogits">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softplus</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:softplus">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softplusGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:softplusGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softsign</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:softsign">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softsignGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:softsignGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">spaceToBatch</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:spaceToBatch">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">spaceToBatchND</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:spaceToBatchND">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">spaceToDepth</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:spaceToDepth">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseAdd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseAddGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseAddGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyAdadelta</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyAdadelta">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyAdagrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyAdagrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyAdagradDA</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyAdagradDA">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyFtrl</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyFtrl">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyMomentum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyMomentum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyProximalAdagrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyProximalAdagrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyProximalGradientDescent</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyProximalGradientDescent">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyRMSProp</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyRMSProp">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseConcat</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseConcat">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseDenseCwiseAdd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseDenseCwiseAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseDenseCwiseDiv</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseDenseCwiseDiv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseDenseCwiseMul</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseDenseCwiseMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseMatMul</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseMatMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseReduceSum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseReduceSum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseReduceSumSparse</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseReduceSumSparse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseReorder</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseReorder">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseReshape</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseReshape">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentMean</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSegmentMean">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentMeanGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSegmentMeanGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentSqrtN</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSegmentSqrtN">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentSqrtNGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSegmentSqrtNGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentSum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSegmentSum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSoftmax</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSoftmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSoftmaxCrossEntropyWithLogits</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSoftmaxCrossEntropyWithLogits">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSparseMaximum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSparseMaximum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSparseMinimum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSparseMinimum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSplit</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSplit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseTensorDenseAdd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseTensorDenseAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseTensorDenseMatMul</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseTensorDenseMatMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseToDense</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseToDense">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">split</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:split">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sqrt</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sqrt">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sqrtGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sqrtGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">square</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:square">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">squaredDifference</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:squaredDifference">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">squeeze</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:squeeze">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stackClose</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stackClose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stackPop</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stackPop">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stackPush</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stackPush">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stopGradient</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stopGradient">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stridedSlice</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stridedSlice">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stridedSliceAssign</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stridedSliceAssign">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stridedSliceGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stridedSliceGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringJoin</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stringJoin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringSplit</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stringSplit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringToHashBucket</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stringToHashBucket">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringToHashBucketFast</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stringToHashBucketFast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringToHashBucketStrong</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stringToHashBucketStrong">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringToNumber</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stringToNumber">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sub</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sub">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">svd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:svd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">switch</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:switch">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tan</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tan">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tanh</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tanh">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tanhGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tanhGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">temporaryVariable</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:temporaryVariable">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayClose</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayClose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayConcat</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayConcat">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayGather</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayGather">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayPack</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayPack">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayRead</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayRead">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayScatter</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayScatter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArraySize</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArraySize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArraySplit</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArraySplit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayUnpack</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayUnpack">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayWrite</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayWrite">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorSummary</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">textLineReader</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:textLineReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tFRecordReader</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tFRecordReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">threadUnsafeUnigramCandidateSampler</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:threadUnsafeUnigramCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tile</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tile">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tileGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tileGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">topK</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:topK">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">topKV2</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:topKV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">transpose</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:transpose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">truncatedNormal</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:truncatedNormal">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">uniformCandidateSampler</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:uniformCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">unique</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:unique">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">uniqueWithCounts</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:uniqueWithCounts">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">unpack</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:unpack">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">unsortedSegmentSum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:unsortedSegmentSum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">variable</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:variable">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">where'</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:where-39-">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">wholeFileReader</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:wholeFileReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">zerosLike</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:zerosLike">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">zeta</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:zeta">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_Arg</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:_Arg">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_HostCast</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:_HostCast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_HostRecv</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:_HostRecv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_HostSend</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:_HostSend">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_Recv</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:_Recv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_Retval</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:_Retval">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">_Send</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:_Send">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-B.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-B.html
new file mode 100644
index 0000000..b7ff19b
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-B.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - B)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - B</p><table><tr><td class="src">barrier</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:barrier">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrierClose</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:barrierClose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrierIncompleteSize</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:barrierIncompleteSize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrierInsertMany</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:barrierInsertMany">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">barrierReadySize</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:barrierReadySize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchCholesky</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchCholesky">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchCholeskyGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchCholeskyGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchFFT</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchFFT">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchFFT2D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchFFT2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchFFT3D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchFFT3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchIFFT</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchIFFT">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchIFFT2D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchIFFT2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchIFFT3D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchIFFT3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatMul</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixBandPart</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixBandPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixDeterminant</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixDeterminant">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixDiag</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixDiag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixDiagPart</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixDiagPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixInverse</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixInverse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixSetDiag</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixSetDiag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixSolve</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixSolve">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixSolveLs</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixSolveLs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchMatrixTriangularSolve</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixTriangularSolve">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchNormWithGlobalNormalization</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchNormWithGlobalNormalization">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchNormWithGlobalNormalizationGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchNormWithGlobalNormalizationGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchSelfAdjointEig</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchSelfAdjointEig">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchSelfAdjointEigV2</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchSelfAdjointEigV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchSvd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchSvd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchToSpace</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchToSpace">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">batchToSpaceND</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:batchToSpaceND">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">betainc</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:betainc">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">biasAdd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:biasAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">biasAddGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:biasAddGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">biasAddV1</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:biasAddV1">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">bitcast</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:bitcast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">broadcastGradientArgs</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:broadcastGradientArgs">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-C.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-C.html
new file mode 100644
index 0000000..9bed483
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-C.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - C)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - C</p><table><tr><td class="src">cast</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">ceil</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:ceil">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">checkNumerics</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:checkNumerics">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cholesky</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cholesky">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">choleskyGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:choleskyGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">complex</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:complex">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">complexAbs</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:complexAbs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">computeAccidentalHits</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:computeAccidentalHits">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">concat</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:concat">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">concatOffset</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:concatOffset">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conj</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conj">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">const</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:const">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">controlTrigger</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:controlTrigger">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv2D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conv2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv2DBackpropFilter</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conv2DBackpropFilter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv2DBackpropInput</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conv2DBackpropInput">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conv3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3DBackpropFilter</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conv3DBackpropFilter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3DBackpropFilterV2</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conv3DBackpropFilterV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3DBackpropInput</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conv3DBackpropInput">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">conv3DBackpropInputV2</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:conv3DBackpropInputV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">copy</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:copy">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">copyHost</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:copyHost">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cos</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cos">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">countUpTo</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:countUpTo">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cropAndResize</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cropAndResize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cropAndResizeGradBoxes</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cropAndResizeGradBoxes">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cropAndResizeGradImage</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cropAndResizeGradImage">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cross</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cross">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cTCBeamSearchDecoder</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cTCBeamSearchDecoder">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cTCGreedyDecoder</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cTCGreedyDecoder">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cTCLoss</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cTCLoss">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cumprod</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cumprod">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">cumsum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:cumsum">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-D.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-D.html
new file mode 100644
index 0000000..ef74313
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-D.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - D)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - D</p><table><tr><td class="src">debugIdentity</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:debugIdentity">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">debugNanCount</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:debugNanCount">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeBase64</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:decodeBase64">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeGif</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:decodeGif">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeJpeg</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:decodeJpeg">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeJSONExample</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:decodeJSONExample">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodePng</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:decodePng">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">decodeRaw</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:decodeRaw">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">deleteSessionTensor</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:deleteSessionTensor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">depthToSpace</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:depthToSpace">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">depthwiseConv2dNative</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:depthwiseConv2dNative">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">depthwiseConv2dNativeBackpropFilter</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:depthwiseConv2dNativeBackpropFilter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">depthwiseConv2dNativeBackpropInput</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:depthwiseConv2dNativeBackpropInput">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">deserializeManySparse</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:deserializeManySparse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">destroyTemporaryVariable</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:destroyTemporaryVariable">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">diag</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:diag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">diagPart</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:diagPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">digamma</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:digamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dilation2D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:dilation2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dilation2DBackpropFilter</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:dilation2DBackpropFilter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dilation2DBackpropInput</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:dilation2DBackpropInput">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">div</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:div">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">drawBoundingBoxes</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:drawBoundingBoxes">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dynamicPartition</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:dynamicPartition">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">dynamicStitch</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:dynamicStitch">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-E.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-E.html
new file mode 100644
index 0000000..1ae3874
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-E.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - E)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - E</p><table><tr><td class="src">editDistance</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:editDistance">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">elu</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:elu">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">eluGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:eluGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">encodeBase64</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:encodeBase64">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">encodeJpeg</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:encodeJpeg">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">encodePng</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:encodePng">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">enter</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:enter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">equal</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:equal">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">erf</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:erf">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">erfc</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:erfc">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">exit</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:exit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">exp</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:exp">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">expandDims</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:expandDims">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">extractGlimpse</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:extractGlimpse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">extractImagePatches</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:extractImagePatches">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-F.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-F.html
new file mode 100644
index 0000000..f511a43
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-F.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - F)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - F</p><table><tr><td class="src">fact</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fact">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fFT</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fFT">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fFT2D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fFT2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fFT3D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fFT3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fIFOQueue</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fIFOQueue">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fill</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fill">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fixedLengthRecordReader</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fixedLengthRecordReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fixedUnigramCandidateSampler</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fixedUnigramCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">floor</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:floor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fractionalAvgPool</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fractionalAvgPool">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fractionalAvgPoolGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fractionalAvgPoolGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fractionalMaxPool</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fractionalMaxPool">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fractionalMaxPoolGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fractionalMaxPoolGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">fusedResizeAndPadConv2D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:fusedResizeAndPadConv2D">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-G.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-G.html
new file mode 100644
index 0000000..3623797
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-G.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - G)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - G</p><table><tr><td class="src">gather</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:gather">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">gatherNd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:gatherNd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">getSessionHandle</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:getSessionHandle">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">getSessionTensor</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:getSessionTensor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">greater</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:greater">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">greaterEqual</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:greaterEqual">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-H.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-H.html
new file mode 100644
index 0000000..12f960e
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-H.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - H)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - H</p><table><tr><td class="src">histogramSummary</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:histogramSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">hSVToRGB</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:hSVToRGB">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-I.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-I.html
new file mode 100644
index 0000000..5076cc3
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-I.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - I)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - I</p><table><tr><td class="src">identity</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:identity">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">identityReader</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:identityReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">iFFT</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:iFFT">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">iFFT2D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:iFFT2D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">iFFT3D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:iFFT3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">igamma</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:igamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">igammac</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:igammac">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">imag</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:imag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">imageSummary</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:imageSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">immutableConst</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:immutableConst">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">initializeTable</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:initializeTable">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">initializeTableFromTextFile</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:initializeTableFromTextFile">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">inTopK</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:inTopK">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">inv</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:inv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">invertPermutation</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:invertPermutation">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">invGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:invGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">isFinite</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:isFinite">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">isInf</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:isInf">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">isNan</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:isNan">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">isVariableInitialized</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:isVariableInitialized">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-L.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-L.html
new file mode 100644
index 0000000..df05ab0
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-L.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - L)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - L</p><table><tr><td class="src">l2Loss</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:l2Loss">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">learnedUnigramCandidateSampler</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:learnedUnigramCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">less</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:less">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lessEqual</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lessEqual">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lgamma</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lgamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">linSpace</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:linSpace">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">listDiff</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:listDiff">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">log</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:log">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logicalAnd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:logicalAnd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logicalNot</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:logicalNot">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logicalOr</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:logicalOr">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logSoftmax</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:logSoftmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">logUniformCandidateSampler</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:logUniformCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableExport</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lookupTableExport">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableFind</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lookupTableFind">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableImport</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lookupTableImport">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableInsert</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lookupTableInsert">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lookupTableSize</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lookupTableSize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">loopCond</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:loopCond">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lRN</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lRN">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">lRNGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:lRNGrad">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-M.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-M.html
new file mode 100644
index 0000000..d43cd4c
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-M.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - M)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - M</p><table><tr><td class="src">matchingFiles</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matchingFiles">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matMul</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixBandPart</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixBandPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixDeterminant</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixDeterminant">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixDiag</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixDiag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixDiagPart</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixDiagPart">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixInverse</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixInverse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixSetDiag</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixSetDiag">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixSolve</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixSolve">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixSolveLs</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixSolveLs">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">matrixTriangularSolve</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:matrixTriangularSolve">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">max</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:max">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maximum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:maximum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPool</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:maxPool">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPool3D</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:maxPool3D">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPool3DGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:maxPool3DGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPoolGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:maxPoolGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPoolGradWithArgmax</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:maxPoolGradWithArgmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">maxPoolWithArgmax</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:maxPoolWithArgmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mean</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:mean">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">merge</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:merge">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mergeSummary</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:mergeSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">min</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:min">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">minimum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:minimum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mirrorPad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:mirrorPad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mirrorPadGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:mirrorPadGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mod</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:mod">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">mul</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:mul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">multinomial</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:multinomial">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-N.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-N.html
new file mode 100644
index 0000000..c2e42d8
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-N.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - N)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - N</p><table><tr><td class="src">neg</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:neg">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">negTrain</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:negTrain">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">nextIteration</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:nextIteration">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">nonMaxSuppression</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:nonMaxSuppression">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">noOp</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:noOp">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">notEqual</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:notEqual">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-O.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-O.html
new file mode 100644
index 0000000..1f3b4cc
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-O.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - O)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - O</p><table><tr><td class="src">oneHot</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:oneHot">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-P.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-P.html
new file mode 100644
index 0000000..ddea867
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-P.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - P)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - P</p><table><tr><td class="src">pack</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:pack">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">pad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:pad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">paddingFIFOQueue</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:paddingFIFOQueue">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">parameterizedTruncatedNormal</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:parameterizedTruncatedNormal">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">parseTensor</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:parseTensor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">placeholder</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:placeholder">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">placeholderWithDefault</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:placeholderWithDefault">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">polygamma</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:polygamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">pow</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:pow">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">priorityQueue</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:priorityQueue">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">prod</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:prod">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-Q.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-Q.html
new file mode 100644
index 0000000..243dcea
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-Q.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - Q)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - Q</p><table><tr><td class="src">quantizeAndDequantize</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:quantizeAndDequantize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">queueClose</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:queueClose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">queueSize</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:queueSize">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-R.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-R.html
new file mode 100644
index 0000000..4296c9f
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-R.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - R)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - R</p><table><tr><td class="src">randomCrop</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:randomCrop">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomGamma</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:randomGamma">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomShuffle</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:randomShuffle">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomShuffleQueue</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:randomShuffleQueue">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomStandardNormal</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:randomStandardNormal">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomUniform</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:randomUniform">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">randomUniformInt</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:randomUniformInt">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">range</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:range">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">rank</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:rank">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerNumRecordsProduced</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:readerNumRecordsProduced">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerNumWorkUnitsCompleted</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:readerNumWorkUnitsCompleted">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerRead</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:readerRead">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerReadUpTo</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:readerReadUpTo">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerReset</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:readerReset">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerRestoreState</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:readerRestoreState">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readerSerializeState</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:readerSerializeState">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">readFile</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:readFile">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">real</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:real">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reduceJoin</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:reduceJoin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refEnter</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:refEnter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refExit</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:refExit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refIdentity</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:refIdentity">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refMerge</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:refMerge">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refNextIteration</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:refNextIteration">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refSelect</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:refSelect">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">refSwitch</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:refSwitch">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">relu</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:relu">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">relu6</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:relu6">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">relu6Grad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:relu6Grad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reluGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:reluGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reshape</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:reshape">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeArea</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:resizeArea">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeBicubic</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:resizeBicubic">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeBilinear</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:resizeBilinear">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeBilinearGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:resizeBilinearGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeNearestNeighbor</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:resizeNearestNeighbor">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">resizeNearestNeighborGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:resizeNearestNeighborGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">restore</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:restore">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">restoreSlice</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:restoreSlice">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reverse</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:reverse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">reverseSequence</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:reverseSequence">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">rGBToHSV</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:rGBToHSV">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">rsqrt</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:rsqrt">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">rsqrtGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:rsqrtGrad">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-S.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-S.html
new file mode 100644
index 0000000..bce9611
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-S.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - S)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - S</p><table><tr><td class="src">sampleDistortedBoundingBox</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sampleDistortedBoundingBox">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scalarSummary</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:scalarSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterAdd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:scatterAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterDiv</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:scatterDiv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterMul</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:scatterMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterSub</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:scatterSub">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">scatterUpdate</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:scatterUpdate">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentMax</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:segmentMax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentMean</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:segmentMean">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentMin</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:segmentMin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentProd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:segmentProd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">segmentSum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:segmentSum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">select</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:select">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">selfAdjointEig</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:selfAdjointEig">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">selfAdjointEigV2</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:selfAdjointEigV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">serializeManySparse</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:serializeManySparse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">serializeSparse</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:serializeSparse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">shape</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:shape">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">shapeN</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:shapeN">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">shardedFilename</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:shardedFilename">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">shardedFilespec</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:shardedFilespec">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sigmoid</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sigmoid">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sigmoidGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sigmoidGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sign</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sign">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sin</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">size</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:size">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">slice</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:slice">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softmax</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:softmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softmaxCrossEntropyWithLogits</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:softmaxCrossEntropyWithLogits">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softplus</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:softplus">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softplusGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:softplusGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softsign</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:softsign">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">softsignGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:softsignGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">spaceToBatch</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:spaceToBatch">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">spaceToBatchND</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:spaceToBatchND">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">spaceToDepth</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:spaceToDepth">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseAdd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseAddGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseAddGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyAdadelta</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyAdadelta">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyAdagrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyAdagrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyAdagradDA</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyAdagradDA">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyFtrl</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyFtrl">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyMomentum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyMomentum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyProximalAdagrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyProximalAdagrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyProximalGradientDescent</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyProximalGradientDescent">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseApplyRMSProp</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyRMSProp">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseConcat</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseConcat">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseDenseCwiseAdd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseDenseCwiseAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseDenseCwiseDiv</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseDenseCwiseDiv">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseDenseCwiseMul</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseDenseCwiseMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseMatMul</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseMatMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseReduceSum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseReduceSum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseReduceSumSparse</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseReduceSumSparse">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseReorder</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseReorder">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseReshape</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseReshape">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentMean</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSegmentMean">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentMeanGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSegmentMeanGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentSqrtN</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSegmentSqrtN">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentSqrtNGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSegmentSqrtNGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSegmentSum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSegmentSum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSoftmax</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSoftmax">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSoftmaxCrossEntropyWithLogits</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSoftmaxCrossEntropyWithLogits">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSparseMaximum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSparseMaximum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSparseMinimum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSparseMinimum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseSplit</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseSplit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseTensorDenseAdd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseTensorDenseAdd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseTensorDenseMatMul</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseTensorDenseMatMul">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sparseToDense</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sparseToDense">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">split</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:split">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sqrt</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sqrt">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sqrtGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sqrtGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">square</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:square">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">squaredDifference</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:squaredDifference">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">squeeze</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:squeeze">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stackClose</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stackClose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stackPop</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stackPop">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stackPush</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stackPush">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stopGradient</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stopGradient">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stridedSlice</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stridedSlice">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stridedSliceAssign</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stridedSliceAssign">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stridedSliceGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stridedSliceGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringJoin</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stringJoin">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringSplit</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stringSplit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringToHashBucket</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stringToHashBucket">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringToHashBucketFast</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stringToHashBucketFast">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringToHashBucketStrong</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stringToHashBucketStrong">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">stringToNumber</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:stringToNumber">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sub</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sub">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">sum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:sum">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">svd</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:svd">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">switch</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:switch">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-T.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-T.html
new file mode 100644
index 0000000..60771c2
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-T.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - T)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - T</p><table><tr><td class="src">tan</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tan">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tanh</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tanh">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tanhGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tanhGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">temporaryVariable</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:temporaryVariable">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayClose</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayClose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayConcat</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayConcat">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayGather</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayGather">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayPack</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayPack">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayRead</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayRead">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayScatter</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayScatter">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArraySize</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArraySize">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArraySplit</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArraySplit">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayUnpack</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayUnpack">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorArrayWrite</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayWrite">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tensorSummary</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tensorSummary">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">textLineReader</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:textLineReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tFRecordReader</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tFRecordReader">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">threadUnsafeUnigramCandidateSampler</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:threadUnsafeUnigramCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tile</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tile">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">tileGrad</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:tileGrad">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">topK</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:topK">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">topKV2</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:topKV2">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">transpose</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:transpose">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">truncatedNormal</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:truncatedNormal">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-U.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-U.html
new file mode 100644
index 0000000..eb6bf96
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-U.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - U)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - U</p><table><tr><td class="src">uniformCandidateSampler</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:uniformCandidateSampler">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">unique</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:unique">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">uniqueWithCounts</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:uniqueWithCounts">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">unpack</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:unpack">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">unsortedSegmentSum</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:unsortedSegmentSum">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-V.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-V.html
new file mode 100644
index 0000000..57902bb
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-V.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - V)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - V</p><table><tr><td class="src">variable</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:variable">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-W.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-W.html
new file mode 100644
index 0000000..c3ad075
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-W.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - W)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - W</p><table><tr><td class="src">where'</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:where-39-">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">wholeFileReader</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:wholeFileReader">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-Z.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-Z.html
new file mode 100644
index 0000000..564dda7
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index-Z.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index - Z)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - Z</p><table><tr><td class="src">zerosLike</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:zerosLike">TensorFlow.GenOps.Core</a></td></tr><tr><td class="src">zeta</td><td class="module"><a href="TensorFlow-GenOps-Core.html#v:zeta">TensorFlow.GenOps.Core</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index.html
new file mode 100644
index 0000000..f2397b4
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/doc-index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops. (Index)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-Q.html">Q</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-W.html">W</a></li><li><a href="doc-index-Z.html">Z</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/frames.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/frames.html
new file mode 100644
index 0000000..1b4e38d
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/frames.html
@@ -0,0 +1,30 @@
+<!DOCTYPE html 
+     PUBLIC "-//W3C//DTD XHTML 1.0 Frameset//EN"
+     "http://www.w3.org/TR/xhtml1/DTD/xhtml1-frameset.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<title></title>
+<script src="haddock-util.js" type="text/javascript"></script>
+<script type="text/javascript"><!--
+/*
+
+  The synopsis frame needs to be updated using javascript, so we hide
+  it by default and only show it if javascript is enabled.
+
+  TODO: provide some means to disable it.
+*/
+function load() {
+  var d = document.getElementById("inner-fs");
+  d.rows = "50%,50%";
+  postReframe();
+}
+--></script>
+</head>
+<frameset id="outer-fs" cols="25%,75%" onload="load()">
+  <frameset id="inner-fs" rows="100%,0%">
+    <frame src="index-frames.html" name="modules" />
+    <frame src="" name="synopsis" />
+  </frameset>
+  <frame src="index.html" name="main" />
+</frameset>
+</html>
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/haddock-util.js b/docs/haddock/tensorflow-core-ops-0.1.0.0/haddock-util.js
new file mode 100644
index 0000000..9a6fccf
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/haddock-util.js
@@ -0,0 +1,344 @@
+// Haddock JavaScript utilities
+
+var rspace = /\s\s+/g,
+	  rtrim = /^\s+|\s+$/g;
+
+function spaced(s) { return (" " + s + " ").replace(rspace, " "); }
+function trim(s)   { return s.replace(rtrim, ""); }
+
+function hasClass(elem, value) {
+  var className = spaced(elem.className || "");
+  return className.indexOf( " " + value + " " ) >= 0;
+}
+
+function addClass(elem, value) {
+  var className = spaced(elem.className || "");
+  if ( className.indexOf( " " + value + " " ) < 0 ) {
+    elem.className = trim(className + " " + value);
+  }
+}
+
+function removeClass(elem, value) {
+  var className = spaced(elem.className || "");
+  className = className.replace(" " + value + " ", " ");
+  elem.className = trim(className);
+}
+
+function toggleClass(elem, valueOn, valueOff, bool) {
+  if (bool == null) { bool = ! hasClass(elem, valueOn); }
+  if (bool) {
+    removeClass(elem, valueOff);
+    addClass(elem, valueOn);
+  }
+  else {
+    removeClass(elem, valueOn);
+    addClass(elem, valueOff);
+  }
+  return bool;
+}
+
+
+function makeClassToggle(valueOn, valueOff)
+{
+  return function(elem, bool) {
+    return toggleClass(elem, valueOn, valueOff, bool);
+  }
+}
+
+toggleShow = makeClassToggle("show", "hide");
+toggleCollapser = makeClassToggle("collapser", "expander");
+
+function toggleSection(id)
+{
+  var b = toggleShow(document.getElementById("section." + id));
+  toggleCollapser(document.getElementById("control." + id), b);
+  rememberCollapsed(id, b);
+  return b;
+}
+
+var collapsed = {};
+function rememberCollapsed(id, b)
+{
+  if(b)
+    delete collapsed[id]
+  else
+    collapsed[id] = null;
+
+  var sections = [];
+  for(var i in collapsed)
+  {
+    if(collapsed.hasOwnProperty(i))
+      sections.push(i);
+  }
+  // cookie specific to this page; don't use setCookie which sets path=/
+  document.cookie = "collapsed=" + escape(sections.join('+'));
+}
+
+function restoreCollapsed()
+{
+  var cookie = getCookie("collapsed");
+  if(!cookie)
+    return;
+
+  var ids = cookie.split('+');
+  for(var i in ids)
+  {
+    if(document.getElementById("section." + ids[i]))
+      toggleSection(ids[i]);
+  }
+}
+
+function setCookie(name, value) {
+  document.cookie = name + "=" + escape(value) + ";path=/;";
+}
+
+function clearCookie(name) {
+  document.cookie = name + "=;path=/;expires=Thu, 01-Jan-1970 00:00:01 GMT;";
+}
+
+function getCookie(name) {
+  var nameEQ = name + "=";
+  var ca = document.cookie.split(';');
+  for(var i=0;i < ca.length;i++) {
+    var c = ca[i];
+    while (c.charAt(0)==' ') c = c.substring(1,c.length);
+    if (c.indexOf(nameEQ) == 0) {
+      return unescape(c.substring(nameEQ.length,c.length));
+    }
+  }
+  return null;
+}
+
+
+
+var max_results = 75; // 50 is not enough to search for map in the base libraries
+var shown_range = null;
+var last_search = null;
+
+function quick_search()
+{
+    perform_search(false);
+}
+
+function full_search()
+{
+    perform_search(true);
+}
+
+
+function perform_search(full)
+{
+    var text = document.getElementById("searchbox").value.toLowerCase();
+    if (text == last_search && !full) return;
+    last_search = text;
+    
+    var table = document.getElementById("indexlist");
+    var status = document.getElementById("searchmsg");
+    var children = table.firstChild.childNodes;
+    
+    // first figure out the first node with the prefix
+    var first = bisect(-1);
+    var last = (first == -1 ? -1 : bisect(1));
+
+    if (first == -1)
+    {
+        table.className = "";
+        status.innerHTML = "No results found, displaying all";
+    }
+    else if (first == 0 && last == children.length - 1)
+    {
+        table.className = "";
+        status.innerHTML = "";
+    }
+    else if (last - first >= max_results && !full)
+    {
+        table.className = "";
+        status.innerHTML = "More than " + max_results + ", press Search to display";
+    }
+    else
+    {
+        // decide what you need to clear/show
+        if (shown_range)
+            setclass(shown_range[0], shown_range[1], "indexrow");
+        setclass(first, last, "indexshow");
+        shown_range = [first, last];
+        table.className = "indexsearch";
+        status.innerHTML = "";
+    }
+
+    
+    function setclass(first, last, status)
+    {
+        for (var i = first; i <= last; i++)
+        {
+            children[i].className = status;
+        }
+    }
+    
+    
+    // do a binary search, treating 0 as ...
+    // return either -1 (no 0's found) or location of most far match
+    function bisect(dir)
+    {
+        var first = 0, finish = children.length - 1;
+        var mid, success = false;
+
+        while (finish - first > 3)
+        {
+            mid = Math.floor((finish + first) / 2);
+
+            var i = checkitem(mid);
+            if (i == 0) i = dir;
+            if (i == -1)
+                finish = mid;
+            else
+                first = mid;
+        }
+        var a = (dir == 1 ? first : finish);
+        var b = (dir == 1 ? finish : first);
+        for (var i = b; i != a - dir; i -= dir)
+        {
+            if (checkitem(i) == 0) return i;
+        }
+        return -1;
+    }    
+    
+    
+    // from an index, decide what the result is
+    // 0 = match, -1 is lower, 1 is higher
+    function checkitem(i)
+    {
+        var s = getitem(i).toLowerCase().substr(0, text.length);
+        if (s == text) return 0;
+        else return (s > text ? -1 : 1);
+    }
+    
+    
+    // from an index, get its string
+    // this abstracts over alternates
+    function getitem(i)
+    {
+        for ( ; i >= 0; i--)
+        {
+            var s = children[i].firstChild.firstChild.data;
+            if (s.indexOf(' ') == -1)
+                return s;
+        }
+        return ""; // should never be reached
+    }
+}
+
+function setSynopsis(filename) {
+    if (parent.window.synopsis) {
+        if (parent.window.synopsis.location.replace) {
+            // In Firefox this avoids adding the change to the history.
+            parent.window.synopsis.location.replace(filename);
+        } else {
+            parent.window.synopsis.location = filename;
+        }
+    }
+}
+
+function addMenuItem(html) {
+  var menu = document.getElementById("page-menu");
+  if (menu) {
+    var btn = menu.firstChild.cloneNode(false);
+    btn.innerHTML = html;
+    menu.appendChild(btn);
+  }
+}
+
+function adjustForFrames() {
+  var bodyCls;
+  
+  if (parent.location.href == window.location.href) {
+    // not in frames, so add Frames button
+    addMenuItem("<a href='#' onclick='reframe();return true;'>Frames</a>");
+    bodyCls = "no-frame";
+  }
+  else {
+    bodyCls = "in-frame";
+  }
+  addClass(document.body, bodyCls);
+}
+
+function reframe() {
+  setCookie("haddock-reframe", document.URL);
+  window.location = "frames.html";
+}
+
+function postReframe() {
+  var s = getCookie("haddock-reframe");
+  if (s) {
+    parent.window.main.location = s;
+    clearCookie("haddock-reframe");
+  }
+}
+
+function styles() {
+  var i, a, es = document.getElementsByTagName("link"), rs = [];
+  for (i = 0; a = es[i]; i++) {
+    if(a.rel.indexOf("style") != -1 && a.title) {
+      rs.push(a);
+    }
+  }
+  return rs;
+}
+
+function addStyleMenu() {
+  var as = styles();
+  var i, a, btns = "";
+  for(i=0; a = as[i]; i++) {
+    btns += "<li><a href='#' onclick=\"setActiveStyleSheet('"
+      + a.title + "'); return false;\">"
+      + a.title + "</a></li>"
+  }
+  if (as.length > 1) {
+    var h = "<div id='style-menu-holder'>"
+      + "<a href='#' onclick='styleMenu(); return false;'>Style &#9662;</a>"
+      + "<ul id='style-menu' class='hide'>" + btns + "</ul>"
+      + "</div>";
+    addMenuItem(h);
+  }
+}
+
+function setActiveStyleSheet(title) {
+  var as = styles();
+  var i, a, found;
+  for(i=0; a = as[i]; i++) {
+    a.disabled = true;
+          // need to do this always, some browsers are edge triggered
+    if(a.title == title) {
+      found = a;
+    }
+  }
+  if (found) {
+    found.disabled = false;
+    setCookie("haddock-style", title);
+  }
+  else {
+    as[0].disabled = false;
+    clearCookie("haddock-style");
+  }
+  styleMenu(false);
+}
+
+function resetStyle() {
+  var s = getCookie("haddock-style");
+  if (s) setActiveStyleSheet(s);
+}
+
+
+function styleMenu(show) {
+  var m = document.getElementById('style-menu');
+  if (m) toggleShow(m, show);
+}
+
+
+function pageLoad() {
+  addStyleMenu();
+  adjustForFrames();
+  resetStyle();
+  restoreCollapsed();
+}
+
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/hslogo-16.png b/docs/haddock/tensorflow-core-ops-0.1.0.0/hslogo-16.png
new file mode 100644
index 0000000000000000000000000000000000000000..0ff8579fbd897417b0d6dad6e920f8882138a7c0
GIT binary patch
literal 1684
zcmV;F25b3=P)<h;3K|Lk000e1NJLTq000*N000mO1^@s6{V(Jg0008+X+uL$Nkc;*
zP;zf(X>4Tx0C)j~RL^S@K@|QrZmG~B2wH0nvUrdpNm;9CMbtL^5n^i$+aI<?kYqDS
z)^0ZI?k2URh~(nYD&9ryp$Gqf9(wR(FG_`4MJV2@C}N=(E%D86Otz)!z|Ma2-Z$@k
zZ+71R4RX>n^?(HA4aZWV5ov6ELTdbo0FI&wK{O>*+w4vx20?>!`FrQsdJlnHR>OPy
zcd~b_n$otK2Za4V;76L-DzNVtaSB-y0*E}{p()372;bw_^6ZZ}PI-92wGS&j#91PI
zKs7DSe@(bk%_Y-7gGe}(^>I=@oY#w#*Bu9GZf3^F5WP>3rn}7Ut74&?PWBFvy`A)a
zPP5)V!Xd&78LdA?xQ(9mjMYElVd13a#D+Z_7&Y|xU=_C-srWU*6kiZcC!$nw*)9$7
zn6CX+@=AhmkT}X@VSsa5NKe;HZuq)~1$`#h6R+ZTR#D-3j}vF!)ZOnz+5)dI4jl{{
z44Mr{P!L4~VVJN`K!!XTF*LGrKO?IK8<Tr7btG!LbYeuYL3=jbJ-1P$-8}v%B5{;M
zwFr{@LH;VQ$xr2Z`O93e*jD$Ht(%&<^58qg<(at}9@o>z<8w`3e3jI8lUGNUta*C8
zn(P`s>{pjD=7Kek#B;Fw@hxAK%$F&Q6vg9J^Xf~4by_hu-=A!MJ3Znq&n~srbFGPs
zH&&aMXZ>nO`|hf|ljc?VPhR!${AbO?W8x_>CU%PFA&Hm8F7cAsOREdwU~R_;ot1_u
z(ruCYB-LPGn!NQdT|ZlRy+(fw^-+`=%+gee_kY4FWHg<*4sZI8+sFJD<oAl_pC|$^
zY~aY5x@}W&?+~G7rEYVs0vEs0eekw!YomTR`~+A$s}`+NHJ>270UUORdLHO0nA4V)
z%{fwsET5CQ>B?eK%uw4yQc~9?*JVo<vzb+5>2}ze(;aRcp*ceL#HUJSllrgm5wQKR
zQu+C;QrUh^8rFfA`ftFz{YAidi-`aL010qNS#tmY4c7nw4c7reD4Tcy00T@(L_t(I
z5sj2vNEA^R$7gqDc6T=2^@fUA2(c`MltuL5<|<Z1YDGjBEr^N`3VJpQ$`C4&9(*Z6
z=|*B$y;-1^F6vXRAd7+(;)77Qps2euv)AeWmj>KW>RWz$&YbU@|M|{$E*8Tu-Ux!w
z1Y*Dr&Ubfr&v-nZaaB{3ilRumrjPmk{sZvQEWlW+{o~IH|8)=s6c#X9S5s5d%<Agu
zgM$153Ttk++ZPA~ZqgqN1|^pNV-|?9-zo48n9+Yiy$*su5czgp*VFj6!1h&DRZR>J
z4@)&QH5|xQY-)^L1n<EM!01(Q6kM8}ojndgRD~2KVIa@>0pTRu0Lx9`08YTjTwn^6
z0;b1+aQ@)n;Em$q;=7BBi)v0zj&o^g>0Whp^_^5IbxIUP8C@y9;R?*Ouu}rmfxbU=
zwtWVNke-m!=`7bYEhWpcI5#)9qp`8E0lr6IQ)ARL3Ui}Af@grj8aN1=r>Cb+prlzO
zNfJs*N_tUm2ZL<f%*@E?H`Ak_J`Q#Qri<;K{{H@Bg@uKWS5{WE+}vCx91geD*4DNm
zJ?D^~dGz7lWtbWH`o<0O#|{R6h9M*GMpjnV?Y_RgatOb~1le#ikPKTNIn%<;=>%5*
zPmL2??da$TR904gL(VDAQ<BE%k@5ZA>-Fv_Dk}Pdw*4T(%*f4MKLRg=4ekMjhe2mW
zMFsBwg%ftWT}0kxRaIk1k7qJ8*#cKB;Ft{i`zVIs-Nqge;!!Ld7#O&Qqu7e0sJmP)
z$MW*>L$vSB&dxp@iA3U9fo)-7!Czlr{|o7<LYh`Fd@X2-b0fIa*Wdy$({Zt@t7`&L
zhoaGF!l5{1P1Qm;<GaSj#=Oa7atf1JZ8n=F9*@uC?b(X48O9kPoH~h;!o2=WM?ZLl
z$Lj;4K~)fAdW@^N%nzT>Hv{1oyg3xsu%gn@(b1>$;SM-ZaQ`HV=V0s;lr%d8bd;xY
zGw<i+<$17ay36gdrqk(Lxc1|{y}dSeM5OIV(FjhBg5#OM<!Z*8(@<7crlVw+B-8R1
zl{3J6w9fYQ^vojauaRgZ=~SE&CK!R3pCe!v3>Nvm3=Iu=tyXIgtJnf@A(2S@M140N
ew{UA~tMxaJq;$xaSSi*30000<MNUMnLSTa0p)&^n

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/index-frames.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/index-frames.html
new file mode 100644
index 0000000..60b888e
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/index-frames.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-list"><p class="caption">Modules</p><ul><li class="module"><a href="TensorFlow-GenOps-Core.html" target="main">TensorFlow.GenOps.Core</a></li></ul></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/index.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/index.html
new file mode 100644
index 0000000..42c5482
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</p></div><div id="content"><div id="description"><h1>tensorflow-core-ops-0.1.0.0: Haskell wrappers for Core Tensorflow Ops.</h1><div class="doc"><p>Code generated signatures for the Ops in libtensorflow_c.</p></div></div><div id="module-list"><p class="caption">Modules</p><ul><li><span id="control.n.1" class="module collapser" onclick="toggleSection('n.1')">TensorFlow</span><ul id="section.n.1" class="show"><li><span id="control.n.1.1" class="module collapser" onclick="toggleSection('n.1.1')">GenOps</span><ul id="section.n.1.1" class="show"><li><span class="module"><a href="TensorFlow-GenOps-Core.html">TensorFlow.GenOps.Core</a></span></li></ul></li></ul></li></ul></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/mini_TensorFlow-GenOps-Core.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/mini_TensorFlow-GenOps-Core.html
new file mode 100644
index 0000000..6d16294
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/mini_TensorFlow-GenOps-Core.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.GenOps.Core</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.GenOps.Core</p></div><div id="interface"><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:_HostRecv" target="main">_HostRecv</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:_Recv" target="main">_Recv</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:_Send" target="main">_Send</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:_Arg" target="main">_Arg</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyRMSProp" target="main">sparseApplyRMSProp</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:applyAdam" target="main">applyAdam</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyMomentum" target="main">sparseApplyMomentum</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:applyMomentum" target="main">applyMomentum</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:applyFtrl" target="main">applyFtrl</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyAdagradDA" target="main">sparseApplyAdagradDA</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyAdagrad" target="main">sparseApplyAdagrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:applyProximalAdagrad" target="main">applyProximalAdagrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:applyAdagrad" target="main">applyAdagrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:applyAdadelta" target="main">applyAdadelta</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyProximalGradientDescent" target="main">sparseApplyProximalGradientDescent</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:applyProximalGradientDescent" target="main">applyProximalGradientDescent</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:encodeBase64" target="main">encodeBase64</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:stringSplit" target="main">stringSplit</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:stringJoin" target="main">stringJoin</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:asString" target="main">asString</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:stringToHashBucketStrong" target="main">stringToHashBucketStrong</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:scatterMul" target="main">scatterMul</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:reduceJoin" target="main">reduceJoin</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:scatterSub" target="main">scatterSub</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:scatterAdd" target="main">scatterAdd</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:scatterUpdate" target="main">scatterUpdate</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:assignSub" target="main">assignSub</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:assignAdd" target="main">assignAdd</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseSegmentMeanGrad" target="main">sparseSegmentMeanGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseSoftmax" target="main">sparseSoftmax</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:matrixSolve" target="main">matrixSolve</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:selfAdjointEigV2" target="main">selfAdjointEigV2</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:selfAdjointEig" target="main">selfAdjointEig</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:applyGradientDescent" target="main">applyGradientDescent</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:stackPush" target="main">stackPush</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:cholesky" target="main">cholesky</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:dynamicStitch" target="main">dynamicStitch</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:readerNumWorkUnitsCompleted" target="main">readerNumWorkUnitsCompleted</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:readerRead" target="main">readerRead</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:fFT2D" target="main">fFT2D</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:fixedLengthRecordReader" target="main">fixedLengthRecordReader</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:placeholder" target="main">placeholder</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:scalarSummary" target="main">scalarSummary</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:softmax" target="main">softmax</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:shardedFilename" target="main">shardedFilename</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:_HostSend" target="main">_HostSend</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sigmoidGrad" target="main">sigmoidGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:nonMaxSuppression" target="main">nonMaxSuppression</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:identityReader" target="main">identityReader</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:extractGlimpse" target="main">extractGlimpse</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:conv3DBackpropInput" target="main">conv3DBackpropInput</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:matrixSolveLs" target="main">matrixSolveLs</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:rGBToHSV" target="main">rGBToHSV</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:decodeGif" target="main">decodeGif</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:adjustContrast" target="main">adjustContrast</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:depthToSpace" target="main">depthToSpace</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixSolve" target="main">batchMatrixSolve</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:erfc" target="main">erfc</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:resizeBilinearGrad" target="main">resizeBilinearGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:fact" target="main">fact</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:deleteSessionTensor" target="main">deleteSessionTensor</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:logicalOr" target="main">logicalOr</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:getSessionTensor" target="main">getSessionTensor</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixInverse" target="main">batchMatrixInverse</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:shardedFilespec" target="main">shardedFilespec</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:decodeBase64" target="main">decodeBase64</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:getSessionHandle" target="main">getSessionHandle</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:initializeTable" target="main">initializeTable</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tan" target="main">tan</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tanh" target="main">tanh</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:applyAdagradDA" target="main">applyAdagradDA</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:stringToHashBucket" target="main">stringToHashBucket</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:eluGrad" target="main">eluGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:fractionalAvgPoolGrad" target="main">fractionalAvgPoolGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:matrixTriangularSolve" target="main">matrixTriangularSolve</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:editDistance" target="main">editDistance</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:barrierIncompleteSize" target="main">barrierIncompleteSize</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:threadUnsafeUnigramCandidateSampler" target="main">threadUnsafeUnigramCandidateSampler</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:barrierReadySize" target="main">barrierReadySize</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:barrierClose" target="main">barrierClose</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:textLineReader" target="main">textLineReader</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:fFT3D" target="main">fFT3D</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:refExit" target="main">refExit</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:exp" target="main">exp</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:restoreSlice" target="main">restoreSlice</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:conj" target="main">conj</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:resizeNearestNeighborGrad" target="main">resizeNearestNeighborGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayClose" target="main">tensorArrayClose</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:atan" target="main">atan</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tensorArraySize" target="main">tensorArraySize</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayConcat" target="main">tensorArrayConcat</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:lRN" target="main">lRN</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:stringToHashBucketFast" target="main">stringToHashBucketFast</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayPack" target="main">tensorArrayPack</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:concatOffset" target="main">concatOffset</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:refEnter" target="main">refEnter</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:softsign" target="main">softsign</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayWrite" target="main">tensorArrayWrite</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:diag" target="main">diag</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:matrixDiagPart" target="main">matrixDiagPart</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:queueSize" target="main">queueSize</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:decodePng" target="main">decodePng</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:ceil" target="main">ceil</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:priorityQueue" target="main">priorityQueue</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:placeholderWithDefault" target="main">placeholderWithDefault</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:cropAndResizeGradImage" target="main">cropAndResizeGradImage</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:readerReset" target="main">readerReset</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:extractImagePatches" target="main">extractImagePatches</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixSetDiag" target="main">batchMatrixSetDiag</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:stackClose" target="main">stackClose</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:quantizeAndDequantize" target="main">quantizeAndDequantize</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:isNan" target="main">isNan</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:where-39-" target="main">where'</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:listDiff" target="main">listDiff</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:stridedSlice" target="main">stridedSlice</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:randomShuffleQueue" target="main">randomShuffleQueue</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tileGrad" target="main">tileGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:stridedSliceAssign" target="main">stridedSliceAssign</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:reshape" target="main">reshape</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:fIFOQueue" target="main">fIFOQueue</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:learnedUnigramCandidateSampler" target="main">learnedUnigramCandidateSampler</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:fractionalAvgPool" target="main">fractionalAvgPool</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:randomCrop" target="main">randomCrop</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:_HostCast" target="main">_HostCast</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:queueClose" target="main">queueClose</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:slice" target="main">slice</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:stridedSliceGrad" target="main">stridedSliceGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseTensorDenseAdd" target="main">sparseTensorDenseAdd</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:size" target="main">size</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:barrier" target="main">barrier</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:lgamma" target="main">lgamma</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:decodeJpeg" target="main">decodeJpeg</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:shapeN" target="main">shapeN</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:uniformCandidateSampler" target="main">uniformCandidateSampler</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:unique" target="main">unique</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:drawBoundingBoxes" target="main">drawBoundingBoxes</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tensorArraySplit" target="main">tensorArraySplit</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:split" target="main">split</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:segmentMax" target="main">segmentMax</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:abort" target="main">abort</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseReorder" target="main">sparseReorder</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:rsqrtGrad" target="main">rsqrtGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:reverseSequence" target="main">reverseSequence</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:readerNumRecordsProduced" target="main">readerNumRecordsProduced</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:deserializeManySparse" target="main">deserializeManySparse</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:immutableConst" target="main">immutableConst</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:minimum" target="main">minimum</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:initializeTableFromTextFile" target="main">initializeTableFromTextFile</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:diagPart" target="main">diagPart</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:log" target="main">log</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayScatter" target="main">tensorArrayScatter</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:rank" target="main">rank</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:identity" target="main">identity</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:adjustContrastv2" target="main">adjustContrastv2</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyProximalAdagrad" target="main">sparseApplyProximalAdagrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:gather" target="main">gather</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:isVariableInitialized" target="main">isVariableInitialized</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:concat" target="main">concat</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:randomUniformInt" target="main">randomUniformInt</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:stopGradient" target="main">stopGradient</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:avgPool" target="main">avgPool</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:wholeFileReader" target="main">wholeFileReader</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:switch" target="main">switch</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:randomStandardNormal" target="main">randomStandardNormal</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sigmoid" target="main">sigmoid</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sampleDistortedBoundingBox" target="main">sampleDistortedBoundingBox</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:greater" target="main">greater</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:refNextIteration" target="main">refNextIteration</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:spaceToDepth" target="main">spaceToDepth</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:controlTrigger" target="main">controlTrigger</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:scatterDiv" target="main">scatterDiv</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:copy" target="main">copy</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:cropAndResizeGradBoxes" target="main">cropAndResizeGradBoxes</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseSegmentMean" target="main">sparseSegmentMean</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:assign" target="main">assign</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:restore" target="main">restore</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:maxPoolGradWithArgmax" target="main">maxPoolGradWithArgmax</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:checkNumerics" target="main">checkNumerics</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:zerosLike" target="main">zerosLike</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:readFile" target="main">readFile</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:transpose" target="main">transpose</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:parseTensor" target="main">parseTensor</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:acos" target="main">acos</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:bitcast" target="main">bitcast</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:lookupTableImport" target="main">lookupTableImport</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:biasAddGrad" target="main">biasAddGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchSelfAdjointEig" target="main">batchSelfAdjointEig</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:prod" target="main">prod</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:resizeBilinear" target="main">resizeBilinear</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayUnpack" target="main">tensorArrayUnpack</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixDeterminant" target="main">batchMatrixDeterminant</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sum" target="main">sum</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:iFFT2D" target="main">iFFT2D</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:fill" target="main">fill</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:fixedUnigramCandidateSampler" target="main">fixedUnigramCandidateSampler</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:dilation2D" target="main">dilation2D</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:polygamma" target="main">polygamma</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:refIdentity" target="main">refIdentity</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:encodePng" target="main">encodePng</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:lookupTableInsert" target="main">lookupTableInsert</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchIFFT2D" target="main">batchIFFT2D</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:uniqueWithCounts" target="main">uniqueWithCounts</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:gatherNd" target="main">gatherNd</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayRead" target="main">tensorArrayRead</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:readerReadUpTo" target="main">readerReadUpTo</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:betainc" target="main">betainc</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixBandPart" target="main">batchMatrixBandPart</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:depthwiseConv2dNativeBackpropInput" target="main">depthwiseConv2dNativeBackpropInput</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:refSelect" target="main">refSelect</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:exit" target="main">exit</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:lookupTableFind" target="main">lookupTableFind</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:squeeze" target="main">squeeze</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:mean" target="main">mean</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:spaceToBatchND" target="main">spaceToBatchND</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:spaceToBatch" target="main">spaceToBatch</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:cTCGreedyDecoder" target="main">cTCGreedyDecoder</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchToSpaceND" target="main">batchToSpaceND</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:pack" target="main">pack</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:oneHot" target="main">oneHot</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:broadcastGradientArgs" target="main">broadcastGradientArgs</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:matrixSetDiag" target="main">matrixSetDiag</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:applyRMSProp" target="main">applyRMSProp</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:const" target="main">const</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:enter" target="main">enter</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:debugIdentity" target="main">debugIdentity</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:debugNanCount" target="main">debugNanCount</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchNormWithGlobalNormalization" target="main">batchNormWithGlobalNormalization</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixDiag" target="main">batchMatrixDiag</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:unpack" target="main">unpack</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseSplit" target="main">sparseSplit</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:mirrorPad" target="main">mirrorPad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixDiagPart" target="main">batchMatrixDiagPart</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:fractionalMaxPoolGrad" target="main">fractionalMaxPoolGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:matchingFiles" target="main">matchingFiles</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tile" target="main">tile</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseSparseMinimum" target="main">sparseSparseMinimum</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:allCandidateSampler" target="main">allCandidateSampler</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:refSwitch" target="main">refSwitch</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:mergeSummary" target="main">mergeSummary</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:logicalNot" target="main">logicalNot</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:lRNGrad" target="main">lRNGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:stringToNumber" target="main">stringToNumber</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseMatMul" target="main">sparseMatMul</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:merge" target="main">merge</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:choleskyGrad" target="main">choleskyGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchCholeskyGrad" target="main">batchCholeskyGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayGather" target="main">tensorArrayGather</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:resizeNearestNeighbor" target="main">resizeNearestNeighbor</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:negTrain" target="main">negTrain</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tensorArrayGrad" target="main">tensorArrayGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:audioSummary" target="main">audioSummary</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:noOp" target="main">noOp</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:nextIteration" target="main">nextIteration</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:softplusGrad" target="main">softplusGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:svd" target="main">svd</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:hSVToRGB" target="main">hSVToRGB</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:parameterizedTruncatedNormal" target="main">parameterizedTruncatedNormal</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:square" target="main">square</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:elu" target="main">elu</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:lookupTableExport" target="main">lookupTableExport</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:lookupTableSize" target="main">lookupTableSize</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:avgPoolGrad" target="main">avgPoolGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:computeAccidentalHits" target="main">computeAccidentalHits</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:cTCLoss" target="main">cTCLoss</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:avgPool3D" target="main">avgPool3D</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:inv" target="main">inv</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:stackPop" target="main">stackPop</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:paddingFIFOQueue" target="main">paddingFIFOQueue</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchSelfAdjointEigV2" target="main">batchSelfAdjointEigV2</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixTriangularSolve" target="main">batchMatrixTriangularSolve</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchMatrixSolveLs" target="main">batchMatrixSolveLs</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchSvd" target="main">batchSvd</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tensorSummary" target="main">tensorSummary</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseSoftmaxCrossEntropyWithLogits" target="main">sparseSoftmaxCrossEntropyWithLogits</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:maxPoolWithArgmax" target="main">maxPoolWithArgmax</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:fFT" target="main">fFT</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:histogramSummary" target="main">histogramSummary</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:pad" target="main">pad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchIFFT3D" target="main">batchIFFT3D</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:imageSummary" target="main">imageSummary</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:segmentSum" target="main">segmentSum</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:encodeJpeg" target="main">encodeJpeg</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchNormWithGlobalNormalizationGrad" target="main">batchNormWithGlobalNormalizationGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:biasAddV1" target="main">biasAddV1</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:invertPermutation" target="main">invertPermutation</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:mirrorPadGrad" target="main">mirrorPadGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:reverse" target="main">reverse</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:conv2D" target="main">conv2D</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:conv2DBackpropInput" target="main">conv2DBackpropInput</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:readerSerializeState" target="main">readerSerializeState</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:temporaryVariable" target="main">temporaryVariable</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:cropAndResize" target="main">cropAndResize</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:maxPoolGrad" target="main">maxPoolGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:fusedResizeAndPadConv2D" target="main">fusedResizeAndPadConv2D</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:randomUniform" target="main">randomUniform</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:depthwiseConv2dNative" target="main">depthwiseConv2dNative</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyAdadelta" target="main">sparseApplyAdadelta</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:depthwiseConv2dNativeBackpropFilter" target="main">depthwiseConv2dNativeBackpropFilter</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:conv3D" target="main">conv3D</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:greaterEqual" target="main">greaterEqual</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseDenseCwiseAdd" target="main">sparseDenseCwiseAdd</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:conv3DBackpropFilter" target="main">conv3DBackpropFilter</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:conv3DBackpropInputV2" target="main">conv3DBackpropInputV2</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:mod" target="main">mod</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:refMerge" target="main">refMerge</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:conv3DBackpropFilterV2" target="main">conv3DBackpropFilterV2</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:serializeManySparse" target="main">serializeManySparse</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:avgPool3DGrad" target="main">avgPool3DGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:maxPool3DGrad" target="main">maxPool3DGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseReduceSum" target="main">sparseReduceSum</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:relu" target="main">relu</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:l2Loss" target="main">l2Loss</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:readerRestoreState" target="main">readerRestoreState</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:shape" target="main">shape</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:softmaxCrossEntropyWithLogits" target="main">softmaxCrossEntropyWithLogits</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:maxPool" target="main">maxPool</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:dilation2DBackpropInput" target="main">dilation2DBackpropInput</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:equal" target="main">equal</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:dilation2DBackpropFilter" target="main">dilation2DBackpropFilter</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:reluGrad" target="main">reluGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:relu6" target="main">relu6</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:resizeBicubic" target="main">resizeBicubic</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:relu6Grad" target="main">relu6Grad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseTensorDenseMatMul" target="main">sparseTensorDenseMatMul</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:softplus" target="main">softplus</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchMatMul" target="main">batchMatMul</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:softsignGrad" target="main">softsignGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:lessEqual" target="main">lessEqual</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:logSoftmax" target="main">logSoftmax</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:inTopK" target="main">inTopK</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:matrixDiag" target="main">matrixDiag</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:maxPool3D" target="main">maxPool3D</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:topK" target="main">topK</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:topKV2" target="main">topKV2</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:fractionalMaxPool" target="main">fractionalMaxPool</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:matrixBandPart" target="main">matrixBandPart</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:decodeRaw" target="main">decodeRaw</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:decodeJSONExample" target="main">decodeJSONExample</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:truncatedNormal" target="main">truncatedNormal</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:randomShuffle" target="main">randomShuffle</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:multinomial" target="main">multinomial</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:randomGamma" target="main">randomGamma</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:addN" target="main">addN</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:max" target="main">max</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:_Retval" target="main">_Retval</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:destroyTemporaryVariable" target="main">destroyTemporaryVariable</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:cast" target="main">cast</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:countUpTo" target="main">countUpTo</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:abs" target="main">abs</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:neg" target="main">neg</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseSparseMaximum" target="main">sparseSparseMaximum</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:invGrad" target="main">invGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sqrt" target="main">sqrt</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:matrixInverse" target="main">matrixInverse</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sqrtGrad" target="main">sqrtGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:expandDims" target="main">expandDims</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:all" target="main">all</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:cTCBeamSearchDecoder" target="main">cTCBeamSearchDecoder</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:rsqrt" target="main">rsqrt</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tanhGrad" target="main">tanhGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sin" target="main">sin</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:matrixDeterminant" target="main">matrixDeterminant</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:cos" target="main">cos</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchToSpace" target="main">batchToSpace</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseToDense" target="main">sparseToDense</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:asin" target="main">asin</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:argMin" target="main">argMin</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:isInf" target="main">isInf</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sign" target="main">sign</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:add" target="main">add</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseApplyFtrl" target="main">sparseApplyFtrl</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sub" target="main">sub</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchFFT3D" target="main">batchFFT3D</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseReduceSumSparse" target="main">sparseReduceSumSparse</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:biasAdd" target="main">biasAdd</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:mul" target="main">mul</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:div" target="main">div</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:loopCond" target="main">loopCond</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:squaredDifference" target="main">squaredDifference</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:maximum" target="main">maximum</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:logUniformCandidateSampler" target="main">logUniformCandidateSampler</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:less" target="main">less</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:pow" target="main">pow</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:igammac" target="main">igammac</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:igamma" target="main">igamma</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:zeta" target="main">zeta</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:imag" target="main">imag</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:complex" target="main">complex</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:notEqual" target="main">notEqual</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:complexAbs" target="main">complexAbs</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:logicalAnd" target="main">logicalAnd</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchFFT" target="main">batchFFT</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:select" target="main">select</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:matMul" target="main">matMul</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:digamma" target="main">digamma</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:conv2DBackpropFilter" target="main">conv2DBackpropFilter</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:min" target="main">min</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:isFinite" target="main">isFinite</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:argMax" target="main">argMax</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:segmentMean" target="main">segmentMean</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:cumprod" target="main">cumprod</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:segmentMin" target="main">segmentMin</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:unsortedSegmentSum" target="main">unsortedSegmentSum</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:tFRecordReader" target="main">tFRecordReader</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseSegmentSum" target="main">sparseSegmentSum</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseSegmentSqrtN" target="main">sparseSegmentSqrtN</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:copyHost" target="main">copyHost</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:variable" target="main">variable</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseSegmentSqrtNGrad" target="main">sparseSegmentSqrtNGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:range" target="main">range</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:any" target="main">any</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:linSpace" target="main">linSpace</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:resizeArea" target="main">resizeArea</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:real" target="main">real</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:iFFT" target="main">iFFT</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:iFFT3D" target="main">iFFT3D</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:cross" target="main">cross</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:cumsum" target="main">cumsum</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchIFFT" target="main">batchIFFT</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:erf" target="main">erf</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:barrierInsertMany" target="main">barrierInsertMany</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:floor" target="main">floor</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchFFT2D" target="main">batchFFT2D</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseAddGrad" target="main">sparseAddGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseAdd" target="main">sparseAdd</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:batchCholesky" target="main">batchCholesky</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:dynamicPartition" target="main">dynamicPartition</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:serializeSparse" target="main">serializeSparse</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseConcat" target="main">sparseConcat</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:segmentProd" target="main">segmentProd</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseReshape" target="main">sparseReshape</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseDenseCwiseMul" target="main">sparseDenseCwiseMul</a></p></div><div class="top"><p class="src"><a href="TensorFlow-GenOps-Core.html#v:sparseDenseCwiseDiv" target="main">sparseDenseCwiseDiv</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/minus.gif b/docs/haddock/tensorflow-core-ops-0.1.0.0/minus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..1deac2fe1a42e35b994f1b855488f392c50f6a89
GIT binary patch
literal 56
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS53HJ0a;6C;`j@u!^I(6E|
Jh4LZ{)&P#)4$%Mr

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/ocean.css b/docs/haddock/tensorflow-core-ops-0.1.0.0/ocean.css
new file mode 100644
index 0000000..1110b40
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/ocean.css
@@ -0,0 +1,600 @@
+/* @group Fundamentals */
+
+* { margin: 0; padding: 0 }
+
+/* Is this portable? */
+html {
+  background-color: white;
+  width: 100%;
+  height: 100%;
+}
+
+body {
+  background: white;
+  color: black;
+  text-align: left;
+  min-height: 100%;
+  position: relative;
+}
+
+p {
+  margin: 0.8em 0;
+}
+
+ul, ol {
+  margin: 0.8em 0 0.8em 2em;
+}
+
+dl {
+  margin: 0.8em 0;
+}
+
+dt {
+  font-weight: bold;
+}
+dd {
+  margin-left: 2em;
+}
+
+a { text-decoration: none; }
+a[href]:link { color: rgb(196,69,29); }
+a[href]:visited { color: rgb(171,105,84); }
+a[href]:hover { text-decoration:underline; }
+
+/* @end */
+
+/* @group Fonts & Sizes */
+
+/* Basic technique & IE workarounds from YUI 3
+   For reasons, see:
+      http://yui.yahooapis.com/3.1.1/build/cssfonts/fonts.css
+ */
+
+body {
+	font:13px/1.4 sans-serif;
+	*font-size:small; /* for IE */
+	*font:x-small; /* for IE in quirks mode */
+}
+
+h1 { font-size: 146.5%; /* 19pt */ }
+h2 { font-size: 131%;   /* 17pt */ }
+h3 { font-size: 116%;   /* 15pt */ }
+h4 { font-size: 100%;   /* 13pt */ }
+h5 { font-size: 100%;   /* 13pt */ }
+
+select, input, button, textarea {
+	font:99% sans-serif;
+}
+
+table {
+	font-size:inherit;
+	font:100%;
+}
+
+pre, code, kbd, samp, tt, .src {
+	font-family:monospace;
+	*font-size:108%;
+	line-height: 124%;
+}
+
+.links, .link {
+  font-size: 85%; /* 11pt */
+}
+
+#module-header .caption {
+  font-size: 182%; /* 24pt */
+}
+
+.info  {
+  font-size: 85%; /* 11pt */
+}
+
+#table-of-contents, #synopsis  {
+  /* font-size: 85%; /* 11pt */
+}
+
+
+/* @end */
+
+/* @group Common */
+
+.caption, h1, h2, h3, h4, h5, h6 {
+  font-weight: bold;
+  color: rgb(78,98,114);
+  margin: 0.8em 0 0.4em;
+}
+
+* + h1, * + h2, * + h3, * + h4, * + h5, * + h6 {
+  margin-top: 2em;
+}
+
+h1 + h2, h2 + h3, h3 + h4, h4 + h5, h5 + h6 {
+  margin-top: inherit;
+}
+
+ul.links {
+  list-style: none;
+  text-align: left;
+  float: right;
+  display: inline-table;
+  margin: 0 0 0 1em;
+}
+
+ul.links li {
+  display: inline;
+  border-left: 1px solid #d5d5d5;
+  white-space: nowrap;
+  padding: 0;
+}
+
+ul.links li a {
+  padding: 0.2em 0.5em;
+}
+
+.hide { display: none; }
+.show { display: inherit; }
+.clear { clear: both; }
+
+.collapser {
+  background-image: url(minus.gif);
+  background-repeat: no-repeat;
+}
+.expander {
+  background-image: url(plus.gif);
+  background-repeat: no-repeat;
+}
+p.caption.collapser,
+p.caption.expander {
+  background-position: 0 0.4em;
+}
+.collapser, .expander {
+  padding-left: 14px;
+  margin-left: -14px;
+  cursor: pointer;
+}
+
+pre {
+  padding: 0.25em;
+  margin: 0.8em 0;
+  background: rgb(229,237,244);
+  overflow: auto;
+  border-bottom: 0.25em solid white;
+  /* white border adds some space below the box to compensate
+     for visual extra space that paragraphs have between baseline
+     and the bounding box */
+}
+
+.src {
+  background: #f0f0f0;
+  padding: 0.2em 0.5em;
+}
+
+.keyword { font-weight: normal; }
+.def { font-weight: bold; }
+
+
+/* @end */
+
+/* @group Page Structure */
+
+#content {
+  margin: 0 auto;
+  padding: 0 2em 6em;
+}
+
+#package-header {
+  background: rgb(41,56,69);
+  border-top: 5px solid rgb(78,98,114);
+  color: #ddd;
+  padding: 0.2em;
+  position: relative;
+  text-align: left;
+}
+
+#package-header .caption {
+  background: url(hslogo-16.png) no-repeat 0em;
+  color: white;
+  margin: 0 2em;
+  font-weight: normal;
+  font-style: normal;
+  padding-left: 2em;
+}
+
+#package-header a:link, #package-header a:visited { color: white; }
+#package-header a:hover { background: rgb(78,98,114); }
+
+#module-header .caption {
+  color: rgb(78,98,114);
+  font-weight: bold;
+  border-bottom: 1px solid #ddd;
+}
+
+table.info {
+  float: right;
+  padding: 0.5em 1em;
+  border: 1px solid #ddd;
+  color: rgb(78,98,114);
+  background-color: #fff;
+  max-width: 40%;
+  border-spacing: 0;
+  position: relative;
+  top: -0.5em;
+  margin: 0 0 0 2em;
+}
+
+.info th {
+	padding: 0 1em 0 0;
+}
+
+div#style-menu-holder {
+  position: relative;
+  z-index: 2;
+  display: inline;
+}
+
+#style-menu {
+  position: absolute;
+  z-index: 1;
+  overflow: visible;
+  background: #374c5e;
+  margin: 0;
+  text-align: center;
+  right: 0;
+  padding: 0;
+  top: 1.25em;
+}
+
+#style-menu li {
+	display: list-item;
+	border-style: none;
+	margin: 0;
+	padding: 0;
+	color: #000;
+	list-style-type: none;
+}
+
+#style-menu li + li {
+	border-top: 1px solid #919191;
+}
+
+#style-menu a {
+  width: 6em;
+  padding: 3px;
+  display: block;
+}
+
+#footer {
+  background: #ddd;
+  border-top: 1px solid #aaa;
+  padding: 0.5em 0;
+  color: #666;
+  text-align: center;
+  position: absolute;
+  bottom: 0;
+  width: 100%;
+  height: 3em;
+}
+
+/* @end */
+
+/* @group Front Matter */
+
+#table-of-contents {
+  float: right;
+  clear: right;
+  background: #faf9dc;
+  border: 1px solid #d8d7ad;
+  padding: 0.5em 1em;
+  max-width: 20em;
+  margin: 0.5em 0 1em 1em;
+}
+
+#table-of-contents .caption {
+  text-align: center;
+  margin: 0;
+}
+
+#table-of-contents ul {
+  list-style: none;
+  margin: 0;
+}
+
+#table-of-contents ul ul {
+  margin-left: 2em;
+}
+
+#description .caption {
+  display: none;
+}
+
+#synopsis {
+  display: none;
+}
+
+.no-frame #synopsis {
+  display: block;
+  position: fixed;
+  right: 0;
+  height: 80%;
+  top: 10%;
+  padding: 0;
+  max-width: 75%;
+}
+
+#synopsis .caption {
+  float: left;
+  width: 29px;
+  color: rgba(255,255,255,0);
+  height: 110px;
+  margin: 0;
+  font-size: 1px;
+  padding: 0;
+}
+
+#synopsis p.caption.collapser {
+  background: url(synopsis.png) no-repeat -64px -8px;
+}
+
+#synopsis p.caption.expander {
+  background: url(synopsis.png) no-repeat 0px -8px;
+}
+
+#synopsis ul {
+  height: 100%;
+  overflow: auto;
+  padding: 0.5em;
+  margin: 0;
+}
+
+#synopsis ul ul {
+  overflow: hidden;
+}
+
+#synopsis ul,
+#synopsis ul li.src {
+  background-color: #faf9dc;
+  white-space: nowrap;
+  list-style: none;
+  margin-left: 0;
+}
+
+/* @end */
+
+/* @group Main Content */
+
+#interface div.top { margin: 2em 0; }
+#interface h1 + div.top,
+#interface h2 + div.top,
+#interface h3 + div.top,
+#interface h4 + div.top,
+#interface h5 + div.top {
+ 	margin-top: 1em;
+}
+#interface p.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface td.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface span.fixity {
+  color: #919191;
+  border-left: 1px solid #919191;
+  padding: 0.2em 0.5em 0.2em 0.5em;
+  margin: 0 -1em 0 1em;
+}
+
+#interface span.rightedge {
+  border-left: 1px solid #919191;
+  padding: 0.2em 0 0.2em 0;
+  margin: 0 0 0 1em;
+}
+
+#interface table { border-spacing: 2px; }
+#interface td {
+  vertical-align: top;
+  padding-left: 0.5em;
+}
+#interface td.src {
+  white-space: nowrap;
+}
+#interface td.doc p {
+  margin: 0;
+}
+#interface td.doc p + p {
+  margin-top: 0.8em;
+}
+
+.clearfix:after {
+  clear: both;
+  content: " ";
+  display: block;
+  height: 0;
+  visibility: hidden;
+}
+
+.subs dl {
+  margin: 0;
+}
+
+.subs dt {
+  float: left;
+  clear: left;
+  display: block;
+  margin: 1px 0;
+}
+
+.subs dd {
+  float: right;
+  width: 90%;
+  display: block;
+  padding-left: 0.5em;
+  margin-bottom: 0.5em;
+}
+
+.subs dd.empty {
+  display: none;
+}
+
+.subs dd p {
+  margin: 0;
+}
+
+/* Render short-style data instances */
+.inst ul {
+  height: 100%;
+  padding: 0.5em;
+  margin: 0;
+}
+
+.inst, .inst li {
+  list-style: none;
+  margin-left: 1em;
+}
+
+/* Workaround for bug in Firefox (issue #384) */
+.inst-left {
+  float: left;
+}
+
+.top p.src {
+  border-top: 1px solid #ccc;
+}
+
+.subs, .doc {
+  /* use this selector for one level of indent */
+  padding-left: 2em;
+}
+
+.warning {
+  color: red;
+}
+
+.arguments {
+  margin-top: -0.4em;
+}
+.arguments .caption {
+  display: none;
+}
+
+.fields { padding-left: 1em; }
+
+.fields .caption { display: none; }
+
+.fields p { margin: 0 0; }
+
+/* this seems bulky to me
+.methods, .constructors {
+  background: #f8f8f8;
+  border: 1px solid #eee;
+}
+*/
+
+/* @end */
+
+/* @group Auxillary Pages */
+
+
+.extension-list {
+    list-style-type: none;
+    margin-left: 0;
+}
+
+#mini {
+  margin: 0 auto;
+  padding: 0 1em 1em;
+}
+
+#mini > * {
+  font-size: 93%; /* 12pt */
+}
+
+#mini #module-list .caption,
+#mini #module-header .caption {
+  font-size: 125%; /* 15pt */
+}
+
+#mini #interface h1,
+#mini #interface h2,
+#mini #interface h3,
+#mini #interface h4 {
+  font-size: 109%; /* 13pt */
+  margin: 1em 0 0;
+}
+
+#mini #interface .top,
+#mini #interface .src {
+  margin: 0;
+}
+
+#mini #module-list ul {
+  list-style: none;
+  margin: 0;
+}
+
+#alphabet ul {
+	list-style: none;
+	padding: 0;
+	margin: 0.5em 0 0;
+	text-align: center;
+}
+
+#alphabet li {
+	display: inline;
+	margin: 0 0.25em;
+}
+
+#alphabet a {
+	font-weight: bold;
+}
+
+#index .caption,
+#module-list .caption { font-size: 131%; /* 17pt */ }
+
+#index table {
+  margin-left: 2em;
+}
+
+#index .src {
+  font-weight: bold;
+}
+#index .alt {
+  font-size: 77%; /* 10pt */
+  font-style: italic;
+  padding-left: 2em;
+}
+
+#index td + td {
+  padding-left: 1em;
+}
+
+#module-list ul {
+  list-style: none;
+  margin: 0 0 0 2em;
+}
+
+#module-list li {
+  clear: right;
+}
+
+#module-list span.collapser,
+#module-list span.expander {
+  background-position: 0 0.3em;
+}
+
+#module-list .package {
+  float: right;
+}
+
+/* @end */
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/plus.gif b/docs/haddock/tensorflow-core-ops-0.1.0.0/plus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..2d15c14173d23f664b955cd24f51c82f5f09d91d
GIT binary patch
literal 59
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS5iS+a@;NG@rPSdH*>gbBX
M^XE!9f*2UA0nx1yDgXcg

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/src/TensorFlow-GenOps-Core.html b/docs/haddock/tensorflow-core-ops-0.1.0.0/src/TensorFlow-GenOps-Core.html
new file mode 100644
index 0000000..eb0a175
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/src/TensorFlow-GenOps-Core.html
@@ -0,0 +1,22618 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/TensorFlow/GenOps/Core.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>{-# LANGUAGE ConstraintKinds #-}</span>
+<a name="line-2"></a><span class='hs-comment'>{-# LANGUAGE DataKinds #-}</span>
+<a name="line-3"></a><span class='hs-comment'>{-# LANGUAGE FlexibleInstances #-}</span>
+<a name="line-4"></a><span class='hs-comment'>{-# LANGUAGE OverloadedStrings #-}</span>
+<a name="line-5"></a><span class='hs-comment'>{-# LANGUAGE RankNTypes #-}</span>
+<a name="line-6"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables #-}</span>
+<a name="line-7"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>GenOps</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span> <span class='hs-keyword'>where</span>
+<a name="line-8"></a>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-layout'>(</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-layout'>(</span><span class='hs-conid'>Complex</span><span class='hs-layout'>)</span>
+<a name="line-11"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span> <span class='hs-layout'>(</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-12"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span> <span class='hs-layout'>(</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Word16</span><span class='hs-layout'>)</span>
+<a name="line-13"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varop'>.~</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>&amp;</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-14"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Build</span>
+<a name="line-15"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>BuildOp</span>
+<a name="line-16"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span>
+<a name="line-17"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span>
+<a name="line-18"></a>
+<a name="line-19"></a><span class='hs-comment'>-- | Receives the named tensor from send_device on recv_device.</span>
+<a name="line-20"></a><span class='hs-comment'>--</span>
+<a name="line-21"></a><span class='hs-comment'>-- _HostRecv requires its input on host memory whereas _Recv requires its</span>
+<a name="line-22"></a><span class='hs-comment'>-- input on device memory.</span>
+<a name="line-23"></a><span class='hs-sel'>_HostRecv</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>tensor_type</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>tensor_type</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-24"></a>             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __send_device_incarnation__: The current incarnation of send_device.</span>
+<a name="line-25"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>tensor_type</span> <span class='hs-comment'>-- ^ __tensor__: The tensor to receive.</span>
+<a name="line-26"></a><span class='hs-sel'>_HostRecv</span> <span class='hs-varid'>send_device_incarnation</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-27"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"_HostRecv"</span>
+<a name="line-28"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"tensor_type"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tensor_type</span><span class='hs-layout'>)</span>
+<a name="line-29"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"send_device_incarnation"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>send_device_incarnation</span><span class='hs-layout'>)</span>
+<a name="line-30"></a>        
+<a name="line-31"></a><span class='hs-comment'>{-
+<a name="line-32"></a>attr { name: "tensor_type" type: "type" }
+<a name="line-33"></a>attr {
+<a name="line-34"></a>  description: "The name of the tensor to receive."
+<a name="line-35"></a>  name: "tensor_name"
+<a name="line-36"></a>  type: "string"
+<a name="line-37"></a>}
+<a name="line-38"></a>attr {
+<a name="line-39"></a>  description: "The name of the device sending the tensor."
+<a name="line-40"></a>  name: "send_device"
+<a name="line-41"></a>  type: "string"
+<a name="line-42"></a>}
+<a name="line-43"></a>attr {
+<a name="line-44"></a>  description: "The current incarnation of send_device."
+<a name="line-45"></a>  name: "send_device_incarnation"
+<a name="line-46"></a>  type: "int"
+<a name="line-47"></a>}
+<a name="line-48"></a>attr {
+<a name="line-49"></a>  description: "The name of the device receiving the tensor."
+<a name="line-50"></a>  name: "recv_device"
+<a name="line-51"></a>  type: "string"
+<a name="line-52"></a>}
+<a name="line-53"></a>attr {
+<a name="line-54"></a>  default_value { b: false }
+<a name="line-55"></a>  description: "If set to true, this indicates that the node was added\nto the graph as a result of a client-side feed or fetch of Tensor data,\nin which case the corresponding send or recv is expected to be managed\nlocally by the caller."
+<a name="line-56"></a>  name: "client_terminated"
+<a name="line-57"></a>  type: "bool"
+<a name="line-58"></a>}
+<a name="line-59"></a>output_arg {
+<a name="line-60"></a>  description: "The tensor to receive."
+<a name="line-61"></a>  name: "tensor"
+<a name="line-62"></a>  type_attr: "tensor_type"
+<a name="line-63"></a>}
+<a name="line-64"></a>-}</span>
+<a name="line-65"></a>
+<a name="line-66"></a><span class='hs-comment'>-- | Receives the named tensor from send_device on recv_device.</span>
+<a name="line-67"></a>
+<a name="line-68"></a><span class='hs-sel'>_Recv</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>tensor_type</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>tensor_type</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-69"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __send_device_incarnation__: The current incarnation of send_device.</span>
+<a name="line-70"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>tensor_type</span> <span class='hs-comment'>-- ^ __tensor__: The tensor to receive.</span>
+<a name="line-71"></a><span class='hs-sel'>_Recv</span> <span class='hs-varid'>send_device_incarnation</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-72"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"_Recv"</span>
+<a name="line-73"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"tensor_type"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tensor_type</span><span class='hs-layout'>)</span>
+<a name="line-74"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"send_device_incarnation"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>send_device_incarnation</span><span class='hs-layout'>)</span>
+<a name="line-75"></a>        
+<a name="line-76"></a><span class='hs-comment'>{-
+<a name="line-77"></a>attr { name: "tensor_type" type: "type" }
+<a name="line-78"></a>attr {
+<a name="line-79"></a>  description: "The name of the tensor to receive."
+<a name="line-80"></a>  name: "tensor_name"
+<a name="line-81"></a>  type: "string"
+<a name="line-82"></a>}
+<a name="line-83"></a>attr {
+<a name="line-84"></a>  description: "The name of the device sending the tensor."
+<a name="line-85"></a>  name: "send_device"
+<a name="line-86"></a>  type: "string"
+<a name="line-87"></a>}
+<a name="line-88"></a>attr {
+<a name="line-89"></a>  description: "The current incarnation of send_device."
+<a name="line-90"></a>  name: "send_device_incarnation"
+<a name="line-91"></a>  type: "int"
+<a name="line-92"></a>}
+<a name="line-93"></a>attr {
+<a name="line-94"></a>  description: "The name of the device receiving the tensor."
+<a name="line-95"></a>  name: "recv_device"
+<a name="line-96"></a>  type: "string"
+<a name="line-97"></a>}
+<a name="line-98"></a>attr {
+<a name="line-99"></a>  default_value { b: false }
+<a name="line-100"></a>  description: "If set to true, this indicates that the node was added\nto the graph as a result of a client-side feed or fetch of Tensor data,\nin which case the corresponding send or recv is expected to be managed\nlocally by the caller."
+<a name="line-101"></a>  name: "client_terminated"
+<a name="line-102"></a>  type: "bool"
+<a name="line-103"></a>}
+<a name="line-104"></a>output_arg {
+<a name="line-105"></a>  description: "The tensor to receive."
+<a name="line-106"></a>  name: "tensor"
+<a name="line-107"></a>  type_attr: "tensor_type"
+<a name="line-108"></a>}
+<a name="line-109"></a>-}</span>
+<a name="line-110"></a>
+<a name="line-111"></a><span class='hs-comment'>-- | Sends the named tensor from send_device to recv_device.</span>
+<a name="line-112"></a>
+<a name="line-113"></a><span class='hs-sel'>_Send</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-114"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __send_device_incarnation__: The current incarnation of send_device.</span>
+<a name="line-115"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __tensor__: The tensor to send.</span>
+<a name="line-116"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-117"></a><span class='hs-sel'>_Send</span> <span class='hs-varid'>send_device_incarnation</span> <span class='hs-varid'>tensor</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-118"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"_Send"</span>
+<a name="line-119"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-120"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"send_device_incarnation"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>send_device_incarnation</span><span class='hs-layout'>)</span>
+<a name="line-121"></a>        <span class='hs-varid'>tensor</span>
+<a name="line-122"></a><span class='hs-comment'>{-
+<a name="line-123"></a>attr { name: "T" type: "type" }
+<a name="line-124"></a>attr {
+<a name="line-125"></a>  description: "The name of the tensor to send."
+<a name="line-126"></a>  name: "tensor_name"
+<a name="line-127"></a>  type: "string"
+<a name="line-128"></a>}
+<a name="line-129"></a>attr {
+<a name="line-130"></a>  description: "The name of the device sending the tensor."
+<a name="line-131"></a>  name: "send_device"
+<a name="line-132"></a>  type: "string"
+<a name="line-133"></a>}
+<a name="line-134"></a>attr {
+<a name="line-135"></a>  description: "The current incarnation of send_device."
+<a name="line-136"></a>  name: "send_device_incarnation"
+<a name="line-137"></a>  type: "int"
+<a name="line-138"></a>}
+<a name="line-139"></a>attr {
+<a name="line-140"></a>  description: "The name of the device receiving the tensor."
+<a name="line-141"></a>  name: "recv_device"
+<a name="line-142"></a>  type: "string"
+<a name="line-143"></a>}
+<a name="line-144"></a>attr {
+<a name="line-145"></a>  default_value { b: false }
+<a name="line-146"></a>  description: "If set to true, this indicates that the node was added\nto the graph as a result of a client-side feed or fetch of Tensor data,\nin which case the corresponding send or recv is expected to be managed\nlocally by the caller."
+<a name="line-147"></a>  name: "client_terminated"
+<a name="line-148"></a>  type: "bool"
+<a name="line-149"></a>}
+<a name="line-150"></a>input_arg {
+<a name="line-151"></a>  description: "The tensor to send." name: "tensor" type_attr: "T"
+<a name="line-152"></a>}
+<a name="line-153"></a>-}</span>
+<a name="line-154"></a>
+<a name="line-155"></a><span class='hs-comment'>-- | A graph node which represents an argument to a function.</span>
+<a name="line-156"></a>
+<a name="line-157"></a><span class='hs-sel'>_Arg</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-158"></a>        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __index__: This argument is the index-th argument of the function.</span>
+<a name="line-159"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The argument.</span>
+<a name="line-160"></a><span class='hs-sel'>_Arg</span> <span class='hs-varid'>index</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-161"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"_Arg"</span>
+<a name="line-162"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-163"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"index"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>index</span><span class='hs-layout'>)</span>
+<a name="line-164"></a>        
+<a name="line-165"></a><span class='hs-comment'>{-
+<a name="line-166"></a>attr { name: "T" type: "type" }
+<a name="line-167"></a>attr {
+<a name="line-168"></a>  description: "This argument is the index-th argument of the function."
+<a name="line-169"></a>  has_minimum: true
+<a name="line-170"></a>  name: "index"
+<a name="line-171"></a>  type: "int"
+<a name="line-172"></a>}
+<a name="line-173"></a>output_arg {
+<a name="line-174"></a>  description: "The argument." name: "output" type_attr: "T"
+<a name="line-175"></a>}
+<a name="line-176"></a>-}</span>
+<a name="line-177"></a>
+<a name="line-178"></a><a name="sparseApplyRMSProp"></a><span class='hs-comment'>-- | Update '*var' according to the RMSProp algorithm.</span>
+<a name="line-179"></a><span class='hs-comment'>--</span>
+<a name="line-180"></a><span class='hs-comment'>-- Note that in dense implement of this algorithm, ms and mom will</span>
+<a name="line-181"></a><span class='hs-comment'>-- update even if the grad is zero, but in this sparse implement, ms</span>
+<a name="line-182"></a><span class='hs-comment'>-- and mom will not update in iterations the grad is zero.</span>
+<a name="line-183"></a><span class='hs-comment'>-- </span>
+<a name="line-184"></a><span class='hs-comment'>-- mean_square = decay * mean_square + (1-decay) * gradient ** 2</span>
+<a name="line-185"></a><span class='hs-comment'>-- Delta = learning_rate * gradient / sqrt(mean_square + epsilon)</span>
+<a name="line-186"></a><span class='hs-comment'>-- </span>
+<a name="line-187"></a><span class='hs-comment'>-- ms &lt;- rho * ms_{t-1} + (1-rho) * grad * grad</span>
+<a name="line-188"></a><span class='hs-comment'>-- mom &lt;- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)</span>
+<a name="line-189"></a><span class='hs-comment'>-- var &lt;- var - mom</span>
+<a name="line-190"></a><span class='hs-definition'>sparseApplyRMSProp</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>v8</span> <span class='hs-varid'>v9</span> <span class='hs-varid'>t</span>
+<a name="line-191"></a>                      <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-192"></a>                                  <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-193"></a>                                          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-194"></a>                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-195"></a>                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-196"></a>                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-197"></a>                                          <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-198"></a>                                  <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-199"></a>                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-200"></a>                      <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-201"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __ms__: Should be from a Variable().</span>
+<a name="line-202"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __mom__: Should be from a Variable().</span>
+<a name="line-203"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr__: Scaling factor. Must be a scalar.</span>
+<a name="line-204"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __rho__: Decay rate. Must be a scalar.</span>
+<a name="line-205"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __momentum__</span>
+<a name="line-206"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __epsilon__: Ridge term. Must be a scalar.</span>
+<a name="line-207"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v8</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: The gradient.</span>
+<a name="line-208"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v9</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __indices__: A vector of indices into the first dimension of var, ms and mom.</span>
+<a name="line-209"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-210"></a><span class='hs-definition'>sparseApplyRMSProp</span> <span class='hs-varid'>var</span> <span class='hs-varid'>ms</span> <span class='hs-varid'>mom</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>rho</span> <span class='hs-varid'>momentum</span> <span class='hs-varid'>epsilon</span> <span class='hs-varid'>grad</span>
+<a name="line-211"></a>                   <span class='hs-varid'>indices</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-212"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseApplyRMSProp"</span>
+<a name="line-213"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-214"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-215"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>ms</span> <span class='hs-varid'>mom</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>rho</span> <span class='hs-varid'>momentum</span> <span class='hs-varid'>epsilon</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>indices</span>
+<a name="line-216"></a><span class='hs-comment'>{-
+<a name="line-217"></a>attr {
+<a name="line-218"></a>  allowed_values {
+<a name="line-219"></a>    list {
+<a name="line-220"></a>      type: DT_FLOAT
+<a name="line-221"></a>      type: DT_DOUBLE
+<a name="line-222"></a>      type: DT_INT64
+<a name="line-223"></a>      type: DT_INT32
+<a name="line-224"></a>      type: DT_UINT8
+<a name="line-225"></a>      type: DT_UINT16
+<a name="line-226"></a>      type: DT_INT16
+<a name="line-227"></a>      type: DT_INT8
+<a name="line-228"></a>      type: DT_COMPLEX64
+<a name="line-229"></a>      type: DT_COMPLEX128
+<a name="line-230"></a>      type: DT_QINT8
+<a name="line-231"></a>      type: DT_QUINT8
+<a name="line-232"></a>      type: DT_QINT32
+<a name="line-233"></a>      type: DT_HALF
+<a name="line-234"></a>    }
+<a name="line-235"></a>  }
+<a name="line-236"></a>  name: "T"
+<a name="line-237"></a>  type: "type"
+<a name="line-238"></a>}
+<a name="line-239"></a>attr {
+<a name="line-240"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-241"></a>  name: "Tindices"
+<a name="line-242"></a>  type: "type"
+<a name="line-243"></a>}
+<a name="line-244"></a>attr {
+<a name="line-245"></a>  default_value { b: false }
+<a name="line-246"></a>  description: "If `True`, updating of the var, m, and v tensors will be protected\nby a lock; otherwise the behavior is undefined, but may exhibit less\ncontention."
+<a name="line-247"></a>  name: "use_locking"
+<a name="line-248"></a>  type: "bool"
+<a name="line-249"></a>}
+<a name="line-250"></a>input_arg {
+<a name="line-251"></a>  description: "Should be from a Variable()."
+<a name="line-252"></a>  is_ref: true
+<a name="line-253"></a>  name: "var"
+<a name="line-254"></a>  type_attr: "T"
+<a name="line-255"></a>}
+<a name="line-256"></a>input_arg {
+<a name="line-257"></a>  description: "Should be from a Variable()."
+<a name="line-258"></a>  is_ref: true
+<a name="line-259"></a>  name: "ms"
+<a name="line-260"></a>  type_attr: "T"
+<a name="line-261"></a>}
+<a name="line-262"></a>input_arg {
+<a name="line-263"></a>  description: "Should be from a Variable()."
+<a name="line-264"></a>  is_ref: true
+<a name="line-265"></a>  name: "mom"
+<a name="line-266"></a>  type_attr: "T"
+<a name="line-267"></a>}
+<a name="line-268"></a>input_arg {
+<a name="line-269"></a>  description: "Scaling factor. Must be a scalar."
+<a name="line-270"></a>  name: "lr"
+<a name="line-271"></a>  type_attr: "T"
+<a name="line-272"></a>}
+<a name="line-273"></a>input_arg {
+<a name="line-274"></a>  description: "Decay rate. Must be a scalar."
+<a name="line-275"></a>  name: "rho"
+<a name="line-276"></a>  type_attr: "T"
+<a name="line-277"></a>}
+<a name="line-278"></a>input_arg { name: "momentum" type_attr: "T" }
+<a name="line-279"></a>input_arg {
+<a name="line-280"></a>  description: "Ridge term. Must be a scalar."
+<a name="line-281"></a>  name: "epsilon"
+<a name="line-282"></a>  type_attr: "T"
+<a name="line-283"></a>}
+<a name="line-284"></a>input_arg {
+<a name="line-285"></a>  description: "The gradient." name: "grad" type_attr: "T"
+<a name="line-286"></a>}
+<a name="line-287"></a>input_arg {
+<a name="line-288"></a>  description: "A vector of indices into the first dimension of var, ms and mom."
+<a name="line-289"></a>  name: "indices"
+<a name="line-290"></a>  type_attr: "Tindices"
+<a name="line-291"></a>}
+<a name="line-292"></a>output_arg {
+<a name="line-293"></a>  description: "Same as \"var\"."
+<a name="line-294"></a>  is_ref: true
+<a name="line-295"></a>  name: "out"
+<a name="line-296"></a>  type_attr: "T"
+<a name="line-297"></a>}
+<a name="line-298"></a>-}</span>
+<a name="line-299"></a>
+<a name="line-300"></a><a name="applyAdam"></a><span class='hs-comment'>-- | Update '*var' according to the Adam algorithm.</span>
+<a name="line-301"></a><span class='hs-comment'>--</span>
+<a name="line-302"></a><span class='hs-comment'>-- lr_t &lt;- learning_rate * sqrt(1 - beta2^t) / (1 - beta1^t)</span>
+<a name="line-303"></a><span class='hs-comment'>-- m_t &lt;- beta1 * m_{t-1} + (1 - beta1) * g_t</span>
+<a name="line-304"></a><span class='hs-comment'>-- v_t &lt;- beta2 * v_{t-1} + (1 - beta2) * g_t * g_t</span>
+<a name="line-305"></a><span class='hs-comment'>-- variable &lt;- variable - lr_t * m_t / (sqrt(v_t) + epsilon)</span>
+<a name="line-306"></a><span class='hs-definition'>applyAdam</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>v8</span> <span class='hs-varid'>v9</span> <span class='hs-varid'>v10</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-307"></a>                                                        <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-308"></a>                                                                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-309"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-310"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-311"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-312"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-313"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-314"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-315"></a>                                                                <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-316"></a>                                                                <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-317"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-318"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __m__: Should be from a Variable().</span>
+<a name="line-319"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __v__: Should be from a Variable().</span>
+<a name="line-320"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __beta1_power__: Must be a scalar.</span>
+<a name="line-321"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __beta2_power__: Must be a scalar.</span>
+<a name="line-322"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr__: Scaling factor. Must be a scalar.</span>
+<a name="line-323"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __beta1__: Momentum factor. Must be a scalar.</span>
+<a name="line-324"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v8</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __beta2__: Momentum factor. Must be a scalar.</span>
+<a name="line-325"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v9</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __epsilon__: Ridge term. Must be a scalar.</span>
+<a name="line-326"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v10</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: The gradient.</span>
+<a name="line-327"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-328"></a><span class='hs-definition'>applyAdam</span> <span class='hs-varid'>var</span> <span class='hs-varid'>m</span> <span class='hs-varid'>v</span> <span class='hs-varid'>beta1_power</span> <span class='hs-varid'>beta2_power</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>beta1</span> <span class='hs-varid'>beta2</span> <span class='hs-varid'>epsilon</span>
+<a name="line-329"></a>          <span class='hs-varid'>grad</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-330"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ApplyAdam"</span>
+<a name="line-331"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-332"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>m</span> <span class='hs-varid'>v</span> <span class='hs-varid'>beta1_power</span> <span class='hs-varid'>beta2_power</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>beta1</span> <span class='hs-varid'>beta2</span> <span class='hs-varid'>epsilon</span> <span class='hs-varid'>grad</span>
+<a name="line-333"></a><span class='hs-comment'>{-
+<a name="line-334"></a>attr {
+<a name="line-335"></a>  allowed_values {
+<a name="line-336"></a>    list {
+<a name="line-337"></a>      type: DT_FLOAT
+<a name="line-338"></a>      type: DT_DOUBLE
+<a name="line-339"></a>      type: DT_INT64
+<a name="line-340"></a>      type: DT_INT32
+<a name="line-341"></a>      type: DT_UINT8
+<a name="line-342"></a>      type: DT_UINT16
+<a name="line-343"></a>      type: DT_INT16
+<a name="line-344"></a>      type: DT_INT8
+<a name="line-345"></a>      type: DT_COMPLEX64
+<a name="line-346"></a>      type: DT_COMPLEX128
+<a name="line-347"></a>      type: DT_QINT8
+<a name="line-348"></a>      type: DT_QUINT8
+<a name="line-349"></a>      type: DT_QINT32
+<a name="line-350"></a>      type: DT_HALF
+<a name="line-351"></a>    }
+<a name="line-352"></a>  }
+<a name="line-353"></a>  name: "T"
+<a name="line-354"></a>  type: "type"
+<a name="line-355"></a>}
+<a name="line-356"></a>attr {
+<a name="line-357"></a>  default_value { b: false }
+<a name="line-358"></a>  description: "If `True`, updating of the var, m, and v tensors will be protected\nby a lock; otherwise the behavior is undefined, but may exhibit less\ncontention."
+<a name="line-359"></a>  name: "use_locking"
+<a name="line-360"></a>  type: "bool"
+<a name="line-361"></a>}
+<a name="line-362"></a>input_arg {
+<a name="line-363"></a>  description: "Should be from a Variable()."
+<a name="line-364"></a>  is_ref: true
+<a name="line-365"></a>  name: "var"
+<a name="line-366"></a>  type_attr: "T"
+<a name="line-367"></a>}
+<a name="line-368"></a>input_arg {
+<a name="line-369"></a>  description: "Should be from a Variable()."
+<a name="line-370"></a>  is_ref: true
+<a name="line-371"></a>  name: "m"
+<a name="line-372"></a>  type_attr: "T"
+<a name="line-373"></a>}
+<a name="line-374"></a>input_arg {
+<a name="line-375"></a>  description: "Should be from a Variable()."
+<a name="line-376"></a>  is_ref: true
+<a name="line-377"></a>  name: "v"
+<a name="line-378"></a>  type_attr: "T"
+<a name="line-379"></a>}
+<a name="line-380"></a>input_arg {
+<a name="line-381"></a>  description: "Must be a scalar." name: "beta1_power" type_attr: "T"
+<a name="line-382"></a>}
+<a name="line-383"></a>input_arg {
+<a name="line-384"></a>  description: "Must be a scalar." name: "beta2_power" type_attr: "T"
+<a name="line-385"></a>}
+<a name="line-386"></a>input_arg {
+<a name="line-387"></a>  description: "Scaling factor. Must be a scalar."
+<a name="line-388"></a>  name: "lr"
+<a name="line-389"></a>  type_attr: "T"
+<a name="line-390"></a>}
+<a name="line-391"></a>input_arg {
+<a name="line-392"></a>  description: "Momentum factor. Must be a scalar."
+<a name="line-393"></a>  name: "beta1"
+<a name="line-394"></a>  type_attr: "T"
+<a name="line-395"></a>}
+<a name="line-396"></a>input_arg {
+<a name="line-397"></a>  description: "Momentum factor. Must be a scalar."
+<a name="line-398"></a>  name: "beta2"
+<a name="line-399"></a>  type_attr: "T"
+<a name="line-400"></a>}
+<a name="line-401"></a>input_arg {
+<a name="line-402"></a>  description: "Ridge term. Must be a scalar."
+<a name="line-403"></a>  name: "epsilon"
+<a name="line-404"></a>  type_attr: "T"
+<a name="line-405"></a>}
+<a name="line-406"></a>input_arg {
+<a name="line-407"></a>  description: "The gradient." name: "grad" type_attr: "T"
+<a name="line-408"></a>}
+<a name="line-409"></a>output_arg {
+<a name="line-410"></a>  description: "Same as \"var\"."
+<a name="line-411"></a>  is_ref: true
+<a name="line-412"></a>  name: "out"
+<a name="line-413"></a>  type_attr: "T"
+<a name="line-414"></a>}
+<a name="line-415"></a>-}</span>
+<a name="line-416"></a>
+<a name="line-417"></a><a name="sparseApplyMomentum"></a><span class='hs-comment'>-- | Update relevant entries in '*var' and '*accum' according to the momentum scheme.</span>
+<a name="line-418"></a><span class='hs-comment'>--</span>
+<a name="line-419"></a><span class='hs-comment'>-- Set use_nesterov = True if you want to use Nesterov momentum.</span>
+<a name="line-420"></a><span class='hs-comment'>-- </span>
+<a name="line-421"></a><span class='hs-comment'>-- That is for rows we have grad for, we update var and accum as follows:</span>
+<a name="line-422"></a><span class='hs-comment'>-- </span>
+<a name="line-423"></a><span class='hs-comment'>-- accum = accum * momentum + grad</span>
+<a name="line-424"></a><span class='hs-comment'>-- var -= lr * accum</span>
+<a name="line-425"></a><span class='hs-definition'>sparseApplyMomentum</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-426"></a>                                                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-427"></a>                                                                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-428"></a>                                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-429"></a>                                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-430"></a>                                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-431"></a>                                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-432"></a>                                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-433"></a>                                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-434"></a>                                                                      <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-435"></a>                                                                      <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-436"></a>                                                              <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-437"></a>                                                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-438"></a>                                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-439"></a>                       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-440"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __accum__: Should be from a Variable().</span>
+<a name="line-441"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr__: Learning rate. Must be a scalar.</span>
+<a name="line-442"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: The gradient.</span>
+<a name="line-443"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __indices__: A vector of indices into the first dimension of var and accum.</span>
+<a name="line-444"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __momentum__: Momentum. Must be a scalar.</span>
+<a name="line-445"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-446"></a><span class='hs-definition'>sparseApplyMomentum</span> <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>momentum</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-447"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseApplyMomentum"</span>
+<a name="line-448"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-449"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-450"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>momentum</span>
+<a name="line-451"></a><span class='hs-comment'>{-
+<a name="line-452"></a>attr {
+<a name="line-453"></a>  allowed_values {
+<a name="line-454"></a>    list {
+<a name="line-455"></a>      type: DT_FLOAT
+<a name="line-456"></a>      type: DT_DOUBLE
+<a name="line-457"></a>      type: DT_INT64
+<a name="line-458"></a>      type: DT_INT32
+<a name="line-459"></a>      type: DT_UINT8
+<a name="line-460"></a>      type: DT_UINT16
+<a name="line-461"></a>      type: DT_INT16
+<a name="line-462"></a>      type: DT_INT8
+<a name="line-463"></a>      type: DT_COMPLEX64
+<a name="line-464"></a>      type: DT_COMPLEX128
+<a name="line-465"></a>      type: DT_QINT8
+<a name="line-466"></a>      type: DT_QUINT8
+<a name="line-467"></a>      type: DT_QINT32
+<a name="line-468"></a>      type: DT_HALF
+<a name="line-469"></a>    }
+<a name="line-470"></a>  }
+<a name="line-471"></a>  name: "T"
+<a name="line-472"></a>  type: "type"
+<a name="line-473"></a>}
+<a name="line-474"></a>attr {
+<a name="line-475"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-476"></a>  name: "Tindices"
+<a name="line-477"></a>  type: "type"
+<a name="line-478"></a>}
+<a name="line-479"></a>attr {
+<a name="line-480"></a>  default_value { b: false }
+<a name="line-481"></a>  description: "If `True`, updating of the var and accum tensors will be protected\nby a lock; otherwise the behavior is undefined, but may exhibit less\ncontention."
+<a name="line-482"></a>  name: "use_locking"
+<a name="line-483"></a>  type: "bool"
+<a name="line-484"></a>}
+<a name="line-485"></a>attr {
+<a name="line-486"></a>  default_value { b: false }
+<a name="line-487"></a>  description: "If `True`, the tensor passed to compute grad will be\nvar - lr * momentum * accum, so in the end, the var you get is actually\nvar - lr * momentum * accum."
+<a name="line-488"></a>  name: "use_nesterov"
+<a name="line-489"></a>  type: "bool"
+<a name="line-490"></a>}
+<a name="line-491"></a>input_arg {
+<a name="line-492"></a>  description: "Should be from a Variable()."
+<a name="line-493"></a>  is_ref: true
+<a name="line-494"></a>  name: "var"
+<a name="line-495"></a>  type_attr: "T"
+<a name="line-496"></a>}
+<a name="line-497"></a>input_arg {
+<a name="line-498"></a>  description: "Should be from a Variable()."
+<a name="line-499"></a>  is_ref: true
+<a name="line-500"></a>  name: "accum"
+<a name="line-501"></a>  type_attr: "T"
+<a name="line-502"></a>}
+<a name="line-503"></a>input_arg {
+<a name="line-504"></a>  description: "Learning rate. Must be a scalar."
+<a name="line-505"></a>  name: "lr"
+<a name="line-506"></a>  type_attr: "T"
+<a name="line-507"></a>}
+<a name="line-508"></a>input_arg {
+<a name="line-509"></a>  description: "The gradient." name: "grad" type_attr: "T"
+<a name="line-510"></a>}
+<a name="line-511"></a>input_arg {
+<a name="line-512"></a>  description: "A vector of indices into the first dimension of var and accum."
+<a name="line-513"></a>  name: "indices"
+<a name="line-514"></a>  type_attr: "Tindices"
+<a name="line-515"></a>}
+<a name="line-516"></a>input_arg {
+<a name="line-517"></a>  description: "Momentum. Must be a scalar."
+<a name="line-518"></a>  name: "momentum"
+<a name="line-519"></a>  type_attr: "T"
+<a name="line-520"></a>}
+<a name="line-521"></a>output_arg {
+<a name="line-522"></a>  description: "Same as \"var\"."
+<a name="line-523"></a>  is_ref: true
+<a name="line-524"></a>  name: "out"
+<a name="line-525"></a>  type_attr: "T"
+<a name="line-526"></a>}
+<a name="line-527"></a>-}</span>
+<a name="line-528"></a>
+<a name="line-529"></a><a name="applyMomentum"></a><span class='hs-comment'>-- | Update '*var' according to the momentum scheme. Set use_nesterov = True if you</span>
+<a name="line-530"></a><span class='hs-comment'>--</span>
+<a name="line-531"></a><span class='hs-comment'>-- want to use Nesterov momentum.</span>
+<a name="line-532"></a><span class='hs-comment'>-- </span>
+<a name="line-533"></a><span class='hs-comment'>-- accum = accum * momentum + grad</span>
+<a name="line-534"></a><span class='hs-comment'>-- var -= lr * accum</span>
+<a name="line-535"></a><span class='hs-definition'>applyMomentum</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-536"></a>                                            <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-537"></a>                                                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-538"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-539"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-540"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-541"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-542"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-543"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-544"></a>                                                    <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-545"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-546"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __accum__: Should be from a Variable().</span>
+<a name="line-547"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr__: Scaling factor. Must be a scalar.</span>
+<a name="line-548"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: The gradient.</span>
+<a name="line-549"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __momentum__: Momentum. Must be a scalar.</span>
+<a name="line-550"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-551"></a><span class='hs-definition'>applyMomentum</span> <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>momentum</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-552"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ApplyMomentum"</span>
+<a name="line-553"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-554"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>momentum</span>
+<a name="line-555"></a><span class='hs-comment'>{-
+<a name="line-556"></a>attr {
+<a name="line-557"></a>  allowed_values {
+<a name="line-558"></a>    list {
+<a name="line-559"></a>      type: DT_FLOAT
+<a name="line-560"></a>      type: DT_DOUBLE
+<a name="line-561"></a>      type: DT_INT64
+<a name="line-562"></a>      type: DT_INT32
+<a name="line-563"></a>      type: DT_UINT8
+<a name="line-564"></a>      type: DT_UINT16
+<a name="line-565"></a>      type: DT_INT16
+<a name="line-566"></a>      type: DT_INT8
+<a name="line-567"></a>      type: DT_COMPLEX64
+<a name="line-568"></a>      type: DT_COMPLEX128
+<a name="line-569"></a>      type: DT_QINT8
+<a name="line-570"></a>      type: DT_QUINT8
+<a name="line-571"></a>      type: DT_QINT32
+<a name="line-572"></a>      type: DT_HALF
+<a name="line-573"></a>    }
+<a name="line-574"></a>  }
+<a name="line-575"></a>  name: "T"
+<a name="line-576"></a>  type: "type"
+<a name="line-577"></a>}
+<a name="line-578"></a>attr {
+<a name="line-579"></a>  default_value { b: false }
+<a name="line-580"></a>  description: "If `True`, updating of the var and accum tensors will be protected\nby a lock; otherwise the behavior is undefined, but may exhibit less\ncontention."
+<a name="line-581"></a>  name: "use_locking"
+<a name="line-582"></a>  type: "bool"
+<a name="line-583"></a>}
+<a name="line-584"></a>attr {
+<a name="line-585"></a>  default_value { b: false }
+<a name="line-586"></a>  description: "If `True`, the tensor passed to compute grad will be\nvar - lr * momentum * accum, so in the end, the var you get is actually\nvar - lr * momentum * accum."
+<a name="line-587"></a>  name: "use_nesterov"
+<a name="line-588"></a>  type: "bool"
+<a name="line-589"></a>}
+<a name="line-590"></a>input_arg {
+<a name="line-591"></a>  description: "Should be from a Variable()."
+<a name="line-592"></a>  is_ref: true
+<a name="line-593"></a>  name: "var"
+<a name="line-594"></a>  type_attr: "T"
+<a name="line-595"></a>}
+<a name="line-596"></a>input_arg {
+<a name="line-597"></a>  description: "Should be from a Variable()."
+<a name="line-598"></a>  is_ref: true
+<a name="line-599"></a>  name: "accum"
+<a name="line-600"></a>  type_attr: "T"
+<a name="line-601"></a>}
+<a name="line-602"></a>input_arg {
+<a name="line-603"></a>  description: "Scaling factor. Must be a scalar."
+<a name="line-604"></a>  name: "lr"
+<a name="line-605"></a>  type_attr: "T"
+<a name="line-606"></a>}
+<a name="line-607"></a>input_arg {
+<a name="line-608"></a>  description: "The gradient." name: "grad" type_attr: "T"
+<a name="line-609"></a>}
+<a name="line-610"></a>input_arg {
+<a name="line-611"></a>  description: "Momentum. Must be a scalar."
+<a name="line-612"></a>  name: "momentum"
+<a name="line-613"></a>  type_attr: "T"
+<a name="line-614"></a>}
+<a name="line-615"></a>output_arg {
+<a name="line-616"></a>  description: "Same as \"var\"."
+<a name="line-617"></a>  is_ref: true
+<a name="line-618"></a>  name: "out"
+<a name="line-619"></a>  type_attr: "T"
+<a name="line-620"></a>}
+<a name="line-621"></a>-}</span>
+<a name="line-622"></a>
+<a name="line-623"></a><a name="applyFtrl"></a><span class='hs-comment'>-- | Update '*var' according to the Ftrl-proximal scheme.</span>
+<a name="line-624"></a><span class='hs-comment'>--</span>
+<a name="line-625"></a><span class='hs-comment'>-- accum_new = accum + grad * grad</span>
+<a name="line-626"></a><span class='hs-comment'>-- linear += grad + (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var</span>
+<a name="line-627"></a><span class='hs-comment'>-- quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2</span>
+<a name="line-628"></a><span class='hs-comment'>-- var = (sign(linear) * l1 - linear) / quadratic if |linear| &gt; l1 else 0.0</span>
+<a name="line-629"></a><span class='hs-comment'>-- accum = accum_new</span>
+<a name="line-630"></a><span class='hs-definition'>applyFtrl</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>v8</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-631"></a>                                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-632"></a>                                                         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-633"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-634"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-635"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-636"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-637"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-638"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-639"></a>                                                         <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-640"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-641"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __accum__: Should be from a Variable().</span>
+<a name="line-642"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __linear__: Should be from a Variable().</span>
+<a name="line-643"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: The gradient.</span>
+<a name="line-644"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr__: Scaling factor. Must be a scalar.</span>
+<a name="line-645"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l1__: L1 regulariation. Must be a scalar.</span>
+<a name="line-646"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l2__: L2 regulariation. Must be a scalar.</span>
+<a name="line-647"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v8</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr_power__: Scaling factor. Must be a scalar.</span>
+<a name="line-648"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-649"></a><span class='hs-definition'>applyFtrl</span> <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>linear</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>l1</span> <span class='hs-varid'>l2</span> <span class='hs-varid'>lr_power</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-650"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ApplyFtrl"</span>
+<a name="line-651"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-652"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>linear</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>l1</span> <span class='hs-varid'>l2</span> <span class='hs-varid'>lr_power</span>
+<a name="line-653"></a><span class='hs-comment'>{-
+<a name="line-654"></a>attr {
+<a name="line-655"></a>  allowed_values {
+<a name="line-656"></a>    list {
+<a name="line-657"></a>      type: DT_FLOAT
+<a name="line-658"></a>      type: DT_DOUBLE
+<a name="line-659"></a>      type: DT_INT64
+<a name="line-660"></a>      type: DT_INT32
+<a name="line-661"></a>      type: DT_UINT8
+<a name="line-662"></a>      type: DT_UINT16
+<a name="line-663"></a>      type: DT_INT16
+<a name="line-664"></a>      type: DT_INT8
+<a name="line-665"></a>      type: DT_COMPLEX64
+<a name="line-666"></a>      type: DT_COMPLEX128
+<a name="line-667"></a>      type: DT_QINT8
+<a name="line-668"></a>      type: DT_QUINT8
+<a name="line-669"></a>      type: DT_QINT32
+<a name="line-670"></a>      type: DT_HALF
+<a name="line-671"></a>    }
+<a name="line-672"></a>  }
+<a name="line-673"></a>  name: "T"
+<a name="line-674"></a>  type: "type"
+<a name="line-675"></a>}
+<a name="line-676"></a>attr {
+<a name="line-677"></a>  default_value { b: false }
+<a name="line-678"></a>  description: "If `True`, updating of the var and accum tensors will be protected\nby a lock; otherwise the behavior is undefined, but may exhibit less\ncontention."
+<a name="line-679"></a>  name: "use_locking"
+<a name="line-680"></a>  type: "bool"
+<a name="line-681"></a>}
+<a name="line-682"></a>input_arg {
+<a name="line-683"></a>  description: "Should be from a Variable()."
+<a name="line-684"></a>  is_ref: true
+<a name="line-685"></a>  name: "var"
+<a name="line-686"></a>  type_attr: "T"
+<a name="line-687"></a>}
+<a name="line-688"></a>input_arg {
+<a name="line-689"></a>  description: "Should be from a Variable()."
+<a name="line-690"></a>  is_ref: true
+<a name="line-691"></a>  name: "accum"
+<a name="line-692"></a>  type_attr: "T"
+<a name="line-693"></a>}
+<a name="line-694"></a>input_arg {
+<a name="line-695"></a>  description: "Should be from a Variable()."
+<a name="line-696"></a>  is_ref: true
+<a name="line-697"></a>  name: "linear"
+<a name="line-698"></a>  type_attr: "T"
+<a name="line-699"></a>}
+<a name="line-700"></a>input_arg {
+<a name="line-701"></a>  description: "The gradient." name: "grad" type_attr: "T"
+<a name="line-702"></a>}
+<a name="line-703"></a>input_arg {
+<a name="line-704"></a>  description: "Scaling factor. Must be a scalar."
+<a name="line-705"></a>  name: "lr"
+<a name="line-706"></a>  type_attr: "T"
+<a name="line-707"></a>}
+<a name="line-708"></a>input_arg {
+<a name="line-709"></a>  description: "L1 regulariation. Must be a scalar."
+<a name="line-710"></a>  name: "l1"
+<a name="line-711"></a>  type_attr: "T"
+<a name="line-712"></a>}
+<a name="line-713"></a>input_arg {
+<a name="line-714"></a>  description: "L2 regulariation. Must be a scalar."
+<a name="line-715"></a>  name: "l2"
+<a name="line-716"></a>  type_attr: "T"
+<a name="line-717"></a>}
+<a name="line-718"></a>input_arg {
+<a name="line-719"></a>  description: "Scaling factor. Must be a scalar."
+<a name="line-720"></a>  name: "lr_power"
+<a name="line-721"></a>  type_attr: "T"
+<a name="line-722"></a>}
+<a name="line-723"></a>output_arg {
+<a name="line-724"></a>  description: "Same as \"var\"."
+<a name="line-725"></a>  is_ref: true
+<a name="line-726"></a>  name: "out"
+<a name="line-727"></a>  type_attr: "T"
+<a name="line-728"></a>}
+<a name="line-729"></a>-}</span>
+<a name="line-730"></a>
+<a name="line-731"></a><span class='hs-comment'>-- | Update entries in '*var' and '*accum' according to the proximal adagrad scheme.</span>
+<a name="line-732"></a>
+<a name="line-733"></a><a name="sparseApplyAdagradDA"></a><span class='hs-definition'>sparseApplyAdagradDA</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>v8</span> <span class='hs-varid'>v9</span> <span class='hs-varid'>t</span>
+<a name="line-734"></a>                        <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-735"></a>                                    <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-736"></a>                                            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-737"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-738"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-739"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-740"></a>                                            <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-741"></a>                                    <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-742"></a>                                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-743"></a>                        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-744"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __gradient_accumulator__: Should be from a Variable().</span>
+<a name="line-745"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __gradient_squared_accumulator__: Should be from a Variable().</span>
+<a name="line-746"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: The gradient.</span>
+<a name="line-747"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __indices__: A vector of indices into the first dimension of var and accum.</span>
+<a name="line-748"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr__: Learning rate. Must be a scalar.</span>
+<a name="line-749"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l1__: L1 regularization. Must be a scalar.</span>
+<a name="line-750"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v8</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l2__: L2 regularization. Must be a scalar.</span>
+<a name="line-751"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v9</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __global_step__: Training step number. Must be a scalar.</span>
+<a name="line-752"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-753"></a><span class='hs-definition'>sparseApplyAdagradDA</span> <span class='hs-varid'>var</span> <span class='hs-varid'>gradient_accumulator</span> <span class='hs-varid'>gradient_squared_accumulator</span> <span class='hs-varid'>grad</span>
+<a name="line-754"></a>                     <span class='hs-varid'>indices</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>l1</span> <span class='hs-varid'>l2</span> <span class='hs-varid'>global_step</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-755"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseApplyAdagradDA"</span>
+<a name="line-756"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-757"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-758"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>gradient_accumulator</span> <span class='hs-varid'>gradient_squared_accumulator</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>l1</span>
+<a name="line-759"></a>        <span class='hs-varid'>l2</span> <span class='hs-varid'>global_step</span>
+<a name="line-760"></a><span class='hs-comment'>{-
+<a name="line-761"></a>attr {
+<a name="line-762"></a>  allowed_values {
+<a name="line-763"></a>    list {
+<a name="line-764"></a>      type: DT_FLOAT
+<a name="line-765"></a>      type: DT_DOUBLE
+<a name="line-766"></a>      type: DT_INT64
+<a name="line-767"></a>      type: DT_INT32
+<a name="line-768"></a>      type: DT_UINT8
+<a name="line-769"></a>      type: DT_UINT16
+<a name="line-770"></a>      type: DT_INT16
+<a name="line-771"></a>      type: DT_INT8
+<a name="line-772"></a>      type: DT_COMPLEX64
+<a name="line-773"></a>      type: DT_COMPLEX128
+<a name="line-774"></a>      type: DT_QINT8
+<a name="line-775"></a>      type: DT_QUINT8
+<a name="line-776"></a>      type: DT_QINT32
+<a name="line-777"></a>      type: DT_HALF
+<a name="line-778"></a>    }
+<a name="line-779"></a>  }
+<a name="line-780"></a>  name: "T"
+<a name="line-781"></a>  type: "type"
+<a name="line-782"></a>}
+<a name="line-783"></a>attr {
+<a name="line-784"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-785"></a>  name: "Tindices"
+<a name="line-786"></a>  type: "type"
+<a name="line-787"></a>}
+<a name="line-788"></a>attr {
+<a name="line-789"></a>  default_value { b: false }
+<a name="line-790"></a>  description: "If True, updating of the var and accum tensors will be protected by\na lock; otherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-791"></a>  name: "use_locking"
+<a name="line-792"></a>  type: "bool"
+<a name="line-793"></a>}
+<a name="line-794"></a>input_arg {
+<a name="line-795"></a>  description: "Should be from a Variable()."
+<a name="line-796"></a>  is_ref: true
+<a name="line-797"></a>  name: "var"
+<a name="line-798"></a>  type_attr: "T"
+<a name="line-799"></a>}
+<a name="line-800"></a>input_arg {
+<a name="line-801"></a>  description: "Should be from a Variable()."
+<a name="line-802"></a>  is_ref: true
+<a name="line-803"></a>  name: "gradient_accumulator"
+<a name="line-804"></a>  type_attr: "T"
+<a name="line-805"></a>}
+<a name="line-806"></a>input_arg {
+<a name="line-807"></a>  description: "Should be from a Variable()."
+<a name="line-808"></a>  is_ref: true
+<a name="line-809"></a>  name: "gradient_squared_accumulator"
+<a name="line-810"></a>  type_attr: "T"
+<a name="line-811"></a>}
+<a name="line-812"></a>input_arg {
+<a name="line-813"></a>  description: "The gradient." name: "grad" type_attr: "T"
+<a name="line-814"></a>}
+<a name="line-815"></a>input_arg {
+<a name="line-816"></a>  description: "A vector of indices into the first dimension of var and accum."
+<a name="line-817"></a>  name: "indices"
+<a name="line-818"></a>  type_attr: "Tindices"
+<a name="line-819"></a>}
+<a name="line-820"></a>input_arg {
+<a name="line-821"></a>  description: "Learning rate. Must be a scalar."
+<a name="line-822"></a>  name: "lr"
+<a name="line-823"></a>  type_attr: "T"
+<a name="line-824"></a>}
+<a name="line-825"></a>input_arg {
+<a name="line-826"></a>  description: "L1 regularization. Must be a scalar."
+<a name="line-827"></a>  name: "l1"
+<a name="line-828"></a>  type_attr: "T"
+<a name="line-829"></a>}
+<a name="line-830"></a>input_arg {
+<a name="line-831"></a>  description: "L2 regularization. Must be a scalar."
+<a name="line-832"></a>  name: "l2"
+<a name="line-833"></a>  type_attr: "T"
+<a name="line-834"></a>}
+<a name="line-835"></a>input_arg {
+<a name="line-836"></a>  description: "Training step number. Must be a scalar."
+<a name="line-837"></a>  name: "global_step"
+<a name="line-838"></a>  type: DT_INT64
+<a name="line-839"></a>}
+<a name="line-840"></a>output_arg {
+<a name="line-841"></a>  description: "Same as \"var\"."
+<a name="line-842"></a>  is_ref: true
+<a name="line-843"></a>  name: "out"
+<a name="line-844"></a>  type_attr: "T"
+<a name="line-845"></a>}
+<a name="line-846"></a>-}</span>
+<a name="line-847"></a>
+<a name="line-848"></a><a name="sparseApplyAdagrad"></a><span class='hs-comment'>-- | Update relevant entries in '*var' and '*accum' according to the adagrad scheme.</span>
+<a name="line-849"></a><span class='hs-comment'>--</span>
+<a name="line-850"></a><span class='hs-comment'>-- That is for rows we have grad for, we update var and accum as follows:</span>
+<a name="line-851"></a><span class='hs-comment'>-- accum += grad * grad</span>
+<a name="line-852"></a><span class='hs-comment'>-- var -= lr * grad * (1 / sqrt(accum))</span>
+<a name="line-853"></a><span class='hs-definition'>sparseApplyAdagrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-854"></a>                                                          <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-855"></a>                                                                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-856"></a>                                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-857"></a>                                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-858"></a>                                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-859"></a>                                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-860"></a>                                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-861"></a>                                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-862"></a>                                                                  <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-863"></a>                                                                  <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-864"></a>                                                          <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-865"></a>                                                          <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-866"></a>                                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-867"></a>                      <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-868"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __accum__: Should be from a Variable().</span>
+<a name="line-869"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr__: Learning rate. Must be a scalar.</span>
+<a name="line-870"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: The gradient.</span>
+<a name="line-871"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __indices__: A vector of indices into the first dimension of var and accum.</span>
+<a name="line-872"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-873"></a><span class='hs-definition'>sparseApplyAdagrad</span> <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>indices</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-874"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseApplyAdagrad"</span>
+<a name="line-875"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-876"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-877"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>indices</span>
+<a name="line-878"></a><span class='hs-comment'>{-
+<a name="line-879"></a>attr {
+<a name="line-880"></a>  allowed_values {
+<a name="line-881"></a>    list {
+<a name="line-882"></a>      type: DT_FLOAT
+<a name="line-883"></a>      type: DT_DOUBLE
+<a name="line-884"></a>      type: DT_INT64
+<a name="line-885"></a>      type: DT_INT32
+<a name="line-886"></a>      type: DT_UINT8
+<a name="line-887"></a>      type: DT_UINT16
+<a name="line-888"></a>      type: DT_INT16
+<a name="line-889"></a>      type: DT_INT8
+<a name="line-890"></a>      type: DT_COMPLEX64
+<a name="line-891"></a>      type: DT_COMPLEX128
+<a name="line-892"></a>      type: DT_QINT8
+<a name="line-893"></a>      type: DT_QUINT8
+<a name="line-894"></a>      type: DT_QINT32
+<a name="line-895"></a>      type: DT_HALF
+<a name="line-896"></a>    }
+<a name="line-897"></a>  }
+<a name="line-898"></a>  name: "T"
+<a name="line-899"></a>  type: "type"
+<a name="line-900"></a>}
+<a name="line-901"></a>attr {
+<a name="line-902"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-903"></a>  name: "Tindices"
+<a name="line-904"></a>  type: "type"
+<a name="line-905"></a>}
+<a name="line-906"></a>attr {
+<a name="line-907"></a>  default_value { b: false }
+<a name="line-908"></a>  description: "If `True`, updating of the var and accum tensors will be protected\nby a lock; otherwise the behavior is undefined, but may exhibit less\ncontention."
+<a name="line-909"></a>  name: "use_locking"
+<a name="line-910"></a>  type: "bool"
+<a name="line-911"></a>}
+<a name="line-912"></a>input_arg {
+<a name="line-913"></a>  description: "Should be from a Variable()."
+<a name="line-914"></a>  is_ref: true
+<a name="line-915"></a>  name: "var"
+<a name="line-916"></a>  type_attr: "T"
+<a name="line-917"></a>}
+<a name="line-918"></a>input_arg {
+<a name="line-919"></a>  description: "Should be from a Variable()."
+<a name="line-920"></a>  is_ref: true
+<a name="line-921"></a>  name: "accum"
+<a name="line-922"></a>  type_attr: "T"
+<a name="line-923"></a>}
+<a name="line-924"></a>input_arg {
+<a name="line-925"></a>  description: "Learning rate. Must be a scalar."
+<a name="line-926"></a>  name: "lr"
+<a name="line-927"></a>  type_attr: "T"
+<a name="line-928"></a>}
+<a name="line-929"></a>input_arg {
+<a name="line-930"></a>  description: "The gradient." name: "grad" type_attr: "T"
+<a name="line-931"></a>}
+<a name="line-932"></a>input_arg {
+<a name="line-933"></a>  description: "A vector of indices into the first dimension of var and accum."
+<a name="line-934"></a>  name: "indices"
+<a name="line-935"></a>  type_attr: "Tindices"
+<a name="line-936"></a>}
+<a name="line-937"></a>output_arg {
+<a name="line-938"></a>  description: "Same as \"var\"."
+<a name="line-939"></a>  is_ref: true
+<a name="line-940"></a>  name: "out"
+<a name="line-941"></a>  type_attr: "T"
+<a name="line-942"></a>}
+<a name="line-943"></a>-}</span>
+<a name="line-944"></a>
+<a name="line-945"></a><a name="applyProximalAdagrad"></a><span class='hs-comment'>-- | Update '*var' and '*accum' according to FOBOS with Adagrad learning rate.</span>
+<a name="line-946"></a><span class='hs-comment'>--</span>
+<a name="line-947"></a><span class='hs-comment'>-- accum += grad * grad</span>
+<a name="line-948"></a><span class='hs-comment'>-- prox_v = var - lr * grad * (1 / sqrt(accum))</span>
+<a name="line-949"></a><span class='hs-comment'>-- var = sign(prox_v)/(1+lr*l2) * max{|prox_v|-lr*l1,0}</span>
+<a name="line-950"></a><span class='hs-definition'>applyProximalAdagrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-951"></a>                                                      <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-952"></a>                                                              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-953"></a>                                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-954"></a>                                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-955"></a>                                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-956"></a>                                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-957"></a>                                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-958"></a>                                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-959"></a>                                                              <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-960"></a>                                                              <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-961"></a>                        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-962"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __accum__: Should be from a Variable().</span>
+<a name="line-963"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr__: Scaling factor. Must be a scalar.</span>
+<a name="line-964"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l1__: L1 regularization. Must be a scalar.</span>
+<a name="line-965"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l2__: L2 regularization. Must be a scalar.</span>
+<a name="line-966"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: The gradient.</span>
+<a name="line-967"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-968"></a><span class='hs-definition'>applyProximalAdagrad</span> <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>l1</span> <span class='hs-varid'>l2</span> <span class='hs-varid'>grad</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-969"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ApplyProximalAdagrad"</span>
+<a name="line-970"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-971"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>l1</span> <span class='hs-varid'>l2</span> <span class='hs-varid'>grad</span>
+<a name="line-972"></a><span class='hs-comment'>{-
+<a name="line-973"></a>attr {
+<a name="line-974"></a>  allowed_values {
+<a name="line-975"></a>    list {
+<a name="line-976"></a>      type: DT_FLOAT
+<a name="line-977"></a>      type: DT_DOUBLE
+<a name="line-978"></a>      type: DT_INT64
+<a name="line-979"></a>      type: DT_INT32
+<a name="line-980"></a>      type: DT_UINT8
+<a name="line-981"></a>      type: DT_UINT16
+<a name="line-982"></a>      type: DT_INT16
+<a name="line-983"></a>      type: DT_INT8
+<a name="line-984"></a>      type: DT_COMPLEX64
+<a name="line-985"></a>      type: DT_COMPLEX128
+<a name="line-986"></a>      type: DT_QINT8
+<a name="line-987"></a>      type: DT_QUINT8
+<a name="line-988"></a>      type: DT_QINT32
+<a name="line-989"></a>      type: DT_HALF
+<a name="line-990"></a>    }
+<a name="line-991"></a>  }
+<a name="line-992"></a>  name: "T"
+<a name="line-993"></a>  type: "type"
+<a name="line-994"></a>}
+<a name="line-995"></a>attr {
+<a name="line-996"></a>  default_value { b: false }
+<a name="line-997"></a>  description: "If True, updating of the var and accum tensors will be protected by\na lock; otherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-998"></a>  name: "use_locking"
+<a name="line-999"></a>  type: "bool"
+<a name="line-1000"></a>}
+<a name="line-1001"></a>input_arg {
+<a name="line-1002"></a>  description: "Should be from a Variable()."
+<a name="line-1003"></a>  is_ref: true
+<a name="line-1004"></a>  name: "var"
+<a name="line-1005"></a>  type_attr: "T"
+<a name="line-1006"></a>}
+<a name="line-1007"></a>input_arg {
+<a name="line-1008"></a>  description: "Should be from a Variable()."
+<a name="line-1009"></a>  is_ref: true
+<a name="line-1010"></a>  name: "accum"
+<a name="line-1011"></a>  type_attr: "T"
+<a name="line-1012"></a>}
+<a name="line-1013"></a>input_arg {
+<a name="line-1014"></a>  description: "Scaling factor. Must be a scalar."
+<a name="line-1015"></a>  name: "lr"
+<a name="line-1016"></a>  type_attr: "T"
+<a name="line-1017"></a>}
+<a name="line-1018"></a>input_arg {
+<a name="line-1019"></a>  description: "L1 regularization. Must be a scalar."
+<a name="line-1020"></a>  name: "l1"
+<a name="line-1021"></a>  type_attr: "T"
+<a name="line-1022"></a>}
+<a name="line-1023"></a>input_arg {
+<a name="line-1024"></a>  description: "L2 regularization. Must be a scalar."
+<a name="line-1025"></a>  name: "l2"
+<a name="line-1026"></a>  type_attr: "T"
+<a name="line-1027"></a>}
+<a name="line-1028"></a>input_arg {
+<a name="line-1029"></a>  description: "The gradient." name: "grad" type_attr: "T"
+<a name="line-1030"></a>}
+<a name="line-1031"></a>output_arg {
+<a name="line-1032"></a>  description: "Same as \"var\"."
+<a name="line-1033"></a>  is_ref: true
+<a name="line-1034"></a>  name: "out"
+<a name="line-1035"></a>  type_attr: "T"
+<a name="line-1036"></a>}
+<a name="line-1037"></a>-}</span>
+<a name="line-1038"></a>
+<a name="line-1039"></a><a name="applyAdagrad"></a><span class='hs-comment'>-- | Update '*var' according to the adagrad scheme.</span>
+<a name="line-1040"></a><span class='hs-comment'>--</span>
+<a name="line-1041"></a><span class='hs-comment'>-- accum += grad * grad</span>
+<a name="line-1042"></a><span class='hs-comment'>-- var -= lr * grad * (1 / sqrt(accum))</span>
+<a name="line-1043"></a><span class='hs-definition'>applyAdagrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-1044"></a>                                        <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1045"></a>                                                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1046"></a>                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-1047"></a>                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-1048"></a>                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-1049"></a>                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-1050"></a>                                                <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1051"></a>                <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-1052"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __accum__: Should be from a Variable().</span>
+<a name="line-1053"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr__: Scaling factor. Must be a scalar.</span>
+<a name="line-1054"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: The gradient.</span>
+<a name="line-1055"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-1056"></a><span class='hs-definition'>applyAdagrad</span> <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>grad</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1057"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ApplyAdagrad"</span>
+<a name="line-1058"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-1059"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>grad</span>
+<a name="line-1060"></a><span class='hs-comment'>{-
+<a name="line-1061"></a>attr {
+<a name="line-1062"></a>  allowed_values {
+<a name="line-1063"></a>    list {
+<a name="line-1064"></a>      type: DT_FLOAT
+<a name="line-1065"></a>      type: DT_DOUBLE
+<a name="line-1066"></a>      type: DT_INT64
+<a name="line-1067"></a>      type: DT_INT32
+<a name="line-1068"></a>      type: DT_UINT8
+<a name="line-1069"></a>      type: DT_UINT16
+<a name="line-1070"></a>      type: DT_INT16
+<a name="line-1071"></a>      type: DT_INT8
+<a name="line-1072"></a>      type: DT_COMPLEX64
+<a name="line-1073"></a>      type: DT_COMPLEX128
+<a name="line-1074"></a>      type: DT_QINT8
+<a name="line-1075"></a>      type: DT_QUINT8
+<a name="line-1076"></a>      type: DT_QINT32
+<a name="line-1077"></a>      type: DT_HALF
+<a name="line-1078"></a>    }
+<a name="line-1079"></a>  }
+<a name="line-1080"></a>  name: "T"
+<a name="line-1081"></a>  type: "type"
+<a name="line-1082"></a>}
+<a name="line-1083"></a>attr {
+<a name="line-1084"></a>  default_value { b: false }
+<a name="line-1085"></a>  description: "If `True`, updating of the var and accum tensors will be protected\nby a lock; otherwise the behavior is undefined, but may exhibit less\ncontention."
+<a name="line-1086"></a>  name: "use_locking"
+<a name="line-1087"></a>  type: "bool"
+<a name="line-1088"></a>}
+<a name="line-1089"></a>input_arg {
+<a name="line-1090"></a>  description: "Should be from a Variable()."
+<a name="line-1091"></a>  is_ref: true
+<a name="line-1092"></a>  name: "var"
+<a name="line-1093"></a>  type_attr: "T"
+<a name="line-1094"></a>}
+<a name="line-1095"></a>input_arg {
+<a name="line-1096"></a>  description: "Should be from a Variable()."
+<a name="line-1097"></a>  is_ref: true
+<a name="line-1098"></a>  name: "accum"
+<a name="line-1099"></a>  type_attr: "T"
+<a name="line-1100"></a>}
+<a name="line-1101"></a>input_arg {
+<a name="line-1102"></a>  description: "Scaling factor. Must be a scalar."
+<a name="line-1103"></a>  name: "lr"
+<a name="line-1104"></a>  type_attr: "T"
+<a name="line-1105"></a>}
+<a name="line-1106"></a>input_arg {
+<a name="line-1107"></a>  description: "The gradient." name: "grad" type_attr: "T"
+<a name="line-1108"></a>}
+<a name="line-1109"></a>output_arg {
+<a name="line-1110"></a>  description: "Same as \"var\"."
+<a name="line-1111"></a>  is_ref: true
+<a name="line-1112"></a>  name: "out"
+<a name="line-1113"></a>  type_attr: "T"
+<a name="line-1114"></a>}
+<a name="line-1115"></a>-}</span>
+<a name="line-1116"></a>
+<a name="line-1117"></a><a name="applyAdadelta"></a><span class='hs-comment'>-- | Update '*var' according to the adadelta scheme.</span>
+<a name="line-1118"></a><span class='hs-comment'>--</span>
+<a name="line-1119"></a><span class='hs-comment'>-- accum = rho() * accum + (1 - rho()) * grad.square();</span>
+<a name="line-1120"></a><span class='hs-comment'>-- update = (update_accum + epsilon).sqrt() * (accum + epsilon()).rsqrt() * grad;</span>
+<a name="line-1121"></a><span class='hs-comment'>-- update_accum = rho() * update_accum + (1 - rho()) * update.square();</span>
+<a name="line-1122"></a><span class='hs-comment'>-- var -= update;</span>
+<a name="line-1123"></a><span class='hs-definition'>applyAdadelta</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-1124"></a>                                                  <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1125"></a>                                                          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1126"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-1127"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-1128"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-1129"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-1130"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-1131"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-1132"></a>                                                          <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1133"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-1134"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __accum__: Should be from a Variable().</span>
+<a name="line-1135"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __accum_update__: Should be from a Variable().</span>
+<a name="line-1136"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr__: Scaling factor. Must be a scalar.</span>
+<a name="line-1137"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __rho__: Decay factor. Must be a scalar.</span>
+<a name="line-1138"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __epsilon__: Constant factor. Must be a scalar.</span>
+<a name="line-1139"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: The gradient.</span>
+<a name="line-1140"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-1141"></a><span class='hs-definition'>applyAdadelta</span> <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>accum_update</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>rho</span> <span class='hs-varid'>epsilon</span> <span class='hs-varid'>grad</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1142"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ApplyAdadelta"</span>
+<a name="line-1143"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-1144"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>accum_update</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>rho</span> <span class='hs-varid'>epsilon</span> <span class='hs-varid'>grad</span>
+<a name="line-1145"></a><span class='hs-comment'>{-
+<a name="line-1146"></a>attr {
+<a name="line-1147"></a>  allowed_values {
+<a name="line-1148"></a>    list {
+<a name="line-1149"></a>      type: DT_FLOAT
+<a name="line-1150"></a>      type: DT_DOUBLE
+<a name="line-1151"></a>      type: DT_INT64
+<a name="line-1152"></a>      type: DT_INT32
+<a name="line-1153"></a>      type: DT_UINT8
+<a name="line-1154"></a>      type: DT_UINT16
+<a name="line-1155"></a>      type: DT_INT16
+<a name="line-1156"></a>      type: DT_INT8
+<a name="line-1157"></a>      type: DT_COMPLEX64
+<a name="line-1158"></a>      type: DT_COMPLEX128
+<a name="line-1159"></a>      type: DT_QINT8
+<a name="line-1160"></a>      type: DT_QUINT8
+<a name="line-1161"></a>      type: DT_QINT32
+<a name="line-1162"></a>      type: DT_HALF
+<a name="line-1163"></a>    }
+<a name="line-1164"></a>  }
+<a name="line-1165"></a>  name: "T"
+<a name="line-1166"></a>  type: "type"
+<a name="line-1167"></a>}
+<a name="line-1168"></a>attr {
+<a name="line-1169"></a>  default_value { b: false }
+<a name="line-1170"></a>  description: "If True, updating of the var, accum and update_accum tensors will be protected by\na lock; otherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-1171"></a>  name: "use_locking"
+<a name="line-1172"></a>  type: "bool"
+<a name="line-1173"></a>}
+<a name="line-1174"></a>input_arg {
+<a name="line-1175"></a>  description: "Should be from a Variable()."
+<a name="line-1176"></a>  is_ref: true
+<a name="line-1177"></a>  name: "var"
+<a name="line-1178"></a>  type_attr: "T"
+<a name="line-1179"></a>}
+<a name="line-1180"></a>input_arg {
+<a name="line-1181"></a>  description: "Should be from a Variable()."
+<a name="line-1182"></a>  is_ref: true
+<a name="line-1183"></a>  name: "accum"
+<a name="line-1184"></a>  type_attr: "T"
+<a name="line-1185"></a>}
+<a name="line-1186"></a>input_arg {
+<a name="line-1187"></a>  description: "Should be from a Variable()."
+<a name="line-1188"></a>  is_ref: true
+<a name="line-1189"></a>  name: "accum_update"
+<a name="line-1190"></a>  type_attr: "T"
+<a name="line-1191"></a>}
+<a name="line-1192"></a>input_arg {
+<a name="line-1193"></a>  description: "Scaling factor. Must be a scalar."
+<a name="line-1194"></a>  name: "lr"
+<a name="line-1195"></a>  type_attr: "T"
+<a name="line-1196"></a>}
+<a name="line-1197"></a>input_arg {
+<a name="line-1198"></a>  description: "Decay factor. Must be a scalar."
+<a name="line-1199"></a>  name: "rho"
+<a name="line-1200"></a>  type_attr: "T"
+<a name="line-1201"></a>}
+<a name="line-1202"></a>input_arg {
+<a name="line-1203"></a>  description: "Constant factor. Must be a scalar."
+<a name="line-1204"></a>  name: "epsilon"
+<a name="line-1205"></a>  type_attr: "T"
+<a name="line-1206"></a>}
+<a name="line-1207"></a>input_arg {
+<a name="line-1208"></a>  description: "The gradient." name: "grad" type_attr: "T"
+<a name="line-1209"></a>}
+<a name="line-1210"></a>output_arg {
+<a name="line-1211"></a>  description: "Same as \"var\"."
+<a name="line-1212"></a>  is_ref: true
+<a name="line-1213"></a>  name: "out"
+<a name="line-1214"></a>  type_attr: "T"
+<a name="line-1215"></a>}
+<a name="line-1216"></a>-}</span>
+<a name="line-1217"></a>
+<a name="line-1218"></a><a name="sparseApplyProximalGradientDescent"></a><span class='hs-comment'>-- | Sparse update '*var' as FOBOS algorithm with fixed learning rate.</span>
+<a name="line-1219"></a><span class='hs-comment'>--</span>
+<a name="line-1220"></a><span class='hs-comment'>-- That is for rows we have grad for, we update var as follows:</span>
+<a name="line-1221"></a><span class='hs-comment'>-- prox_v = var - alpha * grad</span>
+<a name="line-1222"></a><span class='hs-comment'>-- var = sign(prox_v)/(1+alpha*l2) * max{|prox_v|-alpha*l1,0}</span>
+<a name="line-1223"></a><span class='hs-definition'>sparseApplyProximalGradientDescent</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span>
+<a name="line-1224"></a>                                      <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-1225"></a>                                                  <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1226"></a>                                                          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1227"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-1228"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-1229"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-1230"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-1231"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-1232"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-1233"></a>                                                          <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-1234"></a>                                                  <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-1235"></a>                                                  <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-1236"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1237"></a>                                      <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-1238"></a>                                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __alpha__: Scaling factor. Must be a scalar.</span>
+<a name="line-1239"></a>                                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l1__: L1 regularization. Must be a scalar.</span>
+<a name="line-1240"></a>                                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l2__: L2 regularization. Must be a scalar.</span>
+<a name="line-1241"></a>                                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: The gradient.</span>
+<a name="line-1242"></a>                                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __indices__: A vector of indices into the first dimension of var and accum.</span>
+<a name="line-1243"></a>                                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-1244"></a><span class='hs-definition'>sparseApplyProximalGradientDescent</span> <span class='hs-varid'>var</span> <span class='hs-varid'>alpha</span> <span class='hs-varid'>l1</span> <span class='hs-varid'>l2</span> <span class='hs-varid'>grad</span>
+<a name="line-1245"></a>                                   <span class='hs-varid'>indices</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1246"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseApplyProximalGradientDescent"</span>
+<a name="line-1247"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-1248"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-1249"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>alpha</span> <span class='hs-varid'>l1</span> <span class='hs-varid'>l2</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>indices</span>
+<a name="line-1250"></a><span class='hs-comment'>{-
+<a name="line-1251"></a>attr {
+<a name="line-1252"></a>  allowed_values {
+<a name="line-1253"></a>    list {
+<a name="line-1254"></a>      type: DT_FLOAT
+<a name="line-1255"></a>      type: DT_DOUBLE
+<a name="line-1256"></a>      type: DT_INT64
+<a name="line-1257"></a>      type: DT_INT32
+<a name="line-1258"></a>      type: DT_UINT8
+<a name="line-1259"></a>      type: DT_UINT16
+<a name="line-1260"></a>      type: DT_INT16
+<a name="line-1261"></a>      type: DT_INT8
+<a name="line-1262"></a>      type: DT_COMPLEX64
+<a name="line-1263"></a>      type: DT_COMPLEX128
+<a name="line-1264"></a>      type: DT_QINT8
+<a name="line-1265"></a>      type: DT_QUINT8
+<a name="line-1266"></a>      type: DT_QINT32
+<a name="line-1267"></a>      type: DT_HALF
+<a name="line-1268"></a>    }
+<a name="line-1269"></a>  }
+<a name="line-1270"></a>  name: "T"
+<a name="line-1271"></a>  type: "type"
+<a name="line-1272"></a>}
+<a name="line-1273"></a>attr {
+<a name="line-1274"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-1275"></a>  name: "Tindices"
+<a name="line-1276"></a>  type: "type"
+<a name="line-1277"></a>}
+<a name="line-1278"></a>attr {
+<a name="line-1279"></a>  default_value { b: false }
+<a name="line-1280"></a>  description: "If True, the subtraction will be protected by a lock;\notherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-1281"></a>  name: "use_locking"
+<a name="line-1282"></a>  type: "bool"
+<a name="line-1283"></a>}
+<a name="line-1284"></a>input_arg {
+<a name="line-1285"></a>  description: "Should be from a Variable()."
+<a name="line-1286"></a>  is_ref: true
+<a name="line-1287"></a>  name: "var"
+<a name="line-1288"></a>  type_attr: "T"
+<a name="line-1289"></a>}
+<a name="line-1290"></a>input_arg {
+<a name="line-1291"></a>  description: "Scaling factor. Must be a scalar."
+<a name="line-1292"></a>  name: "alpha"
+<a name="line-1293"></a>  type_attr: "T"
+<a name="line-1294"></a>}
+<a name="line-1295"></a>input_arg {
+<a name="line-1296"></a>  description: "L1 regularization. Must be a scalar."
+<a name="line-1297"></a>  name: "l1"
+<a name="line-1298"></a>  type_attr: "T"
+<a name="line-1299"></a>}
+<a name="line-1300"></a>input_arg {
+<a name="line-1301"></a>  description: "L2 regularization. Must be a scalar."
+<a name="line-1302"></a>  name: "l2"
+<a name="line-1303"></a>  type_attr: "T"
+<a name="line-1304"></a>}
+<a name="line-1305"></a>input_arg {
+<a name="line-1306"></a>  description: "The gradient." name: "grad" type_attr: "T"
+<a name="line-1307"></a>}
+<a name="line-1308"></a>input_arg {
+<a name="line-1309"></a>  description: "A vector of indices into the first dimension of var and accum."
+<a name="line-1310"></a>  name: "indices"
+<a name="line-1311"></a>  type_attr: "Tindices"
+<a name="line-1312"></a>}
+<a name="line-1313"></a>output_arg {
+<a name="line-1314"></a>  description: "Same as \"var\"."
+<a name="line-1315"></a>  is_ref: true
+<a name="line-1316"></a>  name: "out"
+<a name="line-1317"></a>  type_attr: "T"
+<a name="line-1318"></a>}
+<a name="line-1319"></a>-}</span>
+<a name="line-1320"></a>
+<a name="line-1321"></a><a name="applyProximalGradientDescent"></a><span class='hs-comment'>-- | Update '*var' as FOBOS algorithm with fixed learning rate.</span>
+<a name="line-1322"></a><span class='hs-comment'>--</span>
+<a name="line-1323"></a><span class='hs-comment'>-- prox_v = var - alpha * delta</span>
+<a name="line-1324"></a><span class='hs-comment'>-- var = sign(prox_v)/(1+alpha*l2) * max{|prox_v|-alpha*l1,0}</span>
+<a name="line-1325"></a><span class='hs-definition'>applyProximalGradientDescent</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-1326"></a>                                                           <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1327"></a>                                                                   <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1328"></a>                                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-1329"></a>                                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-1330"></a>                                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-1331"></a>                                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-1332"></a>                                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-1333"></a>                                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-1334"></a>                                                                   <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-1335"></a>                                                                   <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1336"></a>                                <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-1337"></a>                                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __alpha__: Scaling factor. Must be a scalar.</span>
+<a name="line-1338"></a>                                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l1__: L1 regularization. Must be a scalar.</span>
+<a name="line-1339"></a>                                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l2__: L2 regularization. Must be a scalar.</span>
+<a name="line-1340"></a>                                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __delta__: The change.</span>
+<a name="line-1341"></a>                                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-1342"></a><span class='hs-definition'>applyProximalGradientDescent</span> <span class='hs-varid'>var</span> <span class='hs-varid'>alpha</span> <span class='hs-varid'>l1</span> <span class='hs-varid'>l2</span> <span class='hs-varid'>delta</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1343"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ApplyProximalGradientDescent"</span>
+<a name="line-1344"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-1345"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>alpha</span> <span class='hs-varid'>l1</span> <span class='hs-varid'>l2</span> <span class='hs-varid'>delta</span>
+<a name="line-1346"></a><span class='hs-comment'>{-
+<a name="line-1347"></a>attr {
+<a name="line-1348"></a>  allowed_values {
+<a name="line-1349"></a>    list {
+<a name="line-1350"></a>      type: DT_FLOAT
+<a name="line-1351"></a>      type: DT_DOUBLE
+<a name="line-1352"></a>      type: DT_INT64
+<a name="line-1353"></a>      type: DT_INT32
+<a name="line-1354"></a>      type: DT_UINT8
+<a name="line-1355"></a>      type: DT_UINT16
+<a name="line-1356"></a>      type: DT_INT16
+<a name="line-1357"></a>      type: DT_INT8
+<a name="line-1358"></a>      type: DT_COMPLEX64
+<a name="line-1359"></a>      type: DT_COMPLEX128
+<a name="line-1360"></a>      type: DT_QINT8
+<a name="line-1361"></a>      type: DT_QUINT8
+<a name="line-1362"></a>      type: DT_QINT32
+<a name="line-1363"></a>      type: DT_HALF
+<a name="line-1364"></a>    }
+<a name="line-1365"></a>  }
+<a name="line-1366"></a>  name: "T"
+<a name="line-1367"></a>  type: "type"
+<a name="line-1368"></a>}
+<a name="line-1369"></a>attr {
+<a name="line-1370"></a>  default_value { b: false }
+<a name="line-1371"></a>  description: "If True, the subtraction will be protected by a lock;\notherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-1372"></a>  name: "use_locking"
+<a name="line-1373"></a>  type: "bool"
+<a name="line-1374"></a>}
+<a name="line-1375"></a>input_arg {
+<a name="line-1376"></a>  description: "Should be from a Variable()."
+<a name="line-1377"></a>  is_ref: true
+<a name="line-1378"></a>  name: "var"
+<a name="line-1379"></a>  type_attr: "T"
+<a name="line-1380"></a>}
+<a name="line-1381"></a>input_arg {
+<a name="line-1382"></a>  description: "Scaling factor. Must be a scalar."
+<a name="line-1383"></a>  name: "alpha"
+<a name="line-1384"></a>  type_attr: "T"
+<a name="line-1385"></a>}
+<a name="line-1386"></a>input_arg {
+<a name="line-1387"></a>  description: "L1 regularization. Must be a scalar."
+<a name="line-1388"></a>  name: "l1"
+<a name="line-1389"></a>  type_attr: "T"
+<a name="line-1390"></a>}
+<a name="line-1391"></a>input_arg {
+<a name="line-1392"></a>  description: "L2 regularization. Must be a scalar."
+<a name="line-1393"></a>  name: "l2"
+<a name="line-1394"></a>  type_attr: "T"
+<a name="line-1395"></a>}
+<a name="line-1396"></a>input_arg {
+<a name="line-1397"></a>  description: "The change." name: "delta" type_attr: "T"
+<a name="line-1398"></a>}
+<a name="line-1399"></a>output_arg {
+<a name="line-1400"></a>  description: "Same as \"var\"."
+<a name="line-1401"></a>  is_ref: true
+<a name="line-1402"></a>  name: "out"
+<a name="line-1403"></a>  type_attr: "T"
+<a name="line-1404"></a>}
+<a name="line-1405"></a>-}</span>
+<a name="line-1406"></a>
+<a name="line-1407"></a><a name="encodeBase64"></a><span class='hs-comment'>-- | Encode strings into web-safe base64 format.</span>
+<a name="line-1408"></a><span class='hs-comment'>--</span>
+<a name="line-1409"></a><span class='hs-comment'>-- Refer to the following article for more information on base64 format:</span>
+<a name="line-1410"></a><span class='hs-comment'>-- en.wikipedia.org/wiki/Base64. Base64 strings may have padding with '=' at the</span>
+<a name="line-1411"></a><span class='hs-comment'>-- end so that the encoded has length multiple of 4. See Padding section of the</span>
+<a name="line-1412"></a><span class='hs-comment'>-- link above.</span>
+<a name="line-1413"></a><span class='hs-comment'>-- </span>
+<a name="line-1414"></a><span class='hs-comment'>-- Web-safe means that the encoder uses - and _ instead of + and /.</span>
+<a name="line-1415"></a><span class='hs-definition'>encodeBase64</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __input__: Strings to be encoded.</span>
+<a name="line-1416"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __output__: Input strings encoded in base64.</span>
+<a name="line-1417"></a><span class='hs-definition'>encodeBase64</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1418"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"EncodeBase64"</span><span class='hs-layout'>)</span>
+<a name="line-1419"></a>        <span class='hs-varid'>input</span>
+<a name="line-1420"></a><span class='hs-comment'>{-
+<a name="line-1421"></a>attr {
+<a name="line-1422"></a>  default_value { b: false }
+<a name="line-1423"></a>  description: "Bool whether padding is applied at the ends."
+<a name="line-1424"></a>  name: "pad"
+<a name="line-1425"></a>  type: "bool"
+<a name="line-1426"></a>}
+<a name="line-1427"></a>input_arg {
+<a name="line-1428"></a>  description: "Strings to be encoded." name: "input" type: DT_STRING
+<a name="line-1429"></a>}
+<a name="line-1430"></a>output_arg {
+<a name="line-1431"></a>  description: "Input strings encoded in base64."
+<a name="line-1432"></a>  name: "output"
+<a name="line-1433"></a>  type: DT_STRING
+<a name="line-1434"></a>}
+<a name="line-1435"></a>-}</span>
+<a name="line-1436"></a>
+<a name="line-1437"></a><a name="stringSplit"></a><span class='hs-comment'>-- | Split elements of `input` based on `delimiter` into a `SparseTensor`.</span>
+<a name="line-1438"></a><span class='hs-comment'>--</span>
+<a name="line-1439"></a><span class='hs-comment'>-- Let N be the size of source (typically N will be the batch size). Split each</span>
+<a name="line-1440"></a><span class='hs-comment'>-- element of `input` based on `delimiter` and return a `SparseTensor`</span>
+<a name="line-1441"></a><span class='hs-comment'>-- containing the splitted tokens. Empty tokens are ignored.</span>
+<a name="line-1442"></a><span class='hs-comment'>-- </span>
+<a name="line-1443"></a><span class='hs-comment'>-- `delimiter` can be empty or a single character. If `delimiter` is an empty</span>
+<a name="line-1444"></a><span class='hs-comment'>--  string, each element of `input` is split into individual 1 character strings.</span>
+<a name="line-1445"></a><span class='hs-comment'>-- </span>
+<a name="line-1446"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-1447"></a><span class='hs-comment'>--   N = 2, input[0] is 'hello world' and input[1] is 'a b c', then the output</span>
+<a name="line-1448"></a><span class='hs-comment'>--   will be</span>
+<a name="line-1449"></a><span class='hs-comment'>-- </span>
+<a name="line-1450"></a><span class='hs-comment'>--   indices = [0, 0;</span>
+<a name="line-1451"></a><span class='hs-comment'>--              0, 1;</span>
+<a name="line-1452"></a><span class='hs-comment'>--              1, 0;</span>
+<a name="line-1453"></a><span class='hs-comment'>--              1, 1;</span>
+<a name="line-1454"></a><span class='hs-comment'>--              1, 2]</span>
+<a name="line-1455"></a><span class='hs-comment'>--   shape = [2, 3]</span>
+<a name="line-1456"></a><span class='hs-comment'>--   values = ['hello', 'world', 'a', 'b', 'c']</span>
+<a name="line-1457"></a><span class='hs-definition'>stringSplit</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __input__: 1-D. Strings to split.</span>
+<a name="line-1458"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __delimiter__: 0-D. Delimiter character, or empty string.</span>
+<a name="line-1459"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-1460"></a>                   <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>,</span>
+<a name="line-1461"></a>                   <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-1462"></a>               <span class='hs-comment'>-- ^ (__indices__, __values__, __shape__)</span>
+<a name="line-1463"></a>               <span class='hs-comment'>--</span>
+<a name="line-1464"></a>               <span class='hs-comment'>-- * __indices__: A dense matrix of int64 representing the indices of the sparse tensor.</span>
+<a name="line-1465"></a>               <span class='hs-comment'>--</span>
+<a name="line-1466"></a>               <span class='hs-comment'>-- * __values__: A vector of strings corresponding to the splited values.</span>
+<a name="line-1467"></a>               <span class='hs-comment'>--</span>
+<a name="line-1468"></a>               <span class='hs-comment'>-- * __shape__: a length-2 vector of int64 representing the shape of the sparse</span>
+<a name="line-1469"></a>               <span class='hs-comment'>-- tensor, where the first value is N and the second value is the maximum number</span>
+<a name="line-1470"></a>               <span class='hs-comment'>-- of tokens in a single input entry.</span>
+<a name="line-1471"></a><span class='hs-definition'>stringSplit</span> <span class='hs-varid'>input</span> <span class='hs-varid'>delimiter</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1472"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"StringSplit"</span><span class='hs-layout'>)</span>
+<a name="line-1473"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>delimiter</span>
+<a name="line-1474"></a><span class='hs-comment'>{-
+<a name="line-1475"></a>input_arg {
+<a name="line-1476"></a>  description: "1-D. Strings to split." name: "input" type: DT_STRING
+<a name="line-1477"></a>}
+<a name="line-1478"></a>input_arg {
+<a name="line-1479"></a>  description: "0-D. Delimiter character, or empty string."
+<a name="line-1480"></a>  name: "delimiter"
+<a name="line-1481"></a>  type: DT_STRING
+<a name="line-1482"></a>}
+<a name="line-1483"></a>output_arg {
+<a name="line-1484"></a>  description: "A dense matrix of int64 representing the indices of the sparse tensor."
+<a name="line-1485"></a>  name: "indices"
+<a name="line-1486"></a>  type: DT_INT64
+<a name="line-1487"></a>}
+<a name="line-1488"></a>output_arg {
+<a name="line-1489"></a>  description: "A vector of strings corresponding to the splited values."
+<a name="line-1490"></a>  name: "values"
+<a name="line-1491"></a>  type: DT_STRING
+<a name="line-1492"></a>}
+<a name="line-1493"></a>output_arg {
+<a name="line-1494"></a>  description: "a length-2 vector of int64 representing the shape of the sparse\ntensor, where the first value is N and the second value is the maximum number\nof tokens in a single input entry."
+<a name="line-1495"></a>  name: "shape"
+<a name="line-1496"></a>  type: DT_INT64
+<a name="line-1497"></a>}
+<a name="line-1498"></a>-}</span>
+<a name="line-1499"></a>
+<a name="line-1500"></a><a name="stringJoin"></a><span class='hs-comment'>-- | Joins the strings in the given list of string tensors into one tensor;</span>
+<a name="line-1501"></a><span class='hs-comment'>--</span>
+<a name="line-1502"></a><span class='hs-comment'>-- with the given separator (default is an empty separator).</span>
+<a name="line-1503"></a><span class='hs-definition'>stringJoin</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __inputs__: A list of string tensors.  The tensors must all have the same shape,</span>
+<a name="line-1504"></a>                                                     <span class='hs-comment'>-- or be scalars.  Scalars may be mixed in; these will be broadcast to the shape</span>
+<a name="line-1505"></a>                                                     <span class='hs-comment'>-- of non-scalar inputs.</span>
+<a name="line-1506"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-1507"></a><span class='hs-definition'>stringJoin</span> <span class='hs-varid'>inputs</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"N"</span><span class='hs-layout'>,</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"inputs"</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>inputs</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1508"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"StringJoin"</span>
+<a name="line-1509"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"N"</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>inputs</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-1510"></a>        <span class='hs-varid'>inputs</span>
+<a name="line-1511"></a><span class='hs-comment'>{-
+<a name="line-1512"></a>attr { has_minimum: true minimum: 1 name: "N" type: "int" }
+<a name="line-1513"></a>attr {
+<a name="line-1514"></a>  default_value { s: "" }
+<a name="line-1515"></a>  description: "string, an optional join separator."
+<a name="line-1516"></a>  name: "separator"
+<a name="line-1517"></a>  type: "string"
+<a name="line-1518"></a>}
+<a name="line-1519"></a>input_arg {
+<a name="line-1520"></a>  description: "A list of string tensors.  The tensors must all have the same shape,\nor be scalars.  Scalars may be mixed in; these will be broadcast to the shape\nof non-scalar inputs."
+<a name="line-1521"></a>  name: "inputs"
+<a name="line-1522"></a>  number_attr: "N"
+<a name="line-1523"></a>  type: DT_STRING
+<a name="line-1524"></a>}
+<a name="line-1525"></a>output_arg { name: "output" type: DT_STRING }
+<a name="line-1526"></a>-}</span>
+<a name="line-1527"></a>
+<a name="line-1528"></a><a name="asString"></a><span class='hs-comment'>-- | Converts each entry in the given tensor to strings.  Supports many numeric</span>
+<a name="line-1529"></a><span class='hs-comment'>--</span>
+<a name="line-1530"></a><span class='hs-comment'>-- types and boolean.</span>
+<a name="line-1531"></a><span class='hs-definition'>asString</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1532"></a>                                                 <span class='hs-conid'>Bool</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-1533"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-1534"></a>                                                 <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1535"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-1536"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-1537"></a><span class='hs-definition'>asString</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1538"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"AsString"</span>
+<a name="line-1539"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-1540"></a>        <span class='hs-varid'>input</span>
+<a name="line-1541"></a><span class='hs-comment'>{-
+<a name="line-1542"></a>attr {
+<a name="line-1543"></a>  allowed_values {
+<a name="line-1544"></a>    list {
+<a name="line-1545"></a>      type: DT_INT32
+<a name="line-1546"></a>      type: DT_INT64
+<a name="line-1547"></a>      type: DT_COMPLEX64
+<a name="line-1548"></a>      type: DT_FLOAT
+<a name="line-1549"></a>      type: DT_DOUBLE
+<a name="line-1550"></a>      type: DT_BOOL
+<a name="line-1551"></a>      type: DT_INT8
+<a name="line-1552"></a>    }
+<a name="line-1553"></a>  }
+<a name="line-1554"></a>  name: "T"
+<a name="line-1555"></a>  type: "type"
+<a name="line-1556"></a>}
+<a name="line-1557"></a>attr {
+<a name="line-1558"></a>  default_value { i: -1 }
+<a name="line-1559"></a>  description: "The post-decimal precision to use for floating point numbers.\nOnly used if precision &gt; -1."
+<a name="line-1560"></a>  name: "precision"
+<a name="line-1561"></a>  type: "int"
+<a name="line-1562"></a>}
+<a name="line-1563"></a>attr {
+<a name="line-1564"></a>  default_value { b: false }
+<a name="line-1565"></a>  description: "Use scientific notation for floating point numbers."
+<a name="line-1566"></a>  name: "scientific"
+<a name="line-1567"></a>  type: "bool"
+<a name="line-1568"></a>}
+<a name="line-1569"></a>attr {
+<a name="line-1570"></a>  default_value { b: false }
+<a name="line-1571"></a>  description: "Use shortest representation (either scientific or standard) for\nfloating point numbers."
+<a name="line-1572"></a>  name: "shortest"
+<a name="line-1573"></a>  type: "bool"
+<a name="line-1574"></a>}
+<a name="line-1575"></a>attr {
+<a name="line-1576"></a>  default_value { i: -1 }
+<a name="line-1577"></a>  description: "Pad pre-decimal numbers to this width.\nApplies to both floating point and integer numbers.\nOnly used if width &gt; -1."
+<a name="line-1578"></a>  name: "width"
+<a name="line-1579"></a>  type: "int"
+<a name="line-1580"></a>}
+<a name="line-1581"></a>attr {
+<a name="line-1582"></a>  default_value { s: "" }
+<a name="line-1583"></a>  description: "The value to pad if width &gt; -1.  If empty, pads with spaces.\nAnother typical value is \'0\'.  String cannot be longer than 1 character."
+<a name="line-1584"></a>  name: "fill"
+<a name="line-1585"></a>  type: "string"
+<a name="line-1586"></a>}
+<a name="line-1587"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-1588"></a>output_arg { name: "output" type: DT_STRING }
+<a name="line-1589"></a>-}</span>
+<a name="line-1590"></a>
+<a name="line-1591"></a><a name="stringToHashBucketStrong"></a><span class='hs-comment'>-- | Converts each string in the input Tensor to its hash mod by a number of buckets.</span>
+<a name="line-1592"></a><span class='hs-comment'>--</span>
+<a name="line-1593"></a><span class='hs-comment'>-- The hash function is deterministic on the content of the string within the</span>
+<a name="line-1594"></a><span class='hs-comment'>-- process. The hash function is a keyed hash function, where attribute `key`</span>
+<a name="line-1595"></a><span class='hs-comment'>-- defines the key of the hash function. `key` is an array of 2 elements.</span>
+<a name="line-1596"></a><span class='hs-comment'>-- </span>
+<a name="line-1597"></a><span class='hs-comment'>-- A strong hash is important when inputs may be malicious, e.g. URLs with</span>
+<a name="line-1598"></a><span class='hs-comment'>-- additional components. Adversaries could try to make their inputs hash to the</span>
+<a name="line-1599"></a><span class='hs-comment'>-- same bucket for a denial-of-service attack or to skew the results. A strong</span>
+<a name="line-1600"></a><span class='hs-comment'>-- hash prevents this by making it dificult, if not infeasible, to compute inputs</span>
+<a name="line-1601"></a><span class='hs-comment'>-- that hash to the same bucket. This comes at a cost of roughly 4x higher compute</span>
+<a name="line-1602"></a><span class='hs-comment'>-- time than tf.string_to_hash_bucket_fast.</span>
+<a name="line-1603"></a><span class='hs-definition'>stringToHashBucketStrong</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_buckets__: The number of buckets.</span>
+<a name="line-1604"></a>                            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __input__: The strings to assign a hash bucket.</span>
+<a name="line-1605"></a>                            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __output__: A Tensor of the same shape as the input `string_tensor`.</span>
+<a name="line-1606"></a><span class='hs-definition'>stringToHashBucketStrong</span> <span class='hs-varid'>num_buckets</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1607"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"StringToHashBucketStrong"</span>
+<a name="line-1608"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_buckets"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_buckets</span><span class='hs-layout'>)</span>
+<a name="line-1609"></a>        <span class='hs-varid'>input</span>
+<a name="line-1610"></a><span class='hs-comment'>{-
+<a name="line-1611"></a>attr {
+<a name="line-1612"></a>  description: "The number of buckets."
+<a name="line-1613"></a>  has_minimum: true
+<a name="line-1614"></a>  minimum: 1
+<a name="line-1615"></a>  name: "num_buckets"
+<a name="line-1616"></a>  type: "int"
+<a name="line-1617"></a>}
+<a name="line-1618"></a>attr {
+<a name="line-1619"></a>  description: "The key for the keyed hash function passed as a list of two uint64\nelements."
+<a name="line-1620"></a>  name: "key"
+<a name="line-1621"></a>  type: "list(int)"
+<a name="line-1622"></a>}
+<a name="line-1623"></a>input_arg {
+<a name="line-1624"></a>  description: "The strings to assign a hash bucket."
+<a name="line-1625"></a>  name: "input"
+<a name="line-1626"></a>  type: DT_STRING
+<a name="line-1627"></a>}
+<a name="line-1628"></a>output_arg {
+<a name="line-1629"></a>  description: "A Tensor of the same shape as the input `string_tensor`."
+<a name="line-1630"></a>  name: "output"
+<a name="line-1631"></a>  type: DT_INT64
+<a name="line-1632"></a>}
+<a name="line-1633"></a>-}</span>
+<a name="line-1634"></a>
+<a name="line-1635"></a><a name="scatterMul"></a><span class='hs-comment'>-- | Multiplies sparse updates into a variable reference.</span>
+<a name="line-1636"></a><span class='hs-comment'>--</span>
+<a name="line-1637"></a><span class='hs-comment'>-- This operation computes</span>
+<a name="line-1638"></a><span class='hs-comment'>-- </span>
+<a name="line-1639"></a><span class='hs-comment'>--     # Scalar indices</span>
+<a name="line-1640"></a><span class='hs-comment'>--     ref[indices, ...] *= updates[...]</span>
+<a name="line-1641"></a><span class='hs-comment'>-- </span>
+<a name="line-1642"></a><span class='hs-comment'>--     # Vector indices (for each i)</span>
+<a name="line-1643"></a><span class='hs-comment'>--     ref[indices[i], ...] *= updates[i, ...]</span>
+<a name="line-1644"></a><span class='hs-comment'>-- </span>
+<a name="line-1645"></a><span class='hs-comment'>--     # High rank indices (for each i, ..., j)</span>
+<a name="line-1646"></a><span class='hs-comment'>--     ref[indices[i, ..., j], ...] *= updates[i, ..., j, ...]</span>
+<a name="line-1647"></a><span class='hs-comment'>-- </span>
+<a name="line-1648"></a><span class='hs-comment'>-- This operation outputs `ref` after the update is done.</span>
+<a name="line-1649"></a><span class='hs-comment'>-- This makes it easier to chain operations that need to use the reset value.</span>
+<a name="line-1650"></a><span class='hs-comment'>-- </span>
+<a name="line-1651"></a><span class='hs-comment'>-- Duplicate entries are handled correctly: if multiple `indices` reference</span>
+<a name="line-1652"></a><span class='hs-comment'>-- the same location, their contributions multiply.</span>
+<a name="line-1653"></a><span class='hs-comment'>-- </span>
+<a name="line-1654"></a><span class='hs-comment'>-- Requires `updates.shape = indices.shape + ref.shape[1:]`.</span>
+<a name="line-1655"></a><span class='hs-definition'>scatterMul</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-1656"></a>                                            <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1657"></a>                                                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1658"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-1659"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-1660"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-1661"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-1662"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-1663"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-1664"></a>                                                    <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-1665"></a>                                            <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-1666"></a>                                            <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-1667"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1668"></a>              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __ref__: Should be from a `Variable` node.</span>
+<a name="line-1669"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __indices__: A tensor of indices into the first dimension of `ref`.</span>
+<a name="line-1670"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __updates__: A tensor of updated values to multiply to `ref`.</span>
+<a name="line-1671"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output_ref__: = Same as `ref`.  Returned as a convenience for operations that want</span>
+<a name="line-1672"></a>              <span class='hs-comment'>-- to use the updated values after the update is done.</span>
+<a name="line-1673"></a><span class='hs-definition'>scatterMul</span> <span class='hs-varid'>ref</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>updates</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1674"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ScatterMul"</span>
+<a name="line-1675"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-1676"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-1677"></a>        <span class='hs-varid'>ref</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>updates</span>
+<a name="line-1678"></a><span class='hs-comment'>{-
+<a name="line-1679"></a>attr {
+<a name="line-1680"></a>  allowed_values {
+<a name="line-1681"></a>    list {
+<a name="line-1682"></a>      type: DT_FLOAT
+<a name="line-1683"></a>      type: DT_DOUBLE
+<a name="line-1684"></a>      type: DT_INT64
+<a name="line-1685"></a>      type: DT_INT32
+<a name="line-1686"></a>      type: DT_UINT8
+<a name="line-1687"></a>      type: DT_UINT16
+<a name="line-1688"></a>      type: DT_INT16
+<a name="line-1689"></a>      type: DT_INT8
+<a name="line-1690"></a>      type: DT_COMPLEX64
+<a name="line-1691"></a>      type: DT_COMPLEX128
+<a name="line-1692"></a>      type: DT_QINT8
+<a name="line-1693"></a>      type: DT_QUINT8
+<a name="line-1694"></a>      type: DT_QINT32
+<a name="line-1695"></a>      type: DT_HALF
+<a name="line-1696"></a>    }
+<a name="line-1697"></a>  }
+<a name="line-1698"></a>  name: "T"
+<a name="line-1699"></a>  type: "type"
+<a name="line-1700"></a>}
+<a name="line-1701"></a>attr {
+<a name="line-1702"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-1703"></a>  name: "Tindices"
+<a name="line-1704"></a>  type: "type"
+<a name="line-1705"></a>}
+<a name="line-1706"></a>attr {
+<a name="line-1707"></a>  default_value { b: false }
+<a name="line-1708"></a>  description: "If True, the operation will be protected by a lock;\notherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-1709"></a>  name: "use_locking"
+<a name="line-1710"></a>  type: "bool"
+<a name="line-1711"></a>}
+<a name="line-1712"></a>input_arg {
+<a name="line-1713"></a>  description: "Should be from a `Variable` node."
+<a name="line-1714"></a>  is_ref: true
+<a name="line-1715"></a>  name: "ref"
+<a name="line-1716"></a>  type_attr: "T"
+<a name="line-1717"></a>}
+<a name="line-1718"></a>input_arg {
+<a name="line-1719"></a>  description: "A tensor of indices into the first dimension of `ref`."
+<a name="line-1720"></a>  name: "indices"
+<a name="line-1721"></a>  type_attr: "Tindices"
+<a name="line-1722"></a>}
+<a name="line-1723"></a>input_arg {
+<a name="line-1724"></a>  description: "A tensor of updated values to multiply to `ref`."
+<a name="line-1725"></a>  name: "updates"
+<a name="line-1726"></a>  type_attr: "T"
+<a name="line-1727"></a>}
+<a name="line-1728"></a>output_arg {
+<a name="line-1729"></a>  description: "= Same as `ref`.  Returned as a convenience for operations that want\nto use the updated values after the update is done."
+<a name="line-1730"></a>  is_ref: true
+<a name="line-1731"></a>  name: "output_ref"
+<a name="line-1732"></a>  type_attr: "T"
+<a name="line-1733"></a>}
+<a name="line-1734"></a>-}</span>
+<a name="line-1735"></a>
+<a name="line-1736"></a><a name="reduceJoin"></a><span class='hs-comment'>-- | Joins a string Tensor across the given dimensions.</span>
+<a name="line-1737"></a><span class='hs-comment'>--</span>
+<a name="line-1738"></a><span class='hs-comment'>-- Computes the string join across dimensions in the given string Tensor of shape</span>
+<a name="line-1739"></a><span class='hs-comment'>-- `[d_0, d_1, ..., d_n-1]`.  Returns a new Tensor created by joining the input</span>
+<a name="line-1740"></a><span class='hs-comment'>-- strings with the given separator (default: empty string).  Negative indices are</span>
+<a name="line-1741"></a><span class='hs-comment'>-- counted backwards from the end, with `-1` being equivalent to `n - 1`.  Passing</span>
+<a name="line-1742"></a><span class='hs-comment'>-- an empty `reduction_indices` joins all strings in linear index order and outputs</span>
+<a name="line-1743"></a><span class='hs-comment'>-- a scalar string.</span>
+<a name="line-1744"></a><span class='hs-comment'>-- </span>
+<a name="line-1745"></a><span class='hs-comment'>-- </span>
+<a name="line-1746"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-1747"></a><span class='hs-comment'>-- </span>
+<a name="line-1748"></a><span class='hs-comment'>-- ```</span>
+<a name="line-1749"></a><span class='hs-comment'>-- # tensor `a` is [["a", "b"], ["c", "d"]]</span>
+<a name="line-1750"></a><span class='hs-comment'>-- tf.reduce_join(a, 0) ==&gt; ["ac", "bd"]</span>
+<a name="line-1751"></a><span class='hs-comment'>-- tf.reduce_join(a, 1) ==&gt; ["ab", "cd"]</span>
+<a name="line-1752"></a><span class='hs-comment'>-- tf.reduce_join(a, -2) = tf.reduce_join(a, 0) ==&gt; ["ac", "bd"]</span>
+<a name="line-1753"></a><span class='hs-comment'>-- tf.reduce_join(a, -1) = tf.reduce_join(a, 1) ==&gt; ["ab", "cd"]</span>
+<a name="line-1754"></a><span class='hs-comment'>-- tf.reduce_join(a, 0, keep_dims=True) ==&gt; [["ac", "bd"]]</span>
+<a name="line-1755"></a><span class='hs-comment'>-- tf.reduce_join(a, 1, keep_dims=True) ==&gt; [["ab"], ["cd"]]</span>
+<a name="line-1756"></a><span class='hs-comment'>-- tf.reduce_join(a, 0, separator=".") ==&gt; ["a.c", "b.d"]</span>
+<a name="line-1757"></a><span class='hs-comment'>-- tf.reduce_join(a, [0, 1]) ==&gt; ["acbd"]</span>
+<a name="line-1758"></a><span class='hs-comment'>-- tf.reduce_join(a, [1, 0]) ==&gt; ["abcd"]</span>
+<a name="line-1759"></a><span class='hs-comment'>-- tf.reduce_join(a, []) ==&gt; ["abcd"]</span>
+<a name="line-1760"></a><span class='hs-comment'>-- ```</span>
+<a name="line-1761"></a><span class='hs-definition'>reduceJoin</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __inputs__: The input to be joined.  All reduced indices must have non-zero size.</span>
+<a name="line-1762"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __reduction_indices__: The dimensions to reduce over.  Dimensions are reduced in the</span>
+<a name="line-1763"></a>                                          <span class='hs-comment'>-- order specified.  Omitting `reduction_indices` is equivalent to passing</span>
+<a name="line-1764"></a>                                          <span class='hs-comment'>-- `[n-1, n-2, ..., 0]`.  Negative indices from `-n` to `-1` are supported.</span>
+<a name="line-1765"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __output__: Has shape equal to that of the input with reduced dimensions removed or</span>
+<a name="line-1766"></a>              <span class='hs-comment'>-- set to `1` depending on `keep_dims`.</span>
+<a name="line-1767"></a><span class='hs-definition'>reduceJoin</span> <span class='hs-varid'>inputs</span> <span class='hs-varid'>reduction_indices</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1768"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ReduceJoin"</span><span class='hs-layout'>)</span>
+<a name="line-1769"></a>        <span class='hs-varid'>inputs</span> <span class='hs-varid'>reduction_indices</span>
+<a name="line-1770"></a><span class='hs-comment'>{-
+<a name="line-1771"></a>attr {
+<a name="line-1772"></a>  default_value { b: false }
+<a name="line-1773"></a>  description: "If `True`, retain reduced dimensions with length `1`."
+<a name="line-1774"></a>  name: "keep_dims"
+<a name="line-1775"></a>  type: "bool"
+<a name="line-1776"></a>}
+<a name="line-1777"></a>attr {
+<a name="line-1778"></a>  default_value { s: "" }
+<a name="line-1779"></a>  description: "The separator to use when joining."
+<a name="line-1780"></a>  name: "separator"
+<a name="line-1781"></a>  type: "string"
+<a name="line-1782"></a>}
+<a name="line-1783"></a>input_arg {
+<a name="line-1784"></a>  description: "The input to be joined.  All reduced indices must have non-zero size."
+<a name="line-1785"></a>  name: "inputs"
+<a name="line-1786"></a>  type: DT_STRING
+<a name="line-1787"></a>}
+<a name="line-1788"></a>input_arg {
+<a name="line-1789"></a>  description: "The dimensions to reduce over.  Dimensions are reduced in the\norder specified.  Omitting `reduction_indices` is equivalent to passing\n`[n-1, n-2, ..., 0]`.  Negative indices from `-n` to `-1` are supported."
+<a name="line-1790"></a>  name: "reduction_indices"
+<a name="line-1791"></a>  type: DT_INT32
+<a name="line-1792"></a>}
+<a name="line-1793"></a>output_arg {
+<a name="line-1794"></a>  description: "Has shape equal to that of the input with reduced dimensions removed or\nset to `1` depending on `keep_dims`."
+<a name="line-1795"></a>  name: "output"
+<a name="line-1796"></a>  type: DT_STRING
+<a name="line-1797"></a>}
+<a name="line-1798"></a>-}</span>
+<a name="line-1799"></a>
+<a name="line-1800"></a><a name="scatterSub"></a><span class='hs-comment'>-- | Subtracts sparse updates to a variable reference.</span>
+<a name="line-1801"></a><span class='hs-comment'>--</span>
+<a name="line-1802"></a><span class='hs-comment'>--     # Scalar indices</span>
+<a name="line-1803"></a><span class='hs-comment'>--     ref[indices, ...] -= updates[...]</span>
+<a name="line-1804"></a><span class='hs-comment'>-- </span>
+<a name="line-1805"></a><span class='hs-comment'>--     # Vector indices (for each i)</span>
+<a name="line-1806"></a><span class='hs-comment'>--     ref[indices[i], ...] -= updates[i, ...]</span>
+<a name="line-1807"></a><span class='hs-comment'>-- </span>
+<a name="line-1808"></a><span class='hs-comment'>--     # High rank indices (for each i, ..., j)</span>
+<a name="line-1809"></a><span class='hs-comment'>--     ref[indices[i, ..., j], ...] -= updates[i, ..., j, ...]</span>
+<a name="line-1810"></a><span class='hs-comment'>-- </span>
+<a name="line-1811"></a><span class='hs-comment'>-- This operation outputs `ref` after the update is done.</span>
+<a name="line-1812"></a><span class='hs-comment'>-- This makes it easier to chain operations that need to use the reset value.</span>
+<a name="line-1813"></a><span class='hs-comment'>-- </span>
+<a name="line-1814"></a><span class='hs-comment'>-- Duplicate entries are handled correctly: if multiple `indices` reference</span>
+<a name="line-1815"></a><span class='hs-comment'>-- the same location, their (negated) contributions add.</span>
+<a name="line-1816"></a><span class='hs-comment'>-- </span>
+<a name="line-1817"></a><span class='hs-comment'>-- Requires `updates.shape = indices.shape + ref.shape[1:]`.</span>
+<a name="line-1818"></a><span class='hs-comment'>-- </span>
+<a name="line-1819"></a><span class='hs-comment'>-- &lt;div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;"&gt;</span>
+<a name="line-1820"></a><span class='hs-comment'>-- &lt;img style="width:100%" src="../../images/ScatterSub.png" alt&gt;</span>
+<a name="line-1821"></a><span class='hs-comment'>-- &lt;/div&gt;</span>
+<a name="line-1822"></a><span class='hs-definition'>scatterSub</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-1823"></a>                                            <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1824"></a>                                                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1825"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-1826"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-1827"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-1828"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-1829"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-1830"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-1831"></a>                                                    <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-1832"></a>                                            <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-1833"></a>                                            <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-1834"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1835"></a>              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __ref__: Should be from a `Variable` node.</span>
+<a name="line-1836"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __indices__: A tensor of indices into the first dimension of `ref`.</span>
+<a name="line-1837"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __updates__: A tensor of updated values to subtract from `ref`.</span>
+<a name="line-1838"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output_ref__: = Same as `ref`.  Returned as a convenience for operations that want</span>
+<a name="line-1839"></a>              <span class='hs-comment'>-- to use the updated values after the update is done.</span>
+<a name="line-1840"></a><span class='hs-definition'>scatterSub</span> <span class='hs-varid'>ref</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>updates</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1841"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ScatterSub"</span>
+<a name="line-1842"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-1843"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-1844"></a>        <span class='hs-varid'>ref</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>updates</span>
+<a name="line-1845"></a><span class='hs-comment'>{-
+<a name="line-1846"></a>attr {
+<a name="line-1847"></a>  allowed_values {
+<a name="line-1848"></a>    list {
+<a name="line-1849"></a>      type: DT_FLOAT
+<a name="line-1850"></a>      type: DT_DOUBLE
+<a name="line-1851"></a>      type: DT_INT64
+<a name="line-1852"></a>      type: DT_INT32
+<a name="line-1853"></a>      type: DT_UINT8
+<a name="line-1854"></a>      type: DT_UINT16
+<a name="line-1855"></a>      type: DT_INT16
+<a name="line-1856"></a>      type: DT_INT8
+<a name="line-1857"></a>      type: DT_COMPLEX64
+<a name="line-1858"></a>      type: DT_COMPLEX128
+<a name="line-1859"></a>      type: DT_QINT8
+<a name="line-1860"></a>      type: DT_QUINT8
+<a name="line-1861"></a>      type: DT_QINT32
+<a name="line-1862"></a>      type: DT_HALF
+<a name="line-1863"></a>    }
+<a name="line-1864"></a>  }
+<a name="line-1865"></a>  name: "T"
+<a name="line-1866"></a>  type: "type"
+<a name="line-1867"></a>}
+<a name="line-1868"></a>attr {
+<a name="line-1869"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-1870"></a>  name: "Tindices"
+<a name="line-1871"></a>  type: "type"
+<a name="line-1872"></a>}
+<a name="line-1873"></a>attr {
+<a name="line-1874"></a>  default_value { b: false }
+<a name="line-1875"></a>  description: "If True, the subtraction will be protected by a lock;\notherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-1876"></a>  name: "use_locking"
+<a name="line-1877"></a>  type: "bool"
+<a name="line-1878"></a>}
+<a name="line-1879"></a>input_arg {
+<a name="line-1880"></a>  description: "Should be from a `Variable` node."
+<a name="line-1881"></a>  is_ref: true
+<a name="line-1882"></a>  name: "ref"
+<a name="line-1883"></a>  type_attr: "T"
+<a name="line-1884"></a>}
+<a name="line-1885"></a>input_arg {
+<a name="line-1886"></a>  description: "A tensor of indices into the first dimension of `ref`."
+<a name="line-1887"></a>  name: "indices"
+<a name="line-1888"></a>  type_attr: "Tindices"
+<a name="line-1889"></a>}
+<a name="line-1890"></a>input_arg {
+<a name="line-1891"></a>  description: "A tensor of updated values to subtract from `ref`."
+<a name="line-1892"></a>  name: "updates"
+<a name="line-1893"></a>  type_attr: "T"
+<a name="line-1894"></a>}
+<a name="line-1895"></a>output_arg {
+<a name="line-1896"></a>  description: "= Same as `ref`.  Returned as a convenience for operations that want\nto use the updated values after the update is done."
+<a name="line-1897"></a>  is_ref: true
+<a name="line-1898"></a>  name: "output_ref"
+<a name="line-1899"></a>  type_attr: "T"
+<a name="line-1900"></a>}
+<a name="line-1901"></a>-}</span>
+<a name="line-1902"></a>
+<a name="line-1903"></a><a name="scatterAdd"></a><span class='hs-comment'>-- | Adds sparse updates to a variable reference.</span>
+<a name="line-1904"></a><span class='hs-comment'>--</span>
+<a name="line-1905"></a><span class='hs-comment'>-- This operation computes</span>
+<a name="line-1906"></a><span class='hs-comment'>-- </span>
+<a name="line-1907"></a><span class='hs-comment'>--     # Scalar indices</span>
+<a name="line-1908"></a><span class='hs-comment'>--     ref[indices, ...] += updates[...]</span>
+<a name="line-1909"></a><span class='hs-comment'>-- </span>
+<a name="line-1910"></a><span class='hs-comment'>--     # Vector indices (for each i)</span>
+<a name="line-1911"></a><span class='hs-comment'>--     ref[indices[i], ...] += updates[i, ...]</span>
+<a name="line-1912"></a><span class='hs-comment'>-- </span>
+<a name="line-1913"></a><span class='hs-comment'>--     # High rank indices (for each i, ..., j)</span>
+<a name="line-1914"></a><span class='hs-comment'>--     ref[indices[i, ..., j], ...] += updates[i, ..., j, ...]</span>
+<a name="line-1915"></a><span class='hs-comment'>-- </span>
+<a name="line-1916"></a><span class='hs-comment'>-- This operation outputs `ref` after the update is done.</span>
+<a name="line-1917"></a><span class='hs-comment'>-- This makes it easier to chain operations that need to use the reset value.</span>
+<a name="line-1918"></a><span class='hs-comment'>-- </span>
+<a name="line-1919"></a><span class='hs-comment'>-- Duplicate entries are handled correctly: if multiple `indices` reference</span>
+<a name="line-1920"></a><span class='hs-comment'>-- the same location, their contributions add.</span>
+<a name="line-1921"></a><span class='hs-comment'>-- </span>
+<a name="line-1922"></a><span class='hs-comment'>-- Requires `updates.shape = indices.shape + ref.shape[1:]`.</span>
+<a name="line-1923"></a><span class='hs-comment'>-- </span>
+<a name="line-1924"></a><span class='hs-comment'>-- &lt;div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;"&gt;</span>
+<a name="line-1925"></a><span class='hs-comment'>-- &lt;img style="width:100%" src="../../images/ScatterAdd.png" alt&gt;</span>
+<a name="line-1926"></a><span class='hs-comment'>-- &lt;/div&gt;</span>
+<a name="line-1927"></a><span class='hs-definition'>scatterAdd</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-1928"></a>                                            <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1929"></a>                                                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1930"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-1931"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-1932"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-1933"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-1934"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-1935"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-1936"></a>                                                    <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-1937"></a>                                            <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-1938"></a>                                            <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-1939"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1940"></a>              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __ref__: Should be from a `Variable` node.</span>
+<a name="line-1941"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __indices__: A tensor of indices into the first dimension of `ref`.</span>
+<a name="line-1942"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __updates__: A tensor of updated values to add to `ref`.</span>
+<a name="line-1943"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output_ref__: = Same as `ref`.  Returned as a convenience for operations that want</span>
+<a name="line-1944"></a>              <span class='hs-comment'>-- to use the updated values after the update is done.</span>
+<a name="line-1945"></a><span class='hs-definition'>scatterAdd</span> <span class='hs-varid'>ref</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>updates</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1946"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ScatterAdd"</span>
+<a name="line-1947"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-1948"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-1949"></a>        <span class='hs-varid'>ref</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>updates</span>
+<a name="line-1950"></a><span class='hs-comment'>{-
+<a name="line-1951"></a>attr {
+<a name="line-1952"></a>  allowed_values {
+<a name="line-1953"></a>    list {
+<a name="line-1954"></a>      type: DT_FLOAT
+<a name="line-1955"></a>      type: DT_DOUBLE
+<a name="line-1956"></a>      type: DT_INT64
+<a name="line-1957"></a>      type: DT_INT32
+<a name="line-1958"></a>      type: DT_UINT8
+<a name="line-1959"></a>      type: DT_UINT16
+<a name="line-1960"></a>      type: DT_INT16
+<a name="line-1961"></a>      type: DT_INT8
+<a name="line-1962"></a>      type: DT_COMPLEX64
+<a name="line-1963"></a>      type: DT_COMPLEX128
+<a name="line-1964"></a>      type: DT_QINT8
+<a name="line-1965"></a>      type: DT_QUINT8
+<a name="line-1966"></a>      type: DT_QINT32
+<a name="line-1967"></a>      type: DT_HALF
+<a name="line-1968"></a>    }
+<a name="line-1969"></a>  }
+<a name="line-1970"></a>  name: "T"
+<a name="line-1971"></a>  type: "type"
+<a name="line-1972"></a>}
+<a name="line-1973"></a>attr {
+<a name="line-1974"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-1975"></a>  name: "Tindices"
+<a name="line-1976"></a>  type: "type"
+<a name="line-1977"></a>}
+<a name="line-1978"></a>attr {
+<a name="line-1979"></a>  default_value { b: false }
+<a name="line-1980"></a>  description: "If True, the addition will be protected by a lock;\notherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-1981"></a>  name: "use_locking"
+<a name="line-1982"></a>  type: "bool"
+<a name="line-1983"></a>}
+<a name="line-1984"></a>input_arg {
+<a name="line-1985"></a>  description: "Should be from a `Variable` node."
+<a name="line-1986"></a>  is_ref: true
+<a name="line-1987"></a>  name: "ref"
+<a name="line-1988"></a>  type_attr: "T"
+<a name="line-1989"></a>}
+<a name="line-1990"></a>input_arg {
+<a name="line-1991"></a>  description: "A tensor of indices into the first dimension of `ref`."
+<a name="line-1992"></a>  name: "indices"
+<a name="line-1993"></a>  type_attr: "Tindices"
+<a name="line-1994"></a>}
+<a name="line-1995"></a>input_arg {
+<a name="line-1996"></a>  description: "A tensor of updated values to add to `ref`."
+<a name="line-1997"></a>  name: "updates"
+<a name="line-1998"></a>  type_attr: "T"
+<a name="line-1999"></a>}
+<a name="line-2000"></a>output_arg {
+<a name="line-2001"></a>  description: "= Same as `ref`.  Returned as a convenience for operations that want\nto use the updated values after the update is done."
+<a name="line-2002"></a>  is_ref: true
+<a name="line-2003"></a>  name: "output_ref"
+<a name="line-2004"></a>  type_attr: "T"
+<a name="line-2005"></a>}
+<a name="line-2006"></a>-}</span>
+<a name="line-2007"></a>
+<a name="line-2008"></a><a name="scatterUpdate"></a><span class='hs-comment'>-- | Applies sparse updates to a variable reference.</span>
+<a name="line-2009"></a><span class='hs-comment'>--</span>
+<a name="line-2010"></a><span class='hs-comment'>-- This operation computes</span>
+<a name="line-2011"></a><span class='hs-comment'>-- </span>
+<a name="line-2012"></a><span class='hs-comment'>--     # Scalar indices</span>
+<a name="line-2013"></a><span class='hs-comment'>--     ref[indices, ...] = updates[...]</span>
+<a name="line-2014"></a><span class='hs-comment'>-- </span>
+<a name="line-2015"></a><span class='hs-comment'>--     # Vector indices (for each i)</span>
+<a name="line-2016"></a><span class='hs-comment'>--     ref[indices[i], ...] = updates[i, ...]</span>
+<a name="line-2017"></a><span class='hs-comment'>-- </span>
+<a name="line-2018"></a><span class='hs-comment'>--     # High rank indices (for each i, ..., j)</span>
+<a name="line-2019"></a><span class='hs-comment'>--     ref[indices[i, ..., j], ...] = updates[i, ..., j, ...]</span>
+<a name="line-2020"></a><span class='hs-comment'>-- </span>
+<a name="line-2021"></a><span class='hs-comment'>-- This operation outputs `ref` after the update is done.</span>
+<a name="line-2022"></a><span class='hs-comment'>-- This makes it easier to chain operations that need to use the reset value.</span>
+<a name="line-2023"></a><span class='hs-comment'>-- </span>
+<a name="line-2024"></a><span class='hs-comment'>-- If values in `ref` is to be updated more than once, because there are</span>
+<a name="line-2025"></a><span class='hs-comment'>-- duplicate entires in `indices`, the order at which the updates happen</span>
+<a name="line-2026"></a><span class='hs-comment'>-- for each value is undefined.</span>
+<a name="line-2027"></a><span class='hs-comment'>-- </span>
+<a name="line-2028"></a><span class='hs-comment'>-- Requires `updates.shape = indices.shape + ref.shape[1:]`.</span>
+<a name="line-2029"></a><span class='hs-comment'>-- </span>
+<a name="line-2030"></a><span class='hs-comment'>-- &lt;div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;"&gt;</span>
+<a name="line-2031"></a><span class='hs-comment'>-- &lt;img style="width:100%" src="../../images/ScatterUpdate.png" alt&gt;</span>
+<a name="line-2032"></a><span class='hs-comment'>-- &lt;/div&gt;</span>
+<a name="line-2033"></a><span class='hs-definition'>scatterUpdate</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-2034"></a>                                               <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-2035"></a>                                               <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-2036"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2037"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __ref__: Should be from a `Variable` node.</span>
+<a name="line-2038"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __indices__: A tensor of indices into the first dimension of `ref`.</span>
+<a name="line-2039"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __updates__: A tensor of updated values to store in `ref`.</span>
+<a name="line-2040"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output_ref__: = Same as `ref`.  Returned as a convenience for operations that want</span>
+<a name="line-2041"></a>                 <span class='hs-comment'>-- to use the updated values after the update is done.</span>
+<a name="line-2042"></a><span class='hs-definition'>scatterUpdate</span> <span class='hs-varid'>ref</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>updates</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2043"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ScatterUpdate"</span>
+<a name="line-2044"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-2045"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-2046"></a>        <span class='hs-varid'>ref</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>updates</span>
+<a name="line-2047"></a><span class='hs-comment'>{-
+<a name="line-2048"></a>attr { name: "T" type: "type" }
+<a name="line-2049"></a>attr {
+<a name="line-2050"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-2051"></a>  name: "Tindices"
+<a name="line-2052"></a>  type: "type"
+<a name="line-2053"></a>}
+<a name="line-2054"></a>attr {
+<a name="line-2055"></a>  default_value { b: true }
+<a name="line-2056"></a>  description: "If True, the assignment will be protected by a lock;\notherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-2057"></a>  name: "use_locking"
+<a name="line-2058"></a>  type: "bool"
+<a name="line-2059"></a>}
+<a name="line-2060"></a>input_arg {
+<a name="line-2061"></a>  description: "Should be from a `Variable` node."
+<a name="line-2062"></a>  is_ref: true
+<a name="line-2063"></a>  name: "ref"
+<a name="line-2064"></a>  type_attr: "T"
+<a name="line-2065"></a>}
+<a name="line-2066"></a>input_arg {
+<a name="line-2067"></a>  description: "A tensor of indices into the first dimension of `ref`."
+<a name="line-2068"></a>  name: "indices"
+<a name="line-2069"></a>  type_attr: "Tindices"
+<a name="line-2070"></a>}
+<a name="line-2071"></a>input_arg {
+<a name="line-2072"></a>  description: "A tensor of updated values to store in `ref`."
+<a name="line-2073"></a>  name: "updates"
+<a name="line-2074"></a>  type_attr: "T"
+<a name="line-2075"></a>}
+<a name="line-2076"></a>output_arg {
+<a name="line-2077"></a>  description: "= Same as `ref`.  Returned as a convenience for operations that want\nto use the updated values after the update is done."
+<a name="line-2078"></a>  is_ref: true
+<a name="line-2079"></a>  name: "output_ref"
+<a name="line-2080"></a>  type_attr: "T"
+<a name="line-2081"></a>}
+<a name="line-2082"></a>-}</span>
+<a name="line-2083"></a>
+<a name="line-2084"></a><a name="assignSub"></a><span class='hs-comment'>-- | Update 'ref' by subtracting 'value' from it.</span>
+<a name="line-2085"></a><span class='hs-comment'>--</span>
+<a name="line-2086"></a><span class='hs-comment'>-- This operation outputs "ref" after the update is done.</span>
+<a name="line-2087"></a><span class='hs-comment'>-- This makes it easier to chain operations that need to use the reset value.</span>
+<a name="line-2088"></a><span class='hs-definition'>assignSub</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-2089"></a>                               <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-2090"></a>                                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-2091"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-2092"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-2093"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-2094"></a>                                       <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2095"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __ref__: Should be from a `Variable` node.</span>
+<a name="line-2096"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: The value to be subtracted to the variable.</span>
+<a name="line-2097"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output_ref__: = Same as "ref".  Returned as a convenience for operations that want</span>
+<a name="line-2098"></a>             <span class='hs-comment'>-- to use the new value after the variable has been updated.</span>
+<a name="line-2099"></a><span class='hs-definition'>assignSub</span> <span class='hs-varid'>ref</span> <span class='hs-varid'>value</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2100"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"AssignSub"</span>
+<a name="line-2101"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-2102"></a>        <span class='hs-varid'>ref</span> <span class='hs-varid'>value</span>
+<a name="line-2103"></a><span class='hs-comment'>{-
+<a name="line-2104"></a>attr {
+<a name="line-2105"></a>  allowed_values {
+<a name="line-2106"></a>    list {
+<a name="line-2107"></a>      type: DT_FLOAT
+<a name="line-2108"></a>      type: DT_DOUBLE
+<a name="line-2109"></a>      type: DT_INT64
+<a name="line-2110"></a>      type: DT_INT32
+<a name="line-2111"></a>      type: DT_UINT8
+<a name="line-2112"></a>      type: DT_UINT16
+<a name="line-2113"></a>      type: DT_INT16
+<a name="line-2114"></a>      type: DT_INT8
+<a name="line-2115"></a>      type: DT_COMPLEX64
+<a name="line-2116"></a>      type: DT_COMPLEX128
+<a name="line-2117"></a>      type: DT_QINT8
+<a name="line-2118"></a>      type: DT_QUINT8
+<a name="line-2119"></a>      type: DT_QINT32
+<a name="line-2120"></a>      type: DT_HALF
+<a name="line-2121"></a>    }
+<a name="line-2122"></a>  }
+<a name="line-2123"></a>  name: "T"
+<a name="line-2124"></a>  type: "type"
+<a name="line-2125"></a>}
+<a name="line-2126"></a>attr {
+<a name="line-2127"></a>  default_value { b: false }
+<a name="line-2128"></a>  description: "If True, the subtraction will be protected by a lock;\notherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-2129"></a>  name: "use_locking"
+<a name="line-2130"></a>  type: "bool"
+<a name="line-2131"></a>}
+<a name="line-2132"></a>input_arg {
+<a name="line-2133"></a>  description: "Should be from a `Variable` node."
+<a name="line-2134"></a>  is_ref: true
+<a name="line-2135"></a>  name: "ref"
+<a name="line-2136"></a>  type_attr: "T"
+<a name="line-2137"></a>}
+<a name="line-2138"></a>input_arg {
+<a name="line-2139"></a>  description: "The value to be subtracted to the variable."
+<a name="line-2140"></a>  name: "value"
+<a name="line-2141"></a>  type_attr: "T"
+<a name="line-2142"></a>}
+<a name="line-2143"></a>output_arg {
+<a name="line-2144"></a>  description: "= Same as \"ref\".  Returned as a convenience for operations that want\nto use the new value after the variable has been updated."
+<a name="line-2145"></a>  is_ref: true
+<a name="line-2146"></a>  name: "output_ref"
+<a name="line-2147"></a>  type_attr: "T"
+<a name="line-2148"></a>}
+<a name="line-2149"></a>-}</span>
+<a name="line-2150"></a>
+<a name="line-2151"></a><a name="assignAdd"></a><span class='hs-comment'>-- | Update 'ref' by adding 'value' to it.</span>
+<a name="line-2152"></a><span class='hs-comment'>--</span>
+<a name="line-2153"></a><span class='hs-comment'>-- This operation outputs "ref" after the update is done.</span>
+<a name="line-2154"></a><span class='hs-comment'>-- This makes it easier to chain operations that need to use the reset value.</span>
+<a name="line-2155"></a><span class='hs-definition'>assignAdd</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-2156"></a>                               <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-2157"></a>                                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-2158"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-2159"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-2160"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-2161"></a>                                       <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2162"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __ref__: Should be from a `Variable` node.</span>
+<a name="line-2163"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: The value to be added to the variable.</span>
+<a name="line-2164"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output_ref__: = Same as "ref".  Returned as a convenience for operations that want</span>
+<a name="line-2165"></a>             <span class='hs-comment'>-- to use the new value after the variable has been updated.</span>
+<a name="line-2166"></a><span class='hs-definition'>assignAdd</span> <span class='hs-varid'>ref</span> <span class='hs-varid'>value</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2167"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"AssignAdd"</span>
+<a name="line-2168"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-2169"></a>        <span class='hs-varid'>ref</span> <span class='hs-varid'>value</span>
+<a name="line-2170"></a><span class='hs-comment'>{-
+<a name="line-2171"></a>attr {
+<a name="line-2172"></a>  allowed_values {
+<a name="line-2173"></a>    list {
+<a name="line-2174"></a>      type: DT_FLOAT
+<a name="line-2175"></a>      type: DT_DOUBLE
+<a name="line-2176"></a>      type: DT_INT64
+<a name="line-2177"></a>      type: DT_INT32
+<a name="line-2178"></a>      type: DT_UINT8
+<a name="line-2179"></a>      type: DT_UINT16
+<a name="line-2180"></a>      type: DT_INT16
+<a name="line-2181"></a>      type: DT_INT8
+<a name="line-2182"></a>      type: DT_COMPLEX64
+<a name="line-2183"></a>      type: DT_COMPLEX128
+<a name="line-2184"></a>      type: DT_QINT8
+<a name="line-2185"></a>      type: DT_QUINT8
+<a name="line-2186"></a>      type: DT_QINT32
+<a name="line-2187"></a>      type: DT_HALF
+<a name="line-2188"></a>    }
+<a name="line-2189"></a>  }
+<a name="line-2190"></a>  name: "T"
+<a name="line-2191"></a>  type: "type"
+<a name="line-2192"></a>}
+<a name="line-2193"></a>attr {
+<a name="line-2194"></a>  default_value { b: false }
+<a name="line-2195"></a>  description: "If True, the addition will be protected by a lock;\notherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-2196"></a>  name: "use_locking"
+<a name="line-2197"></a>  type: "bool"
+<a name="line-2198"></a>}
+<a name="line-2199"></a>input_arg {
+<a name="line-2200"></a>  description: "Should be from a `Variable` node."
+<a name="line-2201"></a>  is_ref: true
+<a name="line-2202"></a>  name: "ref"
+<a name="line-2203"></a>  type_attr: "T"
+<a name="line-2204"></a>}
+<a name="line-2205"></a>input_arg {
+<a name="line-2206"></a>  description: "The value to be added to the variable."
+<a name="line-2207"></a>  name: "value"
+<a name="line-2208"></a>  type_attr: "T"
+<a name="line-2209"></a>}
+<a name="line-2210"></a>output_arg {
+<a name="line-2211"></a>  description: "= Same as \"ref\".  Returned as a convenience for operations that want\nto use the new value after the variable has been updated."
+<a name="line-2212"></a>  is_ref: true
+<a name="line-2213"></a>  name: "output_ref"
+<a name="line-2214"></a>  type_attr: "T"
+<a name="line-2215"></a>}
+<a name="line-2216"></a>-}</span>
+<a name="line-2217"></a>
+<a name="line-2218"></a><a name="sparseSegmentMeanGrad"></a><span class='hs-comment'>-- | Computes gradients for SparseSegmentMean.</span>
+<a name="line-2219"></a><span class='hs-comment'>--</span>
+<a name="line-2220"></a><span class='hs-comment'>-- Returns tensor "output" with same shape as grad, except for dimension 0 whose</span>
+<a name="line-2221"></a><span class='hs-comment'>-- value is output_dim0.</span>
+<a name="line-2222"></a><span class='hs-definition'>sparseSegmentMeanGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-2223"></a>                                                      <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-2224"></a>                                                      <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span>
+<a name="line-2225"></a>                                                      <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-2226"></a>                                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2227"></a>                         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: gradient propagated to the SparseSegmentMean op.</span>
+<a name="line-2228"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __indices__: indices passed to the corresponding SparseSegmentMean op.</span>
+<a name="line-2229"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __segment_ids__: segment_ids passed to the corresponding SparseSegmentMean op.</span>
+<a name="line-2230"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __output_dim0__: dimension 0 of "data" passed to SparseSegmentMean op.</span>
+<a name="line-2231"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-2232"></a><span class='hs-definition'>sparseSegmentMeanGrad</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>segment_ids</span> <span class='hs-varid'>output_dim0</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2233"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseSegmentMeanGrad"</span>
+<a name="line-2234"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-2235"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-2236"></a>        <span class='hs-varid'>grad</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>segment_ids</span> <span class='hs-varid'>output_dim0</span>
+<a name="line-2237"></a><span class='hs-comment'>{-
+<a name="line-2238"></a>attr {
+<a name="line-2239"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-2240"></a>  name: "T"
+<a name="line-2241"></a>  type: "type"
+<a name="line-2242"></a>}
+<a name="line-2243"></a>attr {
+<a name="line-2244"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-2245"></a>  default_value { type: DT_INT32 }
+<a name="line-2246"></a>  name: "Tidx"
+<a name="line-2247"></a>  type: "type"
+<a name="line-2248"></a>}
+<a name="line-2249"></a>input_arg {
+<a name="line-2250"></a>  description: "gradient propagated to the SparseSegmentMean op."
+<a name="line-2251"></a>  name: "grad"
+<a name="line-2252"></a>  type_attr: "T"
+<a name="line-2253"></a>}
+<a name="line-2254"></a>input_arg {
+<a name="line-2255"></a>  description: "indices passed to the corresponding SparseSegmentMean op."
+<a name="line-2256"></a>  name: "indices"
+<a name="line-2257"></a>  type_attr: "Tidx"
+<a name="line-2258"></a>}
+<a name="line-2259"></a>input_arg {
+<a name="line-2260"></a>  description: "segment_ids passed to the corresponding SparseSegmentMean op."
+<a name="line-2261"></a>  name: "segment_ids"
+<a name="line-2262"></a>  type: DT_INT32
+<a name="line-2263"></a>}
+<a name="line-2264"></a>input_arg {
+<a name="line-2265"></a>  description: "dimension 0 of \"data\" passed to SparseSegmentMean op."
+<a name="line-2266"></a>  name: "output_dim0"
+<a name="line-2267"></a>  type: DT_INT32
+<a name="line-2268"></a>}
+<a name="line-2269"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-2270"></a>-}</span>
+<a name="line-2271"></a>
+<a name="line-2272"></a><a name="sparseSoftmax"></a><span class='hs-comment'>-- | Applies softmax to a batched N-D `SparseTensor`.</span>
+<a name="line-2273"></a><span class='hs-comment'>--</span>
+<a name="line-2274"></a><span class='hs-comment'>-- The inputs represent an N-D SparseTensor  with logical shape `[..., B, C]`</span>
+<a name="line-2275"></a><span class='hs-comment'>-- (where `N &gt;= 2`), and with indices sorted in the canonical lexicographic order.</span>
+<a name="line-2276"></a><span class='hs-comment'>-- </span>
+<a name="line-2277"></a><span class='hs-comment'>-- This op is equivalent to applying the normal `tf.nn.softmax()` to each innermost</span>
+<a name="line-2278"></a><span class='hs-comment'>-- logical submatrix with shape `[B, C]`, but with the catch that *the implicitly</span>
+<a name="line-2279"></a><span class='hs-comment'>-- zero elements do not participate*.  Specifically, the algorithm is equivalent</span>
+<a name="line-2280"></a><span class='hs-comment'>-- to the following:</span>
+<a name="line-2281"></a><span class='hs-comment'>-- </span>
+<a name="line-2282"></a><span class='hs-comment'>--   (1) Applies `tf.nn.softmax()` to a densified view of each innermost submatrix</span>
+<a name="line-2283"></a><span class='hs-comment'>--       with shape `[B, C]`, along the size-C dimension;</span>
+<a name="line-2284"></a><span class='hs-comment'>--   (2) Masks out the original implicitly-zero locations;</span>
+<a name="line-2285"></a><span class='hs-comment'>--   (3) Renormalizes the remaining elements.</span>
+<a name="line-2286"></a><span class='hs-comment'>-- </span>
+<a name="line-2287"></a><span class='hs-comment'>-- Hence, the `SparseTensor` result has exactly the same non-zero indices and</span>
+<a name="line-2288"></a><span class='hs-comment'>-- shape.</span>
+<a name="line-2289"></a><span class='hs-definition'>sparseSoftmax</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2290"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __sp_indices__: 2-D.  `NNZ x R` matrix with the indices of non-empty values in a</span>
+<a name="line-2291"></a>                                          <span class='hs-comment'>-- SparseTensor, in canonical ordering.</span>
+<a name="line-2292"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __sp_values__: 1-D.  `NNZ` non-empty values corresponding to `sp_indices`.</span>
+<a name="line-2293"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __sp_shape__: 1-D.  Shape of the input SparseTensor.</span>
+<a name="line-2294"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 1-D.  The `NNZ` values for the result `SparseTensor`.</span>
+<a name="line-2295"></a><span class='hs-definition'>sparseSoftmax</span> <span class='hs-varid'>sp_indices</span> <span class='hs-varid'>sp_values</span> <span class='hs-varid'>sp_shape</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2296"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseSoftmax"</span>
+<a name="line-2297"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-2298"></a>        <span class='hs-varid'>sp_indices</span> <span class='hs-varid'>sp_values</span> <span class='hs-varid'>sp_shape</span>
+<a name="line-2299"></a><span class='hs-comment'>{-
+<a name="line-2300"></a>attr {
+<a name="line-2301"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-2302"></a>  name: "T"
+<a name="line-2303"></a>  type: "type"
+<a name="line-2304"></a>}
+<a name="line-2305"></a>input_arg {
+<a name="line-2306"></a>  description: "2-D.  `NNZ x R` matrix with the indices of non-empty values in a\nSparseTensor, in canonical ordering."
+<a name="line-2307"></a>  name: "sp_indices"
+<a name="line-2308"></a>  type: DT_INT64
+<a name="line-2309"></a>}
+<a name="line-2310"></a>input_arg {
+<a name="line-2311"></a>  description: "1-D.  `NNZ` non-empty values corresponding to `sp_indices`."
+<a name="line-2312"></a>  name: "sp_values"
+<a name="line-2313"></a>  type_attr: "T"
+<a name="line-2314"></a>}
+<a name="line-2315"></a>input_arg {
+<a name="line-2316"></a>  description: "1-D.  Shape of the input SparseTensor."
+<a name="line-2317"></a>  name: "sp_shape"
+<a name="line-2318"></a>  type: DT_INT64
+<a name="line-2319"></a>}
+<a name="line-2320"></a>output_arg {
+<a name="line-2321"></a>  description: "1-D.  The `NNZ` values for the result `SparseTensor`."
+<a name="line-2322"></a>  name: "output"
+<a name="line-2323"></a>  type_attr: "T"
+<a name="line-2324"></a>}
+<a name="line-2325"></a>-}</span>
+<a name="line-2326"></a>
+<a name="line-2327"></a><a name="matrixSolve"></a><span class='hs-comment'>-- | Solves systems of linear equations.</span>
+<a name="line-2328"></a><span class='hs-comment'>--</span>
+<a name="line-2329"></a><span class='hs-comment'>-- `Matrix` is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions</span>
+<a name="line-2330"></a><span class='hs-comment'>-- form square matrices. `Rhs` is a tensor of shape `[..., M, K]`. The `output` is</span>
+<a name="line-2331"></a><span class='hs-comment'>-- a tensor shape `[..., M, K]`.  If `adjoint` is `False` then each output matrix</span>
+<a name="line-2332"></a><span class='hs-comment'>-- satisfies `matrix[..., :, :] * output[..., :, :] = rhs[..., :, :]`.</span>
+<a name="line-2333"></a><span class='hs-comment'>-- If `adjoint` is `True` then each output matrix satisfies</span>
+<a name="line-2334"></a><span class='hs-comment'>-- `adjoint(matrix[..., :, :]) * output[..., :, :] = rhs[..., :, :]`.</span>
+<a name="line-2335"></a><span class='hs-definition'>matrixSolve</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2336"></a>               <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __matrix__: Shape is `[..., M, M]`.</span>
+<a name="line-2337"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __rhs__: Shape is `[..., M, K]`.</span>
+<a name="line-2338"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Shape is `[..., M, K]`.</span>
+<a name="line-2339"></a><span class='hs-definition'>matrixSolve</span> <span class='hs-varid'>matrix</span> <span class='hs-varid'>rhs</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2340"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MatrixSolve"</span>
+<a name="line-2341"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-2342"></a>        <span class='hs-varid'>matrix</span> <span class='hs-varid'>rhs</span>
+<a name="line-2343"></a><span class='hs-comment'>{-
+<a name="line-2344"></a>attr {
+<a name="line-2345"></a>  default_value { b: false }
+<a name="line-2346"></a>  description: "Boolean indicating whether to solve with `matrix` or its (block-wise)\nadjoint."
+<a name="line-2347"></a>  name: "adjoint"
+<a name="line-2348"></a>  type: "bool"
+<a name="line-2349"></a>}
+<a name="line-2350"></a>attr {
+<a name="line-2351"></a>  allowed_values { list { type: DT_DOUBLE type: DT_FLOAT } }
+<a name="line-2352"></a>  name: "T"
+<a name="line-2353"></a>  type: "type"
+<a name="line-2354"></a>}
+<a name="line-2355"></a>input_arg {
+<a name="line-2356"></a>  description: "Shape is `[..., M, M]`."
+<a name="line-2357"></a>  name: "matrix"
+<a name="line-2358"></a>  type_attr: "T"
+<a name="line-2359"></a>}
+<a name="line-2360"></a>input_arg {
+<a name="line-2361"></a>  description: "Shape is `[..., M, K]`." name: "rhs" type_attr: "T"
+<a name="line-2362"></a>}
+<a name="line-2363"></a>output_arg {
+<a name="line-2364"></a>  description: "Shape is `[..., M, K]`."
+<a name="line-2365"></a>  name: "output"
+<a name="line-2366"></a>  type_attr: "T"
+<a name="line-2367"></a>}
+<a name="line-2368"></a>-}</span>
+<a name="line-2369"></a>
+<a name="line-2370"></a><a name="selfAdjointEigV2"></a><span class='hs-comment'>-- | Computes the eigen decomposition of one or more square self-adjoint matrices.</span>
+<a name="line-2371"></a><span class='hs-comment'>--</span>
+<a name="line-2372"></a><span class='hs-comment'>-- Computes the eigenvalues and (optionally) eigenvectors of each inner matrix in</span>
+<a name="line-2373"></a><span class='hs-comment'>-- `input` such that `input[..., :, :] = v[..., :, :] * diag(e[..., :])`.</span>
+<a name="line-2374"></a><span class='hs-comment'>-- </span>
+<a name="line-2375"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-2376"></a><span class='hs-comment'>-- # a is a tensor.</span>
+<a name="line-2377"></a><span class='hs-comment'>-- # e is a tensor of eigenvalues.</span>
+<a name="line-2378"></a><span class='hs-comment'>-- # v is a tensor of eigenvectors.</span>
+<a name="line-2379"></a><span class='hs-comment'>-- e, v = self_adjoint_eig(a)</span>
+<a name="line-2380"></a><span class='hs-comment'>-- e = self_adjoint_eig(a, compute_v=False)</span>
+<a name="line-2381"></a><span class='hs-comment'>-- ```</span>
+<a name="line-2382"></a><span class='hs-definition'>selfAdjointEigV2</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2383"></a>                    <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: `Tensor` input of shape `[N, N]`.</span>
+<a name="line-2384"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-2385"></a>                    <span class='hs-comment'>-- ^ (__e__, __v__)</span>
+<a name="line-2386"></a>                    <span class='hs-comment'>--</span>
+<a name="line-2387"></a>                    <span class='hs-comment'>-- * __e__: Eigenvalues. Shape is `[N]`.</span>
+<a name="line-2388"></a>                    <span class='hs-comment'>--</span>
+<a name="line-2389"></a>                    <span class='hs-comment'>-- * __v__: Eigenvectors. Shape is `[N, N]`.</span>
+<a name="line-2390"></a><span class='hs-definition'>selfAdjointEigV2</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2391"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SelfAdjointEigV2"</span>
+<a name="line-2392"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-2393"></a>        <span class='hs-varid'>input</span>
+<a name="line-2394"></a><span class='hs-comment'>{-
+<a name="line-2395"></a>attr {
+<a name="line-2396"></a>  default_value { b: true }
+<a name="line-2397"></a>  description: "If `True` then eigenvectors will be computed and returned in `v`.\nOtherwise, only the eigenvalues will be computed."
+<a name="line-2398"></a>  name: "compute_v"
+<a name="line-2399"></a>  type: "bool"
+<a name="line-2400"></a>}
+<a name="line-2401"></a>attr {
+<a name="line-2402"></a>  allowed_values { list { type: DT_DOUBLE type: DT_FLOAT } }
+<a name="line-2403"></a>  name: "T"
+<a name="line-2404"></a>  type: "type"
+<a name="line-2405"></a>}
+<a name="line-2406"></a>input_arg {
+<a name="line-2407"></a>  description: "`Tensor` input of shape `[N, N]`."
+<a name="line-2408"></a>  name: "input"
+<a name="line-2409"></a>  type_attr: "T"
+<a name="line-2410"></a>}
+<a name="line-2411"></a>output_arg {
+<a name="line-2412"></a>  description: "Eigenvalues. Shape is `[N]`."
+<a name="line-2413"></a>  name: "e"
+<a name="line-2414"></a>  type_attr: "T"
+<a name="line-2415"></a>}
+<a name="line-2416"></a>output_arg {
+<a name="line-2417"></a>  description: "Eigenvectors. Shape is `[N, N]`."
+<a name="line-2418"></a>  name: "v"
+<a name="line-2419"></a>  type_attr: "T"
+<a name="line-2420"></a>}
+<a name="line-2421"></a>-}</span>
+<a name="line-2422"></a>
+<a name="line-2423"></a><a name="selfAdjointEig"></a><span class='hs-comment'>-- | Computes the Eigen Decomposition of a batch of square self-adjoint matrices.</span>
+<a name="line-2424"></a><span class='hs-comment'>--</span>
+<a name="line-2425"></a><span class='hs-comment'>-- The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions</span>
+<a name="line-2426"></a><span class='hs-comment'>-- form square matrices, with the same constraints as the single matrix</span>
+<a name="line-2427"></a><span class='hs-comment'>-- SelfAdjointEig.</span>
+<a name="line-2428"></a><span class='hs-comment'>-- </span>
+<a name="line-2429"></a><span class='hs-comment'>-- The result is a [..., M+1, M] matrix with [..., 0,:] containing the</span>
+<a name="line-2430"></a><span class='hs-comment'>-- eigenvalues, and subsequent [...,1:, :] containing the eigenvectors.</span>
+<a name="line-2431"></a><span class='hs-definition'>selfAdjointEig</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2432"></a>                  <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Shape is `[..., M, M]`.</span>
+<a name="line-2433"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Shape is `[..., M+1, M]`.</span>
+<a name="line-2434"></a><span class='hs-definition'>selfAdjointEig</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2435"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SelfAdjointEig"</span>
+<a name="line-2436"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-2437"></a>        <span class='hs-varid'>input</span>
+<a name="line-2438"></a><span class='hs-comment'>{-
+<a name="line-2439"></a>attr {
+<a name="line-2440"></a>  allowed_values { list { type: DT_DOUBLE type: DT_FLOAT } }
+<a name="line-2441"></a>  name: "T"
+<a name="line-2442"></a>  type: "type"
+<a name="line-2443"></a>}
+<a name="line-2444"></a>input_arg {
+<a name="line-2445"></a>  description: "Shape is `[..., M, M]`." name: "input" type_attr: "T"
+<a name="line-2446"></a>}
+<a name="line-2447"></a>output_arg {
+<a name="line-2448"></a>  description: "Shape is `[..., M+1, M]`."
+<a name="line-2449"></a>  name: "output"
+<a name="line-2450"></a>  type_attr: "T"
+<a name="line-2451"></a>}
+<a name="line-2452"></a>-}</span>
+<a name="line-2453"></a>
+<a name="line-2454"></a><span class='hs-comment'>-- | Update '*var' by subtracting 'alpha' * 'delta' from it.</span>
+<a name="line-2455"></a>
+<a name="line-2456"></a><a name="applyGradientDescent"></a><span class='hs-definition'>applyGradientDescent</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-2457"></a>                                             <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-2458"></a>                                                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-2459"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-2460"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-2461"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-2462"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-2463"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-2464"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-2465"></a>                                                     <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2466"></a>                        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-2467"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __alpha__: Scaling factor. Must be a scalar.</span>
+<a name="line-2468"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __delta__: The change.</span>
+<a name="line-2469"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-2470"></a><span class='hs-definition'>applyGradientDescent</span> <span class='hs-varid'>var</span> <span class='hs-varid'>alpha</span> <span class='hs-varid'>delta</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2471"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ApplyGradientDescent"</span>
+<a name="line-2472"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-2473"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>alpha</span> <span class='hs-varid'>delta</span>
+<a name="line-2474"></a><span class='hs-comment'>{-
+<a name="line-2475"></a>attr {
+<a name="line-2476"></a>  allowed_values {
+<a name="line-2477"></a>    list {
+<a name="line-2478"></a>      type: DT_FLOAT
+<a name="line-2479"></a>      type: DT_DOUBLE
+<a name="line-2480"></a>      type: DT_INT64
+<a name="line-2481"></a>      type: DT_INT32
+<a name="line-2482"></a>      type: DT_UINT8
+<a name="line-2483"></a>      type: DT_UINT16
+<a name="line-2484"></a>      type: DT_INT16
+<a name="line-2485"></a>      type: DT_INT8
+<a name="line-2486"></a>      type: DT_COMPLEX64
+<a name="line-2487"></a>      type: DT_COMPLEX128
+<a name="line-2488"></a>      type: DT_QINT8
+<a name="line-2489"></a>      type: DT_QUINT8
+<a name="line-2490"></a>      type: DT_QINT32
+<a name="line-2491"></a>      type: DT_HALF
+<a name="line-2492"></a>    }
+<a name="line-2493"></a>  }
+<a name="line-2494"></a>  name: "T"
+<a name="line-2495"></a>  type: "type"
+<a name="line-2496"></a>}
+<a name="line-2497"></a>attr {
+<a name="line-2498"></a>  default_value { b: false }
+<a name="line-2499"></a>  description: "If `True`, the subtraction will be protected by a lock;\notherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-2500"></a>  name: "use_locking"
+<a name="line-2501"></a>  type: "bool"
+<a name="line-2502"></a>}
+<a name="line-2503"></a>input_arg {
+<a name="line-2504"></a>  description: "Should be from a Variable()."
+<a name="line-2505"></a>  is_ref: true
+<a name="line-2506"></a>  name: "var"
+<a name="line-2507"></a>  type_attr: "T"
+<a name="line-2508"></a>}
+<a name="line-2509"></a>input_arg {
+<a name="line-2510"></a>  description: "Scaling factor. Must be a scalar."
+<a name="line-2511"></a>  name: "alpha"
+<a name="line-2512"></a>  type_attr: "T"
+<a name="line-2513"></a>}
+<a name="line-2514"></a>input_arg {
+<a name="line-2515"></a>  description: "The change." name: "delta" type_attr: "T"
+<a name="line-2516"></a>}
+<a name="line-2517"></a>output_arg {
+<a name="line-2518"></a>  description: "Same as \"var\"."
+<a name="line-2519"></a>  is_ref: true
+<a name="line-2520"></a>  name: "out"
+<a name="line-2521"></a>  type_attr: "T"
+<a name="line-2522"></a>}
+<a name="line-2523"></a>-}</span>
+<a name="line-2524"></a>
+<a name="line-2525"></a><span class='hs-comment'>-- | Push an element onto the stack.</span>
+<a name="line-2526"></a>
+<a name="line-2527"></a><a name="stackPush"></a><span class='hs-definition'>stackPush</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2528"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a stack.</span>
+<a name="line-2529"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __elem__: The tensor to be pushed onto the stack.</span>
+<a name="line-2530"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The same tensor as the input 'elem'.</span>
+<a name="line-2531"></a><span class='hs-definition'>stackPush</span> <span class='hs-varid'>handle</span> <span class='hs-varid'>elem</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2532"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"StackPush"</span>
+<a name="line-2533"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-2534"></a>        <span class='hs-varid'>handle</span> <span class='hs-varid'>elem</span>
+<a name="line-2535"></a><span class='hs-comment'>{-
+<a name="line-2536"></a>attr { name: "T" type: "type" }
+<a name="line-2537"></a>attr {
+<a name="line-2538"></a>  default_value { b: false }
+<a name="line-2539"></a>  description: "Swap `elem` to CPU. Default to false."
+<a name="line-2540"></a>  name: "swap_memory"
+<a name="line-2541"></a>  type: "bool"
+<a name="line-2542"></a>}
+<a name="line-2543"></a>input_arg {
+<a name="line-2544"></a>  description: "The handle to a stack."
+<a name="line-2545"></a>  is_ref: true
+<a name="line-2546"></a>  name: "handle"
+<a name="line-2547"></a>  type: DT_STRING
+<a name="line-2548"></a>}
+<a name="line-2549"></a>input_arg {
+<a name="line-2550"></a>  description: "The tensor to be pushed onto the stack."
+<a name="line-2551"></a>  name: "elem"
+<a name="line-2552"></a>  type_attr: "T"
+<a name="line-2553"></a>}
+<a name="line-2554"></a>output_arg {
+<a name="line-2555"></a>  description: "The same tensor as the input \'elem\'."
+<a name="line-2556"></a>  name: "output"
+<a name="line-2557"></a>  type_attr: "T"
+<a name="line-2558"></a>}
+<a name="line-2559"></a>-}</span>
+<a name="line-2560"></a>
+<a name="line-2561"></a><a name="cholesky"></a><span class='hs-comment'>-- | Computes the Cholesky decomposition of one or more square matrices.</span>
+<a name="line-2562"></a><span class='hs-comment'>--</span>
+<a name="line-2563"></a><span class='hs-comment'>-- The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions</span>
+<a name="line-2564"></a><span class='hs-comment'>-- form square matrices, with the same constraints as the single matrix Cholesky</span>
+<a name="line-2565"></a><span class='hs-comment'>-- decomposition above. The output is a tensor of the same shape as the input</span>
+<a name="line-2566"></a><span class='hs-comment'>-- containing the Cholesky decompositions for all input submatrices `[..., :, :]`.</span>
+<a name="line-2567"></a><span class='hs-definition'>cholesky</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2568"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Shape is `[..., M, M]`.</span>
+<a name="line-2569"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Shape is `[..., M, M]`.</span>
+<a name="line-2570"></a><span class='hs-definition'>cholesky</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2571"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Cholesky"</span>
+<a name="line-2572"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-2573"></a>        <span class='hs-varid'>input</span>
+<a name="line-2574"></a><span class='hs-comment'>{-
+<a name="line-2575"></a>attr {
+<a name="line-2576"></a>  allowed_values { list { type: DT_DOUBLE type: DT_FLOAT } }
+<a name="line-2577"></a>  name: "T"
+<a name="line-2578"></a>  type: "type"
+<a name="line-2579"></a>}
+<a name="line-2580"></a>input_arg {
+<a name="line-2581"></a>  description: "Shape is `[..., M, M]`." name: "input" type_attr: "T"
+<a name="line-2582"></a>}
+<a name="line-2583"></a>output_arg {
+<a name="line-2584"></a>  description: "Shape is `[..., M, M]`."
+<a name="line-2585"></a>  name: "output"
+<a name="line-2586"></a>  type_attr: "T"
+<a name="line-2587"></a>}
+<a name="line-2588"></a>-}</span>
+<a name="line-2589"></a>
+<a name="line-2590"></a><a name="dynamicStitch"></a><span class='hs-comment'>-- | Interleave the values from the `data` tensors into a single tensor.</span>
+<a name="line-2591"></a><span class='hs-comment'>--</span>
+<a name="line-2592"></a><span class='hs-comment'>-- Builds a merged tensor such that</span>
+<a name="line-2593"></a><span class='hs-comment'>-- </span>
+<a name="line-2594"></a><span class='hs-comment'>--     merged[indices[m][i, ..., j], ...] = data[m][i, ..., j, ...]</span>
+<a name="line-2595"></a><span class='hs-comment'>-- </span>
+<a name="line-2596"></a><span class='hs-comment'>-- For example, if each `indices[m]` is scalar or vector, we have</span>
+<a name="line-2597"></a><span class='hs-comment'>-- </span>
+<a name="line-2598"></a><span class='hs-comment'>--     # Scalar indices</span>
+<a name="line-2599"></a><span class='hs-comment'>--     merged[indices[m], ...] = data[m][...]</span>
+<a name="line-2600"></a><span class='hs-comment'>-- </span>
+<a name="line-2601"></a><span class='hs-comment'>--     # Vector indices</span>
+<a name="line-2602"></a><span class='hs-comment'>--     merged[indices[m][i], ...] = data[m][i, ...]</span>
+<a name="line-2603"></a><span class='hs-comment'>-- </span>
+<a name="line-2604"></a><span class='hs-comment'>-- Each `data[i].shape` must start with the corresponding `indices[i].shape`,</span>
+<a name="line-2605"></a><span class='hs-comment'>-- and the rest of `data[i].shape` must be constant w.r.t. `i`.  That is, we</span>
+<a name="line-2606"></a><span class='hs-comment'>-- must have `data[i].shape = indices[i].shape + constant`.  In terms of this</span>
+<a name="line-2607"></a><span class='hs-comment'>-- `constant`, the output shape is</span>
+<a name="line-2608"></a><span class='hs-comment'>-- </span>
+<a name="line-2609"></a><span class='hs-comment'>--     merged.shape = [max(indices)] + constant</span>
+<a name="line-2610"></a><span class='hs-comment'>-- </span>
+<a name="line-2611"></a><span class='hs-comment'>-- Values are merged in order, so if an index appears in both `indices[m][i]` and</span>
+<a name="line-2612"></a><span class='hs-comment'>-- `indices[n][j]` for `(m,i) &lt; (n,j)` the slice `data[n][j]` will appear in the</span>
+<a name="line-2613"></a><span class='hs-comment'>-- merged result.</span>
+<a name="line-2614"></a><span class='hs-comment'>-- </span>
+<a name="line-2615"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-2616"></a><span class='hs-comment'>-- </span>
+<a name="line-2617"></a><span class='hs-comment'>--     indices[0] = 6</span>
+<a name="line-2618"></a><span class='hs-comment'>--     indices[1] = [4, 1]</span>
+<a name="line-2619"></a><span class='hs-comment'>--     indices[2] = [[5, 2], [0, 3]]</span>
+<a name="line-2620"></a><span class='hs-comment'>--     data[0] = [61, 62]</span>
+<a name="line-2621"></a><span class='hs-comment'>--     data[1] = [[41, 42], [11, 12]]</span>
+<a name="line-2622"></a><span class='hs-comment'>--     data[2] = [[[51, 52], [21, 22]], [[1, 2], [31, 32]]]</span>
+<a name="line-2623"></a><span class='hs-comment'>--     merged = [[1, 2], [11, 12], [21, 22], [31, 32], [41, 42],</span>
+<a name="line-2624"></a><span class='hs-comment'>--               [51, 52], [61, 62]]</span>
+<a name="line-2625"></a><span class='hs-comment'>-- </span>
+<a name="line-2626"></a><span class='hs-comment'>-- &lt;div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;"&gt;</span>
+<a name="line-2627"></a><span class='hs-comment'>-- &lt;img style="width:100%" src="../../images/DynamicStitch.png" alt&gt;</span>
+<a name="line-2628"></a><span class='hs-comment'>-- &lt;/div&gt;</span>
+<a name="line-2629"></a><span class='hs-definition'>dynamicStitch</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2630"></a>                 <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __indices__</span>
+<a name="line-2631"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __data__</span>
+<a name="line-2632"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __merged__</span>
+<a name="line-2633"></a><span class='hs-definition'>dynamicStitch</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>data'</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"N"</span><span class='hs-layout'>,</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"data'"</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>data'</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-2634"></a>                                                    <span class='hs-layout'>(</span><span class='hs-str'>"indices"</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>indices</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2635"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DynamicStitch"</span>
+<a name="line-2636"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-2637"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"N"</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>data'</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-2638"></a>        <span class='hs-varid'>indices</span> <span class='hs-varid'>data'</span>
+<a name="line-2639"></a><span class='hs-comment'>{-
+<a name="line-2640"></a>attr { has_minimum: true minimum: 1 name: "N" type: "int" }
+<a name="line-2641"></a>attr { name: "T" type: "type" }
+<a name="line-2642"></a>input_arg { name: "indices" number_attr: "N" type: DT_INT32 }
+<a name="line-2643"></a>input_arg { name: "data" number_attr: "N" type_attr: "T" }
+<a name="line-2644"></a>output_arg { name: "merged" type_attr: "T" }
+<a name="line-2645"></a>-}</span>
+<a name="line-2646"></a>
+<a name="line-2647"></a><span class='hs-comment'>-- | Returns the number of work units this Reader has finished processing.</span>
+<a name="line-2648"></a>
+<a name="line-2649"></a><a name="readerNumWorkUnitsCompleted"></a><span class='hs-definition'>readerNumWorkUnitsCompleted</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __reader_handle__: Handle to a Reader.</span>
+<a name="line-2650"></a>                               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __units_completed__</span>
+<a name="line-2651"></a><span class='hs-definition'>readerNumWorkUnitsCompleted</span> <span class='hs-varid'>reader_handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2652"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ReaderNumWorkUnitsCompleted"</span><span class='hs-layout'>)</span>
+<a name="line-2653"></a>        <span class='hs-varid'>reader_handle</span>
+<a name="line-2654"></a><span class='hs-comment'>{-
+<a name="line-2655"></a>input_arg {
+<a name="line-2656"></a>  description: "Handle to a Reader."
+<a name="line-2657"></a>  is_ref: true
+<a name="line-2658"></a>  name: "reader_handle"
+<a name="line-2659"></a>  type: DT_STRING
+<a name="line-2660"></a>}
+<a name="line-2661"></a>output_arg { name: "units_completed" type: DT_INT64 }
+<a name="line-2662"></a>-}</span>
+<a name="line-2663"></a>
+<a name="line-2664"></a><a name="readerRead"></a><span class='hs-comment'>-- | Returns the next record (key, value pair) produced by a Reader.</span>
+<a name="line-2665"></a><span class='hs-comment'>--</span>
+<a name="line-2666"></a><span class='hs-comment'>-- Will dequeue from the input queue if necessary (e.g. when the</span>
+<a name="line-2667"></a><span class='hs-comment'>-- Reader needs to start reading from a new file since it has finished</span>
+<a name="line-2668"></a><span class='hs-comment'>-- with the previous file).</span>
+<a name="line-2669"></a><span class='hs-definition'>readerRead</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __reader_handle__: Handle to a Reader.</span>
+<a name="line-2670"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __queue_handle__: Handle to a Queue, with string work items.</span>
+<a name="line-2671"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>,</span>
+<a name="line-2672"></a>                  <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>
+<a name="line-2673"></a>              <span class='hs-comment'>-- ^ (__key__, __value__)</span>
+<a name="line-2674"></a>              <span class='hs-comment'>--</span>
+<a name="line-2675"></a>              <span class='hs-comment'>-- * __key__: A scalar.</span>
+<a name="line-2676"></a>              <span class='hs-comment'>--</span>
+<a name="line-2677"></a>              <span class='hs-comment'>-- * __value__: A scalar.</span>
+<a name="line-2678"></a><span class='hs-definition'>readerRead</span> <span class='hs-varid'>reader_handle</span> <span class='hs-varid'>queue_handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2679"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ReaderRead"</span><span class='hs-layout'>)</span>
+<a name="line-2680"></a>        <span class='hs-varid'>reader_handle</span> <span class='hs-varid'>queue_handle</span>
+<a name="line-2681"></a><span class='hs-comment'>{-
+<a name="line-2682"></a>input_arg {
+<a name="line-2683"></a>  description: "Handle to a Reader."
+<a name="line-2684"></a>  is_ref: true
+<a name="line-2685"></a>  name: "reader_handle"
+<a name="line-2686"></a>  type: DT_STRING
+<a name="line-2687"></a>}
+<a name="line-2688"></a>input_arg {
+<a name="line-2689"></a>  description: "Handle to a Queue, with string work items."
+<a name="line-2690"></a>  is_ref: true
+<a name="line-2691"></a>  name: "queue_handle"
+<a name="line-2692"></a>  type: DT_STRING
+<a name="line-2693"></a>}
+<a name="line-2694"></a>output_arg { description: "A scalar." name: "key" type: DT_STRING }
+<a name="line-2695"></a>output_arg {
+<a name="line-2696"></a>  description: "A scalar." name: "value" type: DT_STRING
+<a name="line-2697"></a>}
+<a name="line-2698"></a>-}</span>
+<a name="line-2699"></a>
+<a name="line-2700"></a><a name="fFT2D"></a><span class='hs-comment'>-- | Compute the 2-dimensional discrete Fourier Transform over the inner-most</span>
+<a name="line-2701"></a><span class='hs-comment'>--</span>
+<a name="line-2702"></a><span class='hs-comment'>-- 2 dimensions of `input`.</span>
+<a name="line-2703"></a><span class='hs-definition'>fFT2D</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __input__: A complex64 tensor.</span>
+<a name="line-2704"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __output__: A complex64 tensor of the same shape as `input`. The inner-most 2</span>
+<a name="line-2705"></a>         <span class='hs-comment'>-- dimensions of `input` are replaced with their 2D Fourier Transform.</span>
+<a name="line-2706"></a><span class='hs-definition'>fFT2D</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2707"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"FFT2D"</span><span class='hs-layout'>)</span>
+<a name="line-2708"></a>        <span class='hs-varid'>input</span>
+<a name="line-2709"></a><span class='hs-comment'>{-
+<a name="line-2710"></a>input_arg {
+<a name="line-2711"></a>  description: "A complex64 tensor." name: "input" type: DT_COMPLEX64
+<a name="line-2712"></a>}
+<a name="line-2713"></a>output_arg {
+<a name="line-2714"></a>  description: "A complex64 tensor of the same shape as `input`. The inner-most 2\ndimensions of `input` are replaced with their 2D Fourier Transform."
+<a name="line-2715"></a>  name: "output"
+<a name="line-2716"></a>  type: DT_COMPLEX64
+<a name="line-2717"></a>}
+<a name="line-2718"></a>-}</span>
+<a name="line-2719"></a>
+<a name="line-2720"></a><span class='hs-comment'>-- | A Reader that outputs fixed-length records from a file.</span>
+<a name="line-2721"></a>
+<a name="line-2722"></a><a name="fixedLengthRecordReader"></a><span class='hs-definition'>fixedLengthRecordReader</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __record_bytes__</span>
+<a name="line-2723"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __reader_handle__: The handle to reference the Reader.</span>
+<a name="line-2724"></a><span class='hs-definition'>fixedLengthRecordReader</span> <span class='hs-varid'>record_bytes</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2725"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"FixedLengthRecordReader"</span>
+<a name="line-2726"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"record_bytes"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>record_bytes</span><span class='hs-layout'>)</span>
+<a name="line-2727"></a>        
+<a name="line-2728"></a><span class='hs-comment'>{-
+<a name="line-2729"></a>attr { default_value { i: 0 } name: "header_bytes" type: "int" }
+<a name="line-2730"></a>attr { name: "record_bytes" type: "int" }
+<a name="line-2731"></a>attr { default_value { i: 0 } name: "footer_bytes" type: "int" }
+<a name="line-2732"></a>attr {
+<a name="line-2733"></a>  default_value { s: "" }
+<a name="line-2734"></a>  description: "If non-empty, this reader is placed in the given container.\nOtherwise, a default container is used."
+<a name="line-2735"></a>  name: "container"
+<a name="line-2736"></a>  type: "string"
+<a name="line-2737"></a>}
+<a name="line-2738"></a>attr {
+<a name="line-2739"></a>  default_value { s: "" }
+<a name="line-2740"></a>  description: "If non-empty, this reader is named in the given bucket\nwith this shared_name. Otherwise, the node name is used instead."
+<a name="line-2741"></a>  name: "shared_name"
+<a name="line-2742"></a>  type: "string"
+<a name="line-2743"></a>}
+<a name="line-2744"></a>output_arg {
+<a name="line-2745"></a>  description: "The handle to reference the Reader."
+<a name="line-2746"></a>  is_ref: true
+<a name="line-2747"></a>  name: "reader_handle"
+<a name="line-2748"></a>  type: DT_STRING
+<a name="line-2749"></a>}
+<a name="line-2750"></a>-}</span>
+<a name="line-2751"></a>
+<a name="line-2752"></a><a name="placeholder"></a><span class='hs-comment'>-- | A placeholder op for a value that will be fed into the computation.</span>
+<a name="line-2753"></a><span class='hs-comment'>--</span>
+<a name="line-2754"></a><span class='hs-comment'>-- N.B. This operation will fail with an error if it is executed. It is</span>
+<a name="line-2755"></a><span class='hs-comment'>-- intended as a way to represent a value that will always be fed, and to</span>
+<a name="line-2756"></a><span class='hs-comment'>-- provide attrs that enable the fed value to be checked at runtime.</span>
+<a name="line-2757"></a><span class='hs-definition'>placeholder</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2758"></a>               <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __output__: A placeholder tensor that must be replaced using the feed mechanism.</span>
+<a name="line-2759"></a><span class='hs-definition'>placeholder</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2760"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Placeholder"</span>
+<a name="line-2761"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-2762"></a>        
+<a name="line-2763"></a><span class='hs-comment'>{-
+<a name="line-2764"></a>attr {
+<a name="line-2765"></a>  description: "The type of elements in the tensor."
+<a name="line-2766"></a>  name: "dtype"
+<a name="line-2767"></a>  type: "type"
+<a name="line-2768"></a>}
+<a name="line-2769"></a>attr {
+<a name="line-2770"></a>  default_value { shape { } }
+<a name="line-2771"></a>  description: "(Optional) The shape of the tensor. If the shape has 0 dimensions, the\nshape is unconstrained."
+<a name="line-2772"></a>  name: "shape"
+<a name="line-2773"></a>  type: "shape"
+<a name="line-2774"></a>}
+<a name="line-2775"></a>output_arg {
+<a name="line-2776"></a>  description: "A placeholder tensor that must be replaced using the feed mechanism."
+<a name="line-2777"></a>  name: "output"
+<a name="line-2778"></a>  type_attr: "dtype"
+<a name="line-2779"></a>}
+<a name="line-2780"></a>-}</span>
+<a name="line-2781"></a>
+<a name="line-2782"></a><a name="scalarSummary"></a><span class='hs-comment'>-- | Outputs a `Summary` protocol buffer with scalar values.</span>
+<a name="line-2783"></a><span class='hs-comment'>--</span>
+<a name="line-2784"></a><span class='hs-comment'>-- The input `tags` and `values` must have the same shape.  The generated summary</span>
+<a name="line-2785"></a><span class='hs-comment'>-- has a summary value for each tag-value pair in `tags` and `values`.</span>
+<a name="line-2786"></a><span class='hs-definition'>scalarSummary</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-2787"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-2788"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-2789"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-2790"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-2791"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-2792"></a>                                                         <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2793"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __tags__: Tags for the summary.</span>
+<a name="line-2794"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __values__: Same shape as `tags.  Values for the summary.</span>
+<a name="line-2795"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __summary__: Scalar.  Serialized `Summary` protocol buffer.</span>
+<a name="line-2796"></a><span class='hs-definition'>scalarSummary</span> <span class='hs-varid'>tags</span> <span class='hs-varid'>values</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2797"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ScalarSummary"</span>
+<a name="line-2798"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-2799"></a>        <span class='hs-varid'>tags</span> <span class='hs-varid'>values</span>
+<a name="line-2800"></a><span class='hs-comment'>{-
+<a name="line-2801"></a>attr {
+<a name="line-2802"></a>  allowed_values {
+<a name="line-2803"></a>    list {
+<a name="line-2804"></a>      type: DT_FLOAT
+<a name="line-2805"></a>      type: DT_DOUBLE
+<a name="line-2806"></a>      type: DT_INT32
+<a name="line-2807"></a>      type: DT_INT64
+<a name="line-2808"></a>      type: DT_UINT8
+<a name="line-2809"></a>      type: DT_INT16
+<a name="line-2810"></a>      type: DT_INT8
+<a name="line-2811"></a>      type: DT_UINT16
+<a name="line-2812"></a>      type: DT_HALF
+<a name="line-2813"></a>    }
+<a name="line-2814"></a>  }
+<a name="line-2815"></a>  name: "T"
+<a name="line-2816"></a>  type: "type"
+<a name="line-2817"></a>}
+<a name="line-2818"></a>input_arg {
+<a name="line-2819"></a>  description: "Tags for the summary." name: "tags" type: DT_STRING
+<a name="line-2820"></a>}
+<a name="line-2821"></a>input_arg {
+<a name="line-2822"></a>  description: "Same shape as `tags.  Values for the summary."
+<a name="line-2823"></a>  name: "values"
+<a name="line-2824"></a>  type_attr: "T"
+<a name="line-2825"></a>}
+<a name="line-2826"></a>output_arg {
+<a name="line-2827"></a>  description: "Scalar.  Serialized `Summary` protocol buffer."
+<a name="line-2828"></a>  name: "summary"
+<a name="line-2829"></a>  type: DT_STRING
+<a name="line-2830"></a>}
+<a name="line-2831"></a>-}</span>
+<a name="line-2832"></a>
+<a name="line-2833"></a><a name="softmax"></a><span class='hs-comment'>-- | Computes softmax activations.</span>
+<a name="line-2834"></a><span class='hs-comment'>--</span>
+<a name="line-2835"></a><span class='hs-comment'>-- For each batch `i` and class `j` we have</span>
+<a name="line-2836"></a><span class='hs-comment'>-- </span>
+<a name="line-2837"></a><span class='hs-comment'>--     softmax[i, j] = exp(logits[i, j]) / sum_j(exp(logits[i, j]))</span>
+<a name="line-2838"></a><span class='hs-definition'>softmax</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-2839"></a>                                                <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2840"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __logits__: 2-D with shape `[batch_size, num_classes]`.</span>
+<a name="line-2841"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __softmax__: Same shape as `logits`.</span>
+<a name="line-2842"></a><span class='hs-definition'>softmax</span> <span class='hs-varid'>logits</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2843"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Softmax"</span>
+<a name="line-2844"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-2845"></a>        <span class='hs-varid'>logits</span>
+<a name="line-2846"></a><span class='hs-comment'>{-
+<a name="line-2847"></a>attr {
+<a name="line-2848"></a>  allowed_values {
+<a name="line-2849"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-2850"></a>  }
+<a name="line-2851"></a>  name: "T"
+<a name="line-2852"></a>  type: "type"
+<a name="line-2853"></a>}
+<a name="line-2854"></a>input_arg {
+<a name="line-2855"></a>  description: "2-D with shape `[batch_size, num_classes]`."
+<a name="line-2856"></a>  name: "logits"
+<a name="line-2857"></a>  type_attr: "T"
+<a name="line-2858"></a>}
+<a name="line-2859"></a>output_arg {
+<a name="line-2860"></a>  description: "Same shape as `logits`."
+<a name="line-2861"></a>  name: "softmax"
+<a name="line-2862"></a>  type_attr: "T"
+<a name="line-2863"></a>}
+<a name="line-2864"></a>-}</span>
+<a name="line-2865"></a>
+<a name="line-2866"></a><a name="shardedFilename"></a><span class='hs-comment'>-- | Generate a sharded filename. The filename is printf formatted as</span>
+<a name="line-2867"></a><span class='hs-comment'>--</span>
+<a name="line-2868"></a><span class='hs-comment'>--    %s-%05d-of-%05d, basename, shard, num_shards.</span>
+<a name="line-2869"></a><span class='hs-definition'>shardedFilename</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __basename__</span>
+<a name="line-2870"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __shard__</span>
+<a name="line-2871"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __num_shards__</span>
+<a name="line-2872"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __filename__</span>
+<a name="line-2873"></a><span class='hs-definition'>shardedFilename</span> <span class='hs-varid'>basename</span> <span class='hs-varid'>shard</span> <span class='hs-varid'>num_shards</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2874"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ShardedFilename"</span><span class='hs-layout'>)</span>
+<a name="line-2875"></a>        <span class='hs-varid'>basename</span> <span class='hs-varid'>shard</span> <span class='hs-varid'>num_shards</span>
+<a name="line-2876"></a><span class='hs-comment'>{-
+<a name="line-2877"></a>input_arg { name: "basename" type: DT_STRING }
+<a name="line-2878"></a>input_arg { name: "shard" type: DT_INT32 }
+<a name="line-2879"></a>input_arg { name: "num_shards" type: DT_INT32 }
+<a name="line-2880"></a>output_arg { name: "filename" type: DT_STRING }
+<a name="line-2881"></a>-}</span>
+<a name="line-2882"></a>
+<a name="line-2883"></a><span class='hs-comment'>-- | Sends the named tensor from send_device to recv_device.</span>
+<a name="line-2884"></a><span class='hs-comment'>--</span>
+<a name="line-2885"></a><span class='hs-comment'>-- _HostSend requires its input on host memory whereas _Send requires its</span>
+<a name="line-2886"></a><span class='hs-comment'>-- input on device memory.</span>
+<a name="line-2887"></a><span class='hs-sel'>_HostSend</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2888"></a>             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __send_device_incarnation__: The current incarnation of send_device.</span>
+<a name="line-2889"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __tensor__: The tensor to send.</span>
+<a name="line-2890"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-2891"></a><span class='hs-sel'>_HostSend</span> <span class='hs-varid'>send_device_incarnation</span> <span class='hs-varid'>tensor</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2892"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"_HostSend"</span>
+<a name="line-2893"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-2894"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"send_device_incarnation"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>send_device_incarnation</span><span class='hs-layout'>)</span>
+<a name="line-2895"></a>        <span class='hs-varid'>tensor</span>
+<a name="line-2896"></a><span class='hs-comment'>{-
+<a name="line-2897"></a>attr { name: "T" type: "type" }
+<a name="line-2898"></a>attr {
+<a name="line-2899"></a>  description: "The name of the tensor to send."
+<a name="line-2900"></a>  name: "tensor_name"
+<a name="line-2901"></a>  type: "string"
+<a name="line-2902"></a>}
+<a name="line-2903"></a>attr {
+<a name="line-2904"></a>  description: "The name of the device sending the tensor."
+<a name="line-2905"></a>  name: "send_device"
+<a name="line-2906"></a>  type: "string"
+<a name="line-2907"></a>}
+<a name="line-2908"></a>attr {
+<a name="line-2909"></a>  description: "The current incarnation of send_device."
+<a name="line-2910"></a>  name: "send_device_incarnation"
+<a name="line-2911"></a>  type: "int"
+<a name="line-2912"></a>}
+<a name="line-2913"></a>attr {
+<a name="line-2914"></a>  description: "The name of the device receiving the tensor."
+<a name="line-2915"></a>  name: "recv_device"
+<a name="line-2916"></a>  type: "string"
+<a name="line-2917"></a>}
+<a name="line-2918"></a>attr {
+<a name="line-2919"></a>  default_value { b: false }
+<a name="line-2920"></a>  description: "If set to true, this indicates that the node was added\nto the graph as a result of a client-side feed or fetch of Tensor data,\nin which case the corresponding send or recv is expected to be managed\nlocally by the caller."
+<a name="line-2921"></a>  name: "client_terminated"
+<a name="line-2922"></a>  type: "bool"
+<a name="line-2923"></a>}
+<a name="line-2924"></a>input_arg {
+<a name="line-2925"></a>  description: "The tensor to send." name: "tensor" type_attr: "T"
+<a name="line-2926"></a>}
+<a name="line-2927"></a>-}</span>
+<a name="line-2928"></a>
+<a name="line-2929"></a><a name="sigmoidGrad"></a><span class='hs-comment'>-- | Computes the gradient of the sigmoid of `x` wrt its input.</span>
+<a name="line-2930"></a><span class='hs-comment'>--</span>
+<a name="line-2931"></a><span class='hs-comment'>-- Specifically, `grad = dy * y * (1 - y)`, where `y = sigmoid(x)`, and</span>
+<a name="line-2932"></a><span class='hs-comment'>-- `dy` is the corresponding input gradient.</span>
+<a name="line-2933"></a><span class='hs-definition'>sigmoidGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-2934"></a>                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-2935"></a>                                         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-2936"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-2937"></a>               <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-2938"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-2939"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-2940"></a><span class='hs-definition'>sigmoidGrad</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2941"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SigmoidGrad"</span>
+<a name="line-2942"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-2943"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-2944"></a><span class='hs-comment'>{-
+<a name="line-2945"></a>attr {
+<a name="line-2946"></a>  allowed_values {
+<a name="line-2947"></a>    list {
+<a name="line-2948"></a>      type: DT_HALF
+<a name="line-2949"></a>      type: DT_FLOAT
+<a name="line-2950"></a>      type: DT_DOUBLE
+<a name="line-2951"></a>      type: DT_COMPLEX64
+<a name="line-2952"></a>      type: DT_COMPLEX128
+<a name="line-2953"></a>    }
+<a name="line-2954"></a>  }
+<a name="line-2955"></a>  name: "T"
+<a name="line-2956"></a>  type: "type"
+<a name="line-2957"></a>}
+<a name="line-2958"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-2959"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-2960"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-2961"></a>-}</span>
+<a name="line-2962"></a>
+<a name="line-2963"></a><a name="nonMaxSuppression"></a><span class='hs-comment'>-- | Greedily selects a subset of bounding boxes in descending order of score,</span>
+<a name="line-2964"></a><span class='hs-comment'>--</span>
+<a name="line-2965"></a><span class='hs-comment'>-- pruning away boxes that have high intersection-over-union (IOU) overlap</span>
+<a name="line-2966"></a><span class='hs-comment'>-- with previously selected boxes.  Bounding boxes are supplied as</span>
+<a name="line-2967"></a><span class='hs-comment'>-- [y1, x1, y2, x2], where (y1, x1) and (y2, x2) are the coordinates of any</span>
+<a name="line-2968"></a><span class='hs-comment'>-- diagonal pair of box corners and the coordinates can be provided as normalized</span>
+<a name="line-2969"></a><span class='hs-comment'>-- (i.e., lying in the interval [0, 1]) or absolute.  Note that this algorithm</span>
+<a name="line-2970"></a><span class='hs-comment'>-- is agnostic to where the origin is in the coordinate system.  Note that this</span>
+<a name="line-2971"></a><span class='hs-comment'>-- algorithm is invariant to orthogonal transformations and translations</span>
+<a name="line-2972"></a><span class='hs-comment'>-- of the coordinate system; thus translating or reflections of the coordinate</span>
+<a name="line-2973"></a><span class='hs-comment'>-- system result in the same boxes being selected by the algorithm.</span>
+<a name="line-2974"></a><span class='hs-comment'>-- </span>
+<a name="line-2975"></a><span class='hs-comment'>-- The output of this operation is a set of integers indexing into the input</span>
+<a name="line-2976"></a><span class='hs-comment'>-- collection of bounding boxes representing the selected boxes.  The bounding</span>
+<a name="line-2977"></a><span class='hs-comment'>-- box coordinates corresponding to the selected indices can then be obtained</span>
+<a name="line-2978"></a><span class='hs-comment'>-- using the tf.gather operation.  For example:</span>
+<a name="line-2979"></a><span class='hs-comment'>-- </span>
+<a name="line-2980"></a><span class='hs-comment'>--   selected_indices = tf.image.non_max_suppression(</span>
+<a name="line-2981"></a><span class='hs-comment'>--       boxes, scores, max_output_size, iou_threshold)</span>
+<a name="line-2982"></a><span class='hs-comment'>--   selected_boxes = tf.gather(boxes, selected_indices)</span>
+<a name="line-2983"></a><span class='hs-definition'>nonMaxSuppression</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __boxes__: A 2-D float tensor of shape `[num_boxes, 4]`.</span>
+<a name="line-2984"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __scores__: A 1-D float tensor of shape `[num_boxes]` representing a single</span>
+<a name="line-2985"></a>                                        <span class='hs-comment'>-- score corresponding to each box (each row of boxes).</span>
+<a name="line-2986"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __max_output_size__: A scalar integer tensor representing the maximum number of</span>
+<a name="line-2987"></a>                                                 <span class='hs-comment'>-- boxes to be selected by non max suppression.</span>
+<a name="line-2988"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __selected_indices__: A 1-D integer tensor of shape `[M]` representing the selected</span>
+<a name="line-2989"></a>                     <span class='hs-comment'>-- indices from the boxes tensor, where `M &lt;= max_output_size`.</span>
+<a name="line-2990"></a><span class='hs-definition'>nonMaxSuppression</span> <span class='hs-varid'>boxes</span> <span class='hs-varid'>scores</span> <span class='hs-varid'>max_output_size</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-2991"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"NonMaxSuppression"</span><span class='hs-layout'>)</span>
+<a name="line-2992"></a>        <span class='hs-varid'>boxes</span> <span class='hs-varid'>scores</span> <span class='hs-varid'>max_output_size</span>
+<a name="line-2993"></a><span class='hs-comment'>{-
+<a name="line-2994"></a>attr {
+<a name="line-2995"></a>  default_value { f: 0.5 }
+<a name="line-2996"></a>  description: "A float representing the threshold for deciding whether boxes\noverlap too much with respect to IOU."
+<a name="line-2997"></a>  name: "iou_threshold"
+<a name="line-2998"></a>  type: "float"
+<a name="line-2999"></a>}
+<a name="line-3000"></a>input_arg {
+<a name="line-3001"></a>  description: "A 2-D float tensor of shape `[num_boxes, 4]`."
+<a name="line-3002"></a>  name: "boxes"
+<a name="line-3003"></a>  type: DT_FLOAT
+<a name="line-3004"></a>}
+<a name="line-3005"></a>input_arg {
+<a name="line-3006"></a>  description: "A 1-D float tensor of shape `[num_boxes]` representing a single\nscore corresponding to each box (each row of boxes)."
+<a name="line-3007"></a>  name: "scores"
+<a name="line-3008"></a>  type: DT_FLOAT
+<a name="line-3009"></a>}
+<a name="line-3010"></a>input_arg {
+<a name="line-3011"></a>  description: "A scalar integer tensor representing the maximum number of\nboxes to be selected by non max suppression."
+<a name="line-3012"></a>  name: "max_output_size"
+<a name="line-3013"></a>  type: DT_INT32
+<a name="line-3014"></a>}
+<a name="line-3015"></a>output_arg {
+<a name="line-3016"></a>  description: "A 1-D integer tensor of shape `[M]` representing the selected\nindices from the boxes tensor, where `M &lt;= max_output_size`."
+<a name="line-3017"></a>  name: "selected_indices"
+<a name="line-3018"></a>  type: DT_INT32
+<a name="line-3019"></a>}
+<a name="line-3020"></a>-}</span>
+<a name="line-3021"></a>
+<a name="line-3022"></a><a name="identityReader"></a><span class='hs-comment'>-- | A Reader that outputs the queued work as both the key and value.</span>
+<a name="line-3023"></a><span class='hs-comment'>--</span>
+<a name="line-3024"></a><span class='hs-comment'>-- To use, enqueue strings in a Queue.  ReaderRead will take the front</span>
+<a name="line-3025"></a><span class='hs-comment'>-- work string and output (work, work).</span>
+<a name="line-3026"></a><span class='hs-definition'>identityReader</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __reader_handle__: The handle to reference the Reader.</span>
+<a name="line-3027"></a><span class='hs-definition'>identityReader</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3028"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"IdentityReader"</span><span class='hs-layout'>)</span>
+<a name="line-3029"></a>        
+<a name="line-3030"></a><span class='hs-comment'>{-
+<a name="line-3031"></a>attr {
+<a name="line-3032"></a>  default_value { s: "" }
+<a name="line-3033"></a>  description: "If non-empty, this reader is placed in the given container.\nOtherwise, a default container is used."
+<a name="line-3034"></a>  name: "container"
+<a name="line-3035"></a>  type: "string"
+<a name="line-3036"></a>}
+<a name="line-3037"></a>attr {
+<a name="line-3038"></a>  default_value { s: "" }
+<a name="line-3039"></a>  description: "If non-empty, this reader is named in the given bucket\nwith this shared_name. Otherwise, the node name is used instead."
+<a name="line-3040"></a>  name: "shared_name"
+<a name="line-3041"></a>  type: "string"
+<a name="line-3042"></a>}
+<a name="line-3043"></a>output_arg {
+<a name="line-3044"></a>  description: "The handle to reference the Reader."
+<a name="line-3045"></a>  is_ref: true
+<a name="line-3046"></a>  name: "reader_handle"
+<a name="line-3047"></a>  type: DT_STRING
+<a name="line-3048"></a>}
+<a name="line-3049"></a>-}</span>
+<a name="line-3050"></a>
+<a name="line-3051"></a><a name="extractGlimpse"></a><span class='hs-comment'>-- | Extracts a glimpse from the input tensor.</span>
+<a name="line-3052"></a><span class='hs-comment'>--</span>
+<a name="line-3053"></a><span class='hs-comment'>-- Returns a set of windows called glimpses extracted at location</span>
+<a name="line-3054"></a><span class='hs-comment'>-- `offsets` from the input tensor. If the windows only partially</span>
+<a name="line-3055"></a><span class='hs-comment'>-- overlaps the inputs, the non overlapping areas will be filled with</span>
+<a name="line-3056"></a><span class='hs-comment'>-- random noise.</span>
+<a name="line-3057"></a><span class='hs-comment'>-- </span>
+<a name="line-3058"></a><span class='hs-comment'>-- The result is a 4-D tensor of shape `[batch_size, glimpse_height,</span>
+<a name="line-3059"></a><span class='hs-comment'>-- glimpse_width, channels]`. The channels and batch dimensions are the</span>
+<a name="line-3060"></a><span class='hs-comment'>-- same as that of the input tensor. The height and width of the output</span>
+<a name="line-3061"></a><span class='hs-comment'>-- windows are specified in the `size` parameter.</span>
+<a name="line-3062"></a><span class='hs-comment'>-- </span>
+<a name="line-3063"></a><span class='hs-comment'>-- The argument `normalized` and `centered` controls how the windows are built:</span>
+<a name="line-3064"></a><span class='hs-comment'>-- </span>
+<a name="line-3065"></a><span class='hs-comment'>-- * If the coordinates are normalized but not centered, 0.0 and 1.0</span>
+<a name="line-3066"></a><span class='hs-comment'>--   correspond to the minimum and maximum of each height and width</span>
+<a name="line-3067"></a><span class='hs-comment'>--   dimension.</span>
+<a name="line-3068"></a><span class='hs-comment'>-- * If the coordinates are both normalized and centered, they range from</span>
+<a name="line-3069"></a><span class='hs-comment'>--   -1.0 to 1.0. The coordinates (-1.0, -1.0) correspond to the upper</span>
+<a name="line-3070"></a><span class='hs-comment'>--   left corner, the lower right corner is located at (1.0, 1.0) and the</span>
+<a name="line-3071"></a><span class='hs-comment'>--   center is at (0, 0).</span>
+<a name="line-3072"></a><span class='hs-comment'>-- * If the coordinates are not normalized they are interpreted as</span>
+<a name="line-3073"></a><span class='hs-comment'>--   numbers of pixels.</span>
+<a name="line-3074"></a><span class='hs-definition'>extractGlimpse</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __input__: A 4-D float tensor of shape `[batch_size, height, width, channels]`.</span>
+<a name="line-3075"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __size__: A 1-D tensor of 2 elements containing the size of the glimpses</span>
+<a name="line-3076"></a>                                              <span class='hs-comment'>-- to extract.  The glimpse height must be specified first, following</span>
+<a name="line-3077"></a>                                              <span class='hs-comment'>-- by the glimpse width.</span>
+<a name="line-3078"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __offsets__: A 2-D integer tensor of shape `[batch_size, 2]` containing</span>
+<a name="line-3079"></a>                                     <span class='hs-comment'>-- the x, y locations of the center of each window.</span>
+<a name="line-3080"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __glimpse__: A tensor representing the glimpses `[batch_size,</span>
+<a name="line-3081"></a>                  <span class='hs-comment'>-- glimpse_height, glimpse_width, channels]`.</span>
+<a name="line-3082"></a><span class='hs-definition'>extractGlimpse</span> <span class='hs-varid'>input</span> <span class='hs-varid'>size</span> <span class='hs-varid'>offsets</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3083"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ExtractGlimpse"</span><span class='hs-layout'>)</span>
+<a name="line-3084"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>size</span> <span class='hs-varid'>offsets</span>
+<a name="line-3085"></a><span class='hs-comment'>{-
+<a name="line-3086"></a>attr {
+<a name="line-3087"></a>  default_value { b: true }
+<a name="line-3088"></a>  description: "indicates if the offset coordinates are centered relative to\nthe image, in which case the (0, 0) offset is relative to the center\nof the input images. If false, the (0,0) offset corresponds to the\nupper left corner of the input images."
+<a name="line-3089"></a>  name: "centered"
+<a name="line-3090"></a>  type: "bool"
+<a name="line-3091"></a>}
+<a name="line-3092"></a>attr {
+<a name="line-3093"></a>  default_value { b: true }
+<a name="line-3094"></a>  description: "indicates if the offset coordinates are normalized."
+<a name="line-3095"></a>  name: "normalized"
+<a name="line-3096"></a>  type: "bool"
+<a name="line-3097"></a>}
+<a name="line-3098"></a>attr {
+<a name="line-3099"></a>  default_value { b: true }
+<a name="line-3100"></a>  description: "indicates if the noise should be generated using a\nuniform distribution or a gaussian distribution."
+<a name="line-3101"></a>  name: "uniform_noise"
+<a name="line-3102"></a>  type: "bool"
+<a name="line-3103"></a>}
+<a name="line-3104"></a>input_arg {
+<a name="line-3105"></a>  description: "A 4-D float tensor of shape `[batch_size, height, width, channels]`."
+<a name="line-3106"></a>  name: "input"
+<a name="line-3107"></a>  type: DT_FLOAT
+<a name="line-3108"></a>}
+<a name="line-3109"></a>input_arg {
+<a name="line-3110"></a>  description: "A 1-D tensor of 2 elements containing the size of the glimpses\nto extract.  The glimpse height must be specified first, following\nby the glimpse width."
+<a name="line-3111"></a>  name: "size"
+<a name="line-3112"></a>  type: DT_INT32
+<a name="line-3113"></a>}
+<a name="line-3114"></a>input_arg {
+<a name="line-3115"></a>  description: "A 2-D integer tensor of shape `[batch_size, 2]` containing\nthe x, y locations of the center of each window."
+<a name="line-3116"></a>  name: "offsets"
+<a name="line-3117"></a>  type: DT_FLOAT
+<a name="line-3118"></a>}
+<a name="line-3119"></a>output_arg {
+<a name="line-3120"></a>  description: "A tensor representing the glimpses `[batch_size,\nglimpse_height, glimpse_width, channels]`."
+<a name="line-3121"></a>  name: "glimpse"
+<a name="line-3122"></a>  type: DT_FLOAT
+<a name="line-3123"></a>}
+<a name="line-3124"></a>-}</span>
+<a name="line-3125"></a>
+<a name="line-3126"></a><span class='hs-comment'>-- | Computes the gradients of 3-D convolution with respect to the input.</span>
+<a name="line-3127"></a>
+<a name="line-3128"></a><a name="conv3DBackpropInput"></a><span class='hs-definition'>conv3DBackpropInput</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-3129"></a>                                            <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-3130"></a>                                                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-3131"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-3132"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-3133"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-3134"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-3135"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-3136"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-3137"></a>                                                    <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-3138"></a>                       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Shape `[batch, depth, rows, cols, in_channels]`.</span>
+<a name="line-3139"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __filter__: Shape `[depth, rows, cols, in_channels, out_channels]`.</span>
+<a name="line-3140"></a>                                      <span class='hs-comment'>-- `in_channels` must match between `input` and `filter`.</span>
+<a name="line-3141"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out_backprop__: Backprop signal of shape `[batch, out_depth, out_rows, out_cols,</span>
+<a name="line-3142"></a>                                      <span class='hs-comment'>-- out_channels]`.</span>
+<a name="line-3143"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-3144"></a><span class='hs-definition'>conv3DBackpropInput</span> <span class='hs-varid'>input</span> <span class='hs-varid'>filter</span> <span class='hs-varid'>out_backprop</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3145"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Conv3DBackpropInput"</span>
+<a name="line-3146"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-3147"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>filter</span> <span class='hs-varid'>out_backprop</span>
+<a name="line-3148"></a><span class='hs-comment'>{-
+<a name="line-3149"></a>attr {
+<a name="line-3150"></a>  allowed_values {
+<a name="line-3151"></a>    list {
+<a name="line-3152"></a>      type: DT_FLOAT
+<a name="line-3153"></a>      type: DT_DOUBLE
+<a name="line-3154"></a>      type: DT_INT64
+<a name="line-3155"></a>      type: DT_INT32
+<a name="line-3156"></a>      type: DT_UINT8
+<a name="line-3157"></a>      type: DT_UINT16
+<a name="line-3158"></a>      type: DT_INT16
+<a name="line-3159"></a>      type: DT_INT8
+<a name="line-3160"></a>      type: DT_COMPLEX64
+<a name="line-3161"></a>      type: DT_COMPLEX128
+<a name="line-3162"></a>      type: DT_QINT8
+<a name="line-3163"></a>      type: DT_QUINT8
+<a name="line-3164"></a>      type: DT_QINT32
+<a name="line-3165"></a>      type: DT_HALF
+<a name="line-3166"></a>    }
+<a name="line-3167"></a>  }
+<a name="line-3168"></a>  name: "T"
+<a name="line-3169"></a>  type: "type"
+<a name="line-3170"></a>}
+<a name="line-3171"></a>attr {
+<a name="line-3172"></a>  description: "1-D tensor of length 5. The stride of the sliding window for each\ndimension of `input`. Must have `strides[0] = strides[4] = 1`."
+<a name="line-3173"></a>  has_minimum: true
+<a name="line-3174"></a>  minimum: 5
+<a name="line-3175"></a>  name: "strides"
+<a name="line-3176"></a>  type: "list(int)"
+<a name="line-3177"></a>}
+<a name="line-3178"></a>attr {
+<a name="line-3179"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-3180"></a>  description: "The type of padding algorithm to use."
+<a name="line-3181"></a>  name: "padding"
+<a name="line-3182"></a>  type: "string"
+<a name="line-3183"></a>}
+<a name="line-3184"></a>input_arg {
+<a name="line-3185"></a>  description: "Shape `[batch, depth, rows, cols, in_channels]`."
+<a name="line-3186"></a>  name: "input"
+<a name="line-3187"></a>  type_attr: "T"
+<a name="line-3188"></a>}
+<a name="line-3189"></a>input_arg {
+<a name="line-3190"></a>  description: "Shape `[depth, rows, cols, in_channels, out_channels]`.\n`in_channels` must match between `input` and `filter`."
+<a name="line-3191"></a>  name: "filter"
+<a name="line-3192"></a>  type_attr: "T"
+<a name="line-3193"></a>}
+<a name="line-3194"></a>input_arg {
+<a name="line-3195"></a>  description: "Backprop signal of shape `[batch, out_depth, out_rows, out_cols,\nout_channels]`."
+<a name="line-3196"></a>  name: "out_backprop"
+<a name="line-3197"></a>  type_attr: "T"
+<a name="line-3198"></a>}
+<a name="line-3199"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-3200"></a>-}</span>
+<a name="line-3201"></a>
+<a name="line-3202"></a><a name="matrixSolveLs"></a><span class='hs-comment'>-- | Solves one or more linear least-squares problems.</span>
+<a name="line-3203"></a><span class='hs-comment'>--</span>
+<a name="line-3204"></a><span class='hs-comment'>-- `matrix` is a tensor of shape `[..., M, N]` whose inner-most 2 dimensions</span>
+<a name="line-3205"></a><span class='hs-comment'>-- form matrices of size `[M, N]`. Rhs is a tensor of shape `[..., M, K]`.</span>
+<a name="line-3206"></a><span class='hs-comment'>-- The output is a tensor shape `[..., N, K]` where each output matrix solves</span>
+<a name="line-3207"></a><span class='hs-comment'>-- each of the equations matrix[..., :, :] * output[..., :, :] = rhs[..., :, :]</span>
+<a name="line-3208"></a><span class='hs-comment'>-- in the least squares sense.</span>
+<a name="line-3209"></a><span class='hs-comment'>-- </span>
+<a name="line-3210"></a><span class='hs-comment'>-- matrix and right-hand sides in the batch:</span>
+<a name="line-3211"></a><span class='hs-comment'>-- </span>
+<a name="line-3212"></a><span class='hs-comment'>-- `matrix`=\\(A \in \Re^{m \times n}\\),</span>
+<a name="line-3213"></a><span class='hs-comment'>-- `rhs`=\\(B  \in \Re^{m \times k}\\),</span>
+<a name="line-3214"></a><span class='hs-comment'>-- `output`=\\(X  \in \Re^{n \times k}\\),</span>
+<a name="line-3215"></a><span class='hs-comment'>-- `l2_regularizer`=\\(\lambda\\).</span>
+<a name="line-3216"></a><span class='hs-comment'>-- </span>
+<a name="line-3217"></a><span class='hs-comment'>-- If `fast` is `True`, then the solution is computed by solving the normal</span>
+<a name="line-3218"></a><span class='hs-comment'>-- equations using Cholesky decomposition. Specifically, if \\(m \ge n\\) then</span>
+<a name="line-3219"></a><span class='hs-comment'>-- \\(X = (A^T A + \lambda I)^{-1} A^T B\\), which solves the least-squares</span>
+<a name="line-3220"></a><span class='hs-comment'>-- problem \\(X = \mathrm{argmin}_{Z \in \Re^{n \times k}} ||A Z - B||_F^2 +</span>
+<a name="line-3221"></a><span class='hs-comment'>-- \lambda ||Z||_F^2\\). If \\(m \lt n\\) then `output` is computed as</span>
+<a name="line-3222"></a><span class='hs-comment'>-- \\(X = A^T (A A^T + \lambda I)^{-1} B\\), which (for \\(\lambda = 0\\)) is the</span>
+<a name="line-3223"></a><span class='hs-comment'>-- minimum-norm solution to the under-determined linear system, i.e.</span>
+<a name="line-3224"></a><span class='hs-comment'>-- \\(X = \mathrm{argmin}_{Z \in \Re^{n \times k}} ||Z||_F^2 \\), subject to</span>
+<a name="line-3225"></a><span class='hs-comment'>-- \\(A Z = B\\). Notice that the fast path is only numerically stable when</span>
+<a name="line-3226"></a><span class='hs-comment'>-- \\(A\\) is numerically full rank and has a condition number</span>
+<a name="line-3227"></a><span class='hs-comment'>-- \\(\mathrm{cond}(A) \lt \frac{1}{\sqrt{\epsilon_{mach}}}\\) or\\(\lambda\\) is</span>
+<a name="line-3228"></a><span class='hs-comment'>-- sufficiently large.</span>
+<a name="line-3229"></a><span class='hs-comment'>-- </span>
+<a name="line-3230"></a><span class='hs-comment'>-- If `fast` is `False` an algorithm based on the numerically robust complete</span>
+<a name="line-3231"></a><span class='hs-comment'>-- orthogonal decomposition is used. This computes the minimum-norm</span>
+<a name="line-3232"></a><span class='hs-comment'>-- least-squares solution, even when \\(A\\) is rank deficient. This path is</span>
+<a name="line-3233"></a><span class='hs-comment'>-- typically 6-7 times slower than the fast path. If `fast` is `False` then</span>
+<a name="line-3234"></a><span class='hs-comment'>-- `l2_regularizer` is ignored.</span>
+<a name="line-3235"></a><span class='hs-definition'>matrixSolveLs</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-3236"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __matrix__: Shape is `[..., M, N]`.</span>
+<a name="line-3237"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __rhs__: Shape is `[..., M, K]`.</span>
+<a name="line-3238"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Double</span> <span class='hs-comment'>-- ^ __l2_regularizer__: Scalar tensor.</span>
+<a name="line-3239"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Shape is `[..., N, K]`.</span>
+<a name="line-3240"></a><span class='hs-definition'>matrixSolveLs</span> <span class='hs-varid'>matrix</span> <span class='hs-varid'>rhs</span> <span class='hs-varid'>l2_regularizer</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3241"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MatrixSolveLs"</span>
+<a name="line-3242"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-3243"></a>        <span class='hs-varid'>matrix</span> <span class='hs-varid'>rhs</span> <span class='hs-varid'>l2_regularizer</span>
+<a name="line-3244"></a><span class='hs-comment'>{-
+<a name="line-3245"></a>attr {
+<a name="line-3246"></a>  allowed_values { list { type: DT_DOUBLE type: DT_FLOAT } }
+<a name="line-3247"></a>  name: "T"
+<a name="line-3248"></a>  type: "type"
+<a name="line-3249"></a>}
+<a name="line-3250"></a>attr { default_value { b: true } name: "fast" type: "bool" }
+<a name="line-3251"></a>input_arg {
+<a name="line-3252"></a>  description: "Shape is `[..., M, N]`."
+<a name="line-3253"></a>  name: "matrix"
+<a name="line-3254"></a>  type_attr: "T"
+<a name="line-3255"></a>}
+<a name="line-3256"></a>input_arg {
+<a name="line-3257"></a>  description: "Shape is `[..., M, K]`." name: "rhs" type_attr: "T"
+<a name="line-3258"></a>}
+<a name="line-3259"></a>input_arg {
+<a name="line-3260"></a>  description: "Scalar tensor."
+<a name="line-3261"></a>  name: "l2_regularizer"
+<a name="line-3262"></a>  type: DT_DOUBLE
+<a name="line-3263"></a>}
+<a name="line-3264"></a>output_arg {
+<a name="line-3265"></a>  description: "Shape is `[..., N, K]`."
+<a name="line-3266"></a>  name: "output"
+<a name="line-3267"></a>  type_attr: "T"
+<a name="line-3268"></a>}
+<a name="line-3269"></a>-}</span>
+<a name="line-3270"></a>
+<a name="line-3271"></a><a name="rGBToHSV"></a><span class='hs-comment'>-- | Converts one or more images from RGB to HSV.</span>
+<a name="line-3272"></a><span class='hs-comment'>--</span>
+<a name="line-3273"></a><span class='hs-comment'>-- Outputs a tensor of the same shape as the `images` tensor, containing the HSV</span>
+<a name="line-3274"></a><span class='hs-comment'>-- value of the pixels. The output is only well defined if the value in `images`</span>
+<a name="line-3275"></a><span class='hs-comment'>-- are in `[0,1]`.</span>
+<a name="line-3276"></a><span class='hs-comment'>-- </span>
+<a name="line-3277"></a><span class='hs-comment'>-- `output[..., 0]` contains hue, `output[..., 1]` contains saturation, and</span>
+<a name="line-3278"></a><span class='hs-comment'>-- `output[..., 2]` contains value. All HSV values are in `[0,1]`. A hue of 0</span>
+<a name="line-3279"></a><span class='hs-comment'>-- corresponds to pure red, hue 1/3 is pure green, and 2/3 is pure blue.</span>
+<a name="line-3280"></a><span class='hs-definition'>rGBToHSV</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-3281"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __images__: 1-D or higher rank. RGB data to convert. Last dimension must be size 3.</span>
+<a name="line-3282"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: `images` converted to HSV.</span>
+<a name="line-3283"></a><span class='hs-definition'>rGBToHSV</span> <span class='hs-varid'>images</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3284"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RGBToHSV"</span>
+<a name="line-3285"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-3286"></a>        <span class='hs-varid'>images</span>
+<a name="line-3287"></a><span class='hs-comment'>{-
+<a name="line-3288"></a>attr {
+<a name="line-3289"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-3290"></a>  default_value { type: DT_FLOAT }
+<a name="line-3291"></a>  name: "T"
+<a name="line-3292"></a>  type: "type"
+<a name="line-3293"></a>}
+<a name="line-3294"></a>input_arg {
+<a name="line-3295"></a>  description: "1-D or higher rank. RGB data to convert. Last dimension must be size 3."
+<a name="line-3296"></a>  name: "images"
+<a name="line-3297"></a>  type_attr: "T"
+<a name="line-3298"></a>}
+<a name="line-3299"></a>output_arg {
+<a name="line-3300"></a>  description: "`images` converted to HSV."
+<a name="line-3301"></a>  name: "output"
+<a name="line-3302"></a>  type_attr: "T"
+<a name="line-3303"></a>}
+<a name="line-3304"></a>-}</span>
+<a name="line-3305"></a>
+<a name="line-3306"></a><a name="decodeGif"></a><span class='hs-comment'>-- | Decode the first frame of a GIF-encoded image to a uint8 tensor.</span>
+<a name="line-3307"></a><span class='hs-comment'>--</span>
+<a name="line-3308"></a><span class='hs-comment'>-- GIF with frame or transparency compression are not supported</span>
+<a name="line-3309"></a><span class='hs-comment'>-- convert animated GIF from compressed to uncompressed by:</span>
+<a name="line-3310"></a><span class='hs-comment'>-- </span>
+<a name="line-3311"></a><span class='hs-comment'>-- convert $src.gif -coalesce $dst.gif</span>
+<a name="line-3312"></a><span class='hs-definition'>decodeGif</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __contents__: 0-D.  The GIF-encoded image.</span>
+<a name="line-3313"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span> <span class='hs-comment'>-- ^ __image__: 4-D with shape `[num_frames, height, width, 3]`. RGB order</span>
+<a name="line-3314"></a><span class='hs-definition'>decodeGif</span> <span class='hs-varid'>contents</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3315"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DecodeGif"</span><span class='hs-layout'>)</span>
+<a name="line-3316"></a>        <span class='hs-varid'>contents</span>
+<a name="line-3317"></a><span class='hs-comment'>{-
+<a name="line-3318"></a>input_arg {
+<a name="line-3319"></a>  description: "0-D.  The GIF-encoded image."
+<a name="line-3320"></a>  name: "contents"
+<a name="line-3321"></a>  type: DT_STRING
+<a name="line-3322"></a>}
+<a name="line-3323"></a>output_arg {
+<a name="line-3324"></a>  description: "4-D with shape `[num_frames, height, width, 3]`. RGB order"
+<a name="line-3325"></a>  name: "image"
+<a name="line-3326"></a>  type: DT_UINT8
+<a name="line-3327"></a>}
+<a name="line-3328"></a>-}</span>
+<a name="line-3329"></a>
+<a name="line-3330"></a><span class='hs-comment'>-- | Deprecated. Disallowed in GraphDef version &gt;= 2.</span>
+<a name="line-3331"></a>
+<a name="line-3332"></a><a name="adjustContrast"></a><span class='hs-definition'>adjustContrast</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-3333"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-3334"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-3335"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-3336"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-3337"></a>                                                                <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-3338"></a>                                                                <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-3339"></a>                  <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __images__</span>
+<a name="line-3340"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __contrast_factor__</span>
+<a name="line-3341"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __min_value__</span>
+<a name="line-3342"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __max_value__</span>
+<a name="line-3343"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-3344"></a><span class='hs-definition'>adjustContrast</span> <span class='hs-varid'>images</span> <span class='hs-varid'>contrast_factor</span> <span class='hs-varid'>min_value</span> <span class='hs-varid'>max_value</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3345"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"AdjustContrast"</span>
+<a name="line-3346"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-3347"></a>        <span class='hs-varid'>images</span> <span class='hs-varid'>contrast_factor</span> <span class='hs-varid'>min_value</span> <span class='hs-varid'>max_value</span>
+<a name="line-3348"></a><span class='hs-comment'>{-
+<a name="line-3349"></a>attr {
+<a name="line-3350"></a>  allowed_values {
+<a name="line-3351"></a>    list {
+<a name="line-3352"></a>      type: DT_UINT8
+<a name="line-3353"></a>      type: DT_INT8
+<a name="line-3354"></a>      type: DT_INT16
+<a name="line-3355"></a>      type: DT_INT32
+<a name="line-3356"></a>      type: DT_INT64
+<a name="line-3357"></a>      type: DT_FLOAT
+<a name="line-3358"></a>      type: DT_DOUBLE
+<a name="line-3359"></a>    }
+<a name="line-3360"></a>  }
+<a name="line-3361"></a>  name: "T"
+<a name="line-3362"></a>  type: "type"
+<a name="line-3363"></a>}
+<a name="line-3364"></a>input_arg { name: "images" type_attr: "T" }
+<a name="line-3365"></a>input_arg { name: "contrast_factor" type: DT_FLOAT }
+<a name="line-3366"></a>input_arg { name: "min_value" type: DT_FLOAT }
+<a name="line-3367"></a>input_arg { name: "max_value" type: DT_FLOAT }
+<a name="line-3368"></a>output_arg { name: "output" type: DT_FLOAT }
+<a name="line-3369"></a>-}</span>
+<a name="line-3370"></a>
+<a name="line-3371"></a><a name="depthToSpace"></a><span class='hs-comment'>-- | DepthToSpace for tensors of type T.</span>
+<a name="line-3372"></a><span class='hs-comment'>--</span>
+<a name="line-3373"></a><span class='hs-comment'>-- Rearranges data from depth into blocks of spatial data.</span>
+<a name="line-3374"></a><span class='hs-comment'>-- This is the reverse transformation of SpaceToDepth. More specifically,</span>
+<a name="line-3375"></a><span class='hs-comment'>-- this op outputs a copy of the input tensor where values from the `depth`</span>
+<a name="line-3376"></a><span class='hs-comment'>-- dimension are moved in spatial blocks to the `height` and `width` dimensions.</span>
+<a name="line-3377"></a><span class='hs-comment'>-- The attr `block_size` indicates the input block size and how the data is moved.</span>
+<a name="line-3378"></a><span class='hs-comment'>-- </span>
+<a name="line-3379"></a><span class='hs-comment'>--   * Chunks of data of size `block_size * block_size` from depth are rearranged</span>
+<a name="line-3380"></a><span class='hs-comment'>--     into non-overlapping blocks of size `block_size x block_size`</span>
+<a name="line-3381"></a><span class='hs-comment'>--   * The width the output tensor is `input_depth * block_size`, whereas the</span>
+<a name="line-3382"></a><span class='hs-comment'>--     height is `input_height * block_size`.</span>
+<a name="line-3383"></a><span class='hs-comment'>--   * The depth of the input tensor must be divisible by</span>
+<a name="line-3384"></a><span class='hs-comment'>--     `block_size * block_size`.</span>
+<a name="line-3385"></a><span class='hs-comment'>-- </span>
+<a name="line-3386"></a><span class='hs-comment'>-- That is, assuming the input is in the shape:</span>
+<a name="line-3387"></a><span class='hs-comment'>-- `[batch, height, width, depth]`,</span>
+<a name="line-3388"></a><span class='hs-comment'>-- the shape of the output will be:</span>
+<a name="line-3389"></a><span class='hs-comment'>-- `[batch, height*block_size, width*block_size, depth/(block_size*block_size)]`</span>
+<a name="line-3390"></a><span class='hs-comment'>-- </span>
+<a name="line-3391"></a><span class='hs-comment'>-- This operation requires that the input tensor be of rank 4, and that</span>
+<a name="line-3392"></a><span class='hs-comment'>-- `block_size` be &gt;=1 and that `block_size * block_size` be a divisor of the</span>
+<a name="line-3393"></a><span class='hs-comment'>-- input depth.</span>
+<a name="line-3394"></a><span class='hs-comment'>-- </span>
+<a name="line-3395"></a><span class='hs-comment'>-- This operation is useful for resizing the activations between convolutions</span>
+<a name="line-3396"></a><span class='hs-comment'>-- (but keeping all data), e.g. instead of pooling. It is also useful for training</span>
+<a name="line-3397"></a><span class='hs-comment'>-- purely convolutional models.</span>
+<a name="line-3398"></a><span class='hs-comment'>-- </span>
+<a name="line-3399"></a><span class='hs-comment'>-- For example, given this input of shape `[1, 1, 1, 4]`, and a block size of 2:</span>
+<a name="line-3400"></a><span class='hs-comment'>-- </span>
+<a name="line-3401"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-3402"></a><span class='hs-comment'>-- x = [[[[1, 2, 3, 4]]]]</span>
+<a name="line-3403"></a><span class='hs-comment'>-- </span>
+<a name="line-3404"></a><span class='hs-comment'>-- ```</span>
+<a name="line-3405"></a><span class='hs-comment'>-- </span>
+<a name="line-3406"></a><span class='hs-comment'>-- This operation will output a tensor of shape `[1, 2, 2, 1]`:</span>
+<a name="line-3407"></a><span class='hs-comment'>-- </span>
+<a name="line-3408"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-3409"></a><span class='hs-comment'>--    [[[[1], [2]],</span>
+<a name="line-3410"></a><span class='hs-comment'>--      [[3], [4]]]]</span>
+<a name="line-3411"></a><span class='hs-comment'>-- ```</span>
+<a name="line-3412"></a><span class='hs-comment'>-- </span>
+<a name="line-3413"></a><span class='hs-comment'>-- Here, the input has a batch of 1 and each batch element has shape `[1, 1, 4]`,</span>
+<a name="line-3414"></a><span class='hs-comment'>-- the corresponding output will have 2x2 elements and will have a depth of</span>
+<a name="line-3415"></a><span class='hs-comment'>-- 1 channel (1 = `4 / (block_size * block_size)`).</span>
+<a name="line-3416"></a><span class='hs-comment'>-- The output element shape is `[2, 2, 1]`.</span>
+<a name="line-3417"></a><span class='hs-comment'>-- </span>
+<a name="line-3418"></a><span class='hs-comment'>-- For an input tensor with larger depth, here of shape `[1, 1, 1, 12]`, e.g.</span>
+<a name="line-3419"></a><span class='hs-comment'>-- </span>
+<a name="line-3420"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-3421"></a><span class='hs-comment'>-- x = [[[[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]]]]</span>
+<a name="line-3422"></a><span class='hs-comment'>-- ```</span>
+<a name="line-3423"></a><span class='hs-comment'>-- </span>
+<a name="line-3424"></a><span class='hs-comment'>-- This operation, for block size of 2, will return the following tensor of shape</span>
+<a name="line-3425"></a><span class='hs-comment'>-- `[1, 2, 2, 3]`</span>
+<a name="line-3426"></a><span class='hs-comment'>-- </span>
+<a name="line-3427"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-3428"></a><span class='hs-comment'>--    [[[[1, 2, 3], [4, 5, 6]],</span>
+<a name="line-3429"></a><span class='hs-comment'>--      [[7, 8, 9], [10, 11, 12]]]]</span>
+<a name="line-3430"></a><span class='hs-comment'>-- </span>
+<a name="line-3431"></a><span class='hs-comment'>-- ```</span>
+<a name="line-3432"></a><span class='hs-comment'>-- </span>
+<a name="line-3433"></a><span class='hs-comment'>-- Similarly, for the following input of shape `[1 2 2 4]`, and a block size of 2:</span>
+<a name="line-3434"></a><span class='hs-comment'>-- </span>
+<a name="line-3435"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-3436"></a><span class='hs-comment'>-- x =  [[[[1, 2, 3, 4],</span>
+<a name="line-3437"></a><span class='hs-comment'>--        [5, 6, 7, 8]],</span>
+<a name="line-3438"></a><span class='hs-comment'>--       [[9, 10, 11, 12],</span>
+<a name="line-3439"></a><span class='hs-comment'>--        [13, 14, 15, 16]]]]</span>
+<a name="line-3440"></a><span class='hs-comment'>-- ```</span>
+<a name="line-3441"></a><span class='hs-comment'>-- </span>
+<a name="line-3442"></a><span class='hs-comment'>-- the operator will return the following tensor of shape `[1 4 4 1]`:</span>
+<a name="line-3443"></a><span class='hs-comment'>-- </span>
+<a name="line-3444"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-3445"></a><span class='hs-comment'>-- x = [[ [1],   [2],  [5],  [6]],</span>
+<a name="line-3446"></a><span class='hs-comment'>--      [ [3],   [4],  [7],  [8]],</span>
+<a name="line-3447"></a><span class='hs-comment'>--      [ [9],  [10], [13],  [14]],</span>
+<a name="line-3448"></a><span class='hs-comment'>--      [ [11], [12], [15],  [16]]]</span>
+<a name="line-3449"></a><span class='hs-comment'>-- </span>
+<a name="line-3450"></a><span class='hs-comment'>-- ```</span>
+<a name="line-3451"></a><span class='hs-definition'>depthToSpace</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-3452"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __block_size__: The size of the spatial block, same as in Space2Depth.</span>
+<a name="line-3453"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-3454"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-3455"></a><span class='hs-definition'>depthToSpace</span> <span class='hs-varid'>block_size</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3456"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DepthToSpace"</span>
+<a name="line-3457"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-3458"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"block_size"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>block_size</span><span class='hs-layout'>)</span>
+<a name="line-3459"></a>        <span class='hs-varid'>input</span>
+<a name="line-3460"></a><span class='hs-comment'>{-
+<a name="line-3461"></a>attr { name: "T" type: "type" }
+<a name="line-3462"></a>attr {
+<a name="line-3463"></a>  description: "The size of the spatial block, same as in Space2Depth."
+<a name="line-3464"></a>  has_minimum: true
+<a name="line-3465"></a>  minimum: 2
+<a name="line-3466"></a>  name: "block_size"
+<a name="line-3467"></a>  type: "int"
+<a name="line-3468"></a>}
+<a name="line-3469"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-3470"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-3471"></a>-}</span>
+<a name="line-3472"></a>
+<a name="line-3473"></a><span class='hs-comment'>-- | </span>
+<a name="line-3474"></a>
+<a name="line-3475"></a><a name="batchMatrixSolve"></a><span class='hs-definition'>batchMatrixSolve</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-3476"></a>                    <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __matrix__</span>
+<a name="line-3477"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __rhs__</span>
+<a name="line-3478"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-3479"></a><span class='hs-definition'>batchMatrixSolve</span> <span class='hs-varid'>matrix</span> <span class='hs-varid'>rhs</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3480"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchMatrixSolve"</span>
+<a name="line-3481"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-3482"></a>        <span class='hs-varid'>matrix</span> <span class='hs-varid'>rhs</span>
+<a name="line-3483"></a><span class='hs-comment'>{-
+<a name="line-3484"></a>attr { default_value { b: false } name: "adjoint" type: "bool" }
+<a name="line-3485"></a>attr {
+<a name="line-3486"></a>  allowed_values { list { type: DT_DOUBLE type: DT_FLOAT } }
+<a name="line-3487"></a>  name: "T"
+<a name="line-3488"></a>  type: "type"
+<a name="line-3489"></a>}
+<a name="line-3490"></a>input_arg { name: "matrix" type_attr: "T" }
+<a name="line-3491"></a>input_arg { name: "rhs" type_attr: "T" }
+<a name="line-3492"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-3493"></a>-}</span>
+<a name="line-3494"></a>
+<a name="line-3495"></a><span class='hs-comment'>-- | Computes the complementary error function of `x` element-wise.</span>
+<a name="line-3496"></a>
+<a name="line-3497"></a><a name="erfc"></a><span class='hs-definition'>erfc</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-3498"></a>                                             <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-3499"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-3500"></a><span class='hs-definition'>erfc</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3501"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Erfc"</span>
+<a name="line-3502"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-3503"></a>        <span class='hs-varid'>x</span>
+<a name="line-3504"></a><span class='hs-comment'>{-
+<a name="line-3505"></a>attr {
+<a name="line-3506"></a>  allowed_values {
+<a name="line-3507"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-3508"></a>  }
+<a name="line-3509"></a>  name: "T"
+<a name="line-3510"></a>  type: "type"
+<a name="line-3511"></a>}
+<a name="line-3512"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-3513"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-3514"></a>-}</span>
+<a name="line-3515"></a>
+<a name="line-3516"></a><span class='hs-comment'>-- | Computes the gradient of bilinear interpolation.</span>
+<a name="line-3517"></a>
+<a name="line-3518"></a><a name="resizeBilinearGrad"></a><span class='hs-definition'>resizeBilinearGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-3519"></a>                                                              <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-3520"></a>                                                              <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-3521"></a>                      <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __grads__: 4-D with shape `[batch, height, width, channels]`.</span>
+<a name="line-3522"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __original_image__: 4-D with shape `[batch, orig_height, orig_width, channels]`,</span>
+<a name="line-3523"></a>                                     <span class='hs-comment'>-- The image tensor that was resized.</span>
+<a name="line-3524"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 4-D with shape `[batch, orig_height, orig_width, channels]`.</span>
+<a name="line-3525"></a>                      <span class='hs-comment'>-- Gradients with respect to the input image. Input image must have been</span>
+<a name="line-3526"></a>                      <span class='hs-comment'>-- float or double.</span>
+<a name="line-3527"></a><span class='hs-definition'>resizeBilinearGrad</span> <span class='hs-varid'>grads</span> <span class='hs-varid'>original_image</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3528"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ResizeBilinearGrad"</span>
+<a name="line-3529"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-3530"></a>        <span class='hs-varid'>grads</span> <span class='hs-varid'>original_image</span>
+<a name="line-3531"></a><span class='hs-comment'>{-
+<a name="line-3532"></a>attr {
+<a name="line-3533"></a>  allowed_values {
+<a name="line-3534"></a>    list { type: DT_FLOAT type: DT_HALF type: DT_DOUBLE }
+<a name="line-3535"></a>  }
+<a name="line-3536"></a>  name: "T"
+<a name="line-3537"></a>  type: "type"
+<a name="line-3538"></a>}
+<a name="line-3539"></a>attr {
+<a name="line-3540"></a>  default_value { b: false }
+<a name="line-3541"></a>  description: "If true, rescale grads by (orig_height - 1) / (height - 1), which\nexactly aligns the 4 corners of grads and original_image. If false, rescale by\norig_height / height. Treat similarly the width dimension."
+<a name="line-3542"></a>  name: "align_corners"
+<a name="line-3543"></a>  type: "bool"
+<a name="line-3544"></a>}
+<a name="line-3545"></a>input_arg {
+<a name="line-3546"></a>  description: "4-D with shape `[batch, height, width, channels]`."
+<a name="line-3547"></a>  name: "grads"
+<a name="line-3548"></a>  type: DT_FLOAT
+<a name="line-3549"></a>}
+<a name="line-3550"></a>input_arg {
+<a name="line-3551"></a>  description: "4-D with shape `[batch, orig_height, orig_width, channels]`,\nThe image tensor that was resized."
+<a name="line-3552"></a>  name: "original_image"
+<a name="line-3553"></a>  type_attr: "T"
+<a name="line-3554"></a>}
+<a name="line-3555"></a>output_arg {
+<a name="line-3556"></a>  description: "4-D with shape `[batch, orig_height, orig_width, channels]`.\nGradients with respect to the input image. Input image must have been\nfloat or double."
+<a name="line-3557"></a>  name: "output"
+<a name="line-3558"></a>  type_attr: "T"
+<a name="line-3559"></a>}
+<a name="line-3560"></a>-}</span>
+<a name="line-3561"></a>
+<a name="line-3562"></a><span class='hs-comment'>-- | Output a fact about factorials.</span>
+<a name="line-3563"></a>
+<a name="line-3564"></a><a name="fact"></a><span class='hs-definition'>fact</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __fact__</span>
+<a name="line-3565"></a><span class='hs-definition'>fact</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3566"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Fact"</span><span class='hs-layout'>)</span>
+<a name="line-3567"></a>        
+<a name="line-3568"></a><span class='hs-comment'>{-
+<a name="line-3569"></a>output_arg { name: "fact" type: DT_STRING }
+<a name="line-3570"></a>-}</span>
+<a name="line-3571"></a>
+<a name="line-3572"></a><span class='hs-comment'>-- | Delete the tensor specified by its handle in the session.</span>
+<a name="line-3573"></a>
+<a name="line-3574"></a><a name="deleteSessionTensor"></a><span class='hs-definition'>deleteSessionTensor</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle for a tensor stored in the session state.</span>
+<a name="line-3575"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-3576"></a><span class='hs-definition'>deleteSessionTensor</span> <span class='hs-varid'>handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3577"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DeleteSessionTensor"</span><span class='hs-layout'>)</span>
+<a name="line-3578"></a>        <span class='hs-varid'>handle</span>
+<a name="line-3579"></a><span class='hs-comment'>{-
+<a name="line-3580"></a>input_arg {
+<a name="line-3581"></a>  description: "The handle for a tensor stored in the session state."
+<a name="line-3582"></a>  name: "handle"
+<a name="line-3583"></a>  type: DT_STRING
+<a name="line-3584"></a>}
+<a name="line-3585"></a>-}</span>
+<a name="line-3586"></a>
+<a name="line-3587"></a><a name="logicalOr"></a><span class='hs-comment'>-- | Returns the truth value of x OR y element-wise.</span>
+<a name="line-3588"></a><span class='hs-comment'>--</span>
+<a name="line-3589"></a><span class='hs-comment'>-- *NOTE*: `LogicalOr` supports broadcasting. More about broadcasting</span>
+<a name="line-3590"></a><span class='hs-comment'>-- [here](<a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)">http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)</a></span>
+<a name="line-3591"></a><span class='hs-definition'>logicalOr</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-3592"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-3593"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-3594"></a><span class='hs-definition'>logicalOr</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3595"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"LogicalOr"</span><span class='hs-layout'>)</span>
+<a name="line-3596"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-3597"></a><span class='hs-comment'>{-
+<a name="line-3598"></a>input_arg { name: "x" type: DT_BOOL }
+<a name="line-3599"></a>input_arg { name: "y" type: DT_BOOL }
+<a name="line-3600"></a>output_arg { name: "z" type: DT_BOOL }
+<a name="line-3601"></a>-}</span>
+<a name="line-3602"></a>
+<a name="line-3603"></a><span class='hs-comment'>-- | Get the value of the tensor specified by its handle.</span>
+<a name="line-3604"></a>
+<a name="line-3605"></a><a name="getSessionTensor"></a><span class='hs-definition'>getSessionTensor</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-3606"></a>                    <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle for a tensor stored in the session state.</span>
+<a name="line-3607"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __value__: The tensor for the given handle.</span>
+<a name="line-3608"></a><span class='hs-definition'>getSessionTensor</span> <span class='hs-varid'>handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3609"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"GetSessionTensor"</span>
+<a name="line-3610"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-3611"></a>        <span class='hs-varid'>handle</span>
+<a name="line-3612"></a><span class='hs-comment'>{-
+<a name="line-3613"></a>attr {
+<a name="line-3614"></a>  description: "The type of the output value."
+<a name="line-3615"></a>  name: "dtype"
+<a name="line-3616"></a>  type: "type"
+<a name="line-3617"></a>}
+<a name="line-3618"></a>input_arg {
+<a name="line-3619"></a>  description: "The handle for a tensor stored in the session state."
+<a name="line-3620"></a>  name: "handle"
+<a name="line-3621"></a>  type: DT_STRING
+<a name="line-3622"></a>}
+<a name="line-3623"></a>output_arg {
+<a name="line-3624"></a>  description: "The tensor for the given handle."
+<a name="line-3625"></a>  name: "value"
+<a name="line-3626"></a>  type_attr: "dtype"
+<a name="line-3627"></a>}
+<a name="line-3628"></a>-}</span>
+<a name="line-3629"></a>
+<a name="line-3630"></a><span class='hs-comment'>-- | </span>
+<a name="line-3631"></a>
+<a name="line-3632"></a><a name="batchMatrixInverse"></a><span class='hs-definition'>batchMatrixInverse</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-3633"></a>                      <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-3634"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-3635"></a><span class='hs-definition'>batchMatrixInverse</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3636"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchMatrixInverse"</span>
+<a name="line-3637"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-3638"></a>        <span class='hs-varid'>input</span>
+<a name="line-3639"></a><span class='hs-comment'>{-
+<a name="line-3640"></a>attr { default_value { b: false } name: "adjoint" type: "bool" }
+<a name="line-3641"></a>attr {
+<a name="line-3642"></a>  allowed_values { list { type: DT_DOUBLE type: DT_FLOAT } }
+<a name="line-3643"></a>  name: "T"
+<a name="line-3644"></a>  type: "type"
+<a name="line-3645"></a>}
+<a name="line-3646"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-3647"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-3648"></a>-}</span>
+<a name="line-3649"></a>
+<a name="line-3650"></a><span class='hs-comment'>-- | Generate a glob pattern matching all sharded file names.</span>
+<a name="line-3651"></a>
+<a name="line-3652"></a><a name="shardedFilespec"></a><span class='hs-definition'>shardedFilespec</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __basename__</span>
+<a name="line-3653"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __num_shards__</span>
+<a name="line-3654"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __filename__</span>
+<a name="line-3655"></a><span class='hs-definition'>shardedFilespec</span> <span class='hs-varid'>basename</span> <span class='hs-varid'>num_shards</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3656"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ShardedFilespec"</span><span class='hs-layout'>)</span>
+<a name="line-3657"></a>        <span class='hs-varid'>basename</span> <span class='hs-varid'>num_shards</span>
+<a name="line-3658"></a><span class='hs-comment'>{-
+<a name="line-3659"></a>input_arg { name: "basename" type: DT_STRING }
+<a name="line-3660"></a>input_arg { name: "num_shards" type: DT_INT32 }
+<a name="line-3661"></a>output_arg { name: "filename" type: DT_STRING }
+<a name="line-3662"></a>-}</span>
+<a name="line-3663"></a>
+<a name="line-3664"></a><a name="decodeBase64"></a><span class='hs-comment'>-- | Decode web-safe base64-encoded strings.</span>
+<a name="line-3665"></a><span class='hs-comment'>--</span>
+<a name="line-3666"></a><span class='hs-comment'>-- Input may or may not have padding at the end. See EncodeBase64 for padding.</span>
+<a name="line-3667"></a><span class='hs-comment'>-- Web-safe means that input must use - and _ instead of + and /.</span>
+<a name="line-3668"></a><span class='hs-definition'>decodeBase64</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __input__: Base64 strings to decode.</span>
+<a name="line-3669"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __output__: Decoded strings.</span>
+<a name="line-3670"></a><span class='hs-definition'>decodeBase64</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3671"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DecodeBase64"</span><span class='hs-layout'>)</span>
+<a name="line-3672"></a>        <span class='hs-varid'>input</span>
+<a name="line-3673"></a><span class='hs-comment'>{-
+<a name="line-3674"></a>input_arg {
+<a name="line-3675"></a>  description: "Base64 strings to decode."
+<a name="line-3676"></a>  name: "input"
+<a name="line-3677"></a>  type: DT_STRING
+<a name="line-3678"></a>}
+<a name="line-3679"></a>output_arg {
+<a name="line-3680"></a>  description: "Decoded strings." name: "output" type: DT_STRING
+<a name="line-3681"></a>}
+<a name="line-3682"></a>-}</span>
+<a name="line-3683"></a>
+<a name="line-3684"></a><span class='hs-comment'>-- | Store the input tensor in the state of the current session.</span>
+<a name="line-3685"></a>
+<a name="line-3686"></a><a name="getSessionHandle"></a><span class='hs-definition'>getSessionHandle</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-3687"></a>                    <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: The tensor to be stored.</span>
+<a name="line-3688"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle for the tensor stored in the session state.</span>
+<a name="line-3689"></a><span class='hs-definition'>getSessionHandle</span> <span class='hs-varid'>value</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3690"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"GetSessionHandle"</span>
+<a name="line-3691"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-3692"></a>        <span class='hs-varid'>value</span>
+<a name="line-3693"></a><span class='hs-comment'>{-
+<a name="line-3694"></a>attr { name: "T" type: "type" }
+<a name="line-3695"></a>input_arg {
+<a name="line-3696"></a>  description: "The tensor to be stored."
+<a name="line-3697"></a>  name: "value"
+<a name="line-3698"></a>  type_attr: "T"
+<a name="line-3699"></a>}
+<a name="line-3700"></a>output_arg {
+<a name="line-3701"></a>  description: "The handle for the tensor stored in the session state."
+<a name="line-3702"></a>  name: "handle"
+<a name="line-3703"></a>  type: DT_STRING
+<a name="line-3704"></a>}
+<a name="line-3705"></a>-}</span>
+<a name="line-3706"></a>
+<a name="line-3707"></a><span class='hs-comment'>-- | Table initializer that takes two tensors for keys and values respectively.</span>
+<a name="line-3708"></a>
+<a name="line-3709"></a><a name="initializeTable"></a><span class='hs-definition'>initializeTable</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>tkey</span> <span class='hs-varid'>tval</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>tkey</span><span class='hs-layout'>,</span>
+<a name="line-3710"></a>                                                <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tval</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-3711"></a>                   <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __table_handle__: Handle to a table which will be initialized.</span>
+<a name="line-3712"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tkey</span> <span class='hs-comment'>-- ^ __keys__: Keys of type Tkey.</span>
+<a name="line-3713"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>tval</span> <span class='hs-comment'>-- ^ __values__: Values of type Tval.</span>
+<a name="line-3714"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-3715"></a><span class='hs-definition'>initializeTable</span> <span class='hs-varid'>table_handle</span> <span class='hs-varid'>keys</span> <span class='hs-varid'>values</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3716"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"InitializeTable"</span>
+<a name="line-3717"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tkey"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tkey</span><span class='hs-layout'>)</span>
+<a name="line-3718"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tval"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tval</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-3719"></a>        <span class='hs-varid'>table_handle</span> <span class='hs-varid'>keys</span> <span class='hs-varid'>values</span>
+<a name="line-3720"></a><span class='hs-comment'>{-
+<a name="line-3721"></a>attr { name: "Tkey" type: "type" }
+<a name="line-3722"></a>attr { name: "Tval" type: "type" }
+<a name="line-3723"></a>input_arg {
+<a name="line-3724"></a>  description: "Handle to a table which will be initialized."
+<a name="line-3725"></a>  is_ref: true
+<a name="line-3726"></a>  name: "table_handle"
+<a name="line-3727"></a>  type: DT_STRING
+<a name="line-3728"></a>}
+<a name="line-3729"></a>input_arg {
+<a name="line-3730"></a>  description: "Keys of type Tkey." name: "keys" type_attr: "Tkey"
+<a name="line-3731"></a>}
+<a name="line-3732"></a>input_arg {
+<a name="line-3733"></a>  description: "Values of type Tval."
+<a name="line-3734"></a>  name: "values"
+<a name="line-3735"></a>  type_attr: "Tval"
+<a name="line-3736"></a>}
+<a name="line-3737"></a>-}</span>
+<a name="line-3738"></a>
+<a name="line-3739"></a><span class='hs-comment'>-- | Computes tan of x element-wise.</span>
+<a name="line-3740"></a>
+<a name="line-3741"></a><a name="tan"></a><span class='hs-definition'>tan</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-3742"></a>                                            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-3743"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-3744"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-3745"></a>                                            <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-3746"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-3747"></a><span class='hs-definition'>tan</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3748"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Tan"</span>
+<a name="line-3749"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-3750"></a>        <span class='hs-varid'>x</span>
+<a name="line-3751"></a><span class='hs-comment'>{-
+<a name="line-3752"></a>attr {
+<a name="line-3753"></a>  allowed_values {
+<a name="line-3754"></a>    list {
+<a name="line-3755"></a>      type: DT_HALF
+<a name="line-3756"></a>      type: DT_FLOAT
+<a name="line-3757"></a>      type: DT_DOUBLE
+<a name="line-3758"></a>      type: DT_INT32
+<a name="line-3759"></a>      type: DT_INT64
+<a name="line-3760"></a>      type: DT_COMPLEX64
+<a name="line-3761"></a>      type: DT_COMPLEX128
+<a name="line-3762"></a>    }
+<a name="line-3763"></a>  }
+<a name="line-3764"></a>  name: "T"
+<a name="line-3765"></a>  type: "type"
+<a name="line-3766"></a>}
+<a name="line-3767"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-3768"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-3769"></a>-}</span>
+<a name="line-3770"></a>
+<a name="line-3771"></a><span class='hs-comment'>-- | Computes hyperbolic tangent of `x` element-wise.</span>
+<a name="line-3772"></a>
+<a name="line-3773"></a><a name="tanh"></a><span class='hs-definition'>tanh</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-3774"></a>                                             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-3775"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-3776"></a>                                             <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-3777"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-3778"></a><span class='hs-definition'>tanh</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3779"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Tanh"</span>
+<a name="line-3780"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-3781"></a>        <span class='hs-varid'>x</span>
+<a name="line-3782"></a><span class='hs-comment'>{-
+<a name="line-3783"></a>attr {
+<a name="line-3784"></a>  allowed_values {
+<a name="line-3785"></a>    list {
+<a name="line-3786"></a>      type: DT_HALF
+<a name="line-3787"></a>      type: DT_FLOAT
+<a name="line-3788"></a>      type: DT_DOUBLE
+<a name="line-3789"></a>      type: DT_COMPLEX64
+<a name="line-3790"></a>      type: DT_COMPLEX128
+<a name="line-3791"></a>    }
+<a name="line-3792"></a>  }
+<a name="line-3793"></a>  name: "T"
+<a name="line-3794"></a>  type: "type"
+<a name="line-3795"></a>}
+<a name="line-3796"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-3797"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-3798"></a>-}</span>
+<a name="line-3799"></a>
+<a name="line-3800"></a><span class='hs-comment'>-- | Update '*var' according to the proximal adagrad scheme.</span>
+<a name="line-3801"></a>
+<a name="line-3802"></a><a name="applyAdagradDA"></a><span class='hs-definition'>applyAdagradDA</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>v8</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-3803"></a>                                                      <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-3804"></a>                                                              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-3805"></a>                                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-3806"></a>                                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-3807"></a>                                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-3808"></a>                                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-3809"></a>                                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-3810"></a>                                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-3811"></a>                                                              <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-3812"></a>                                                              <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-3813"></a>                  <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-3814"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __gradient_accumulator__: Should be from a Variable().</span>
+<a name="line-3815"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __gradient_squared_accumulator__: Should be from a Variable().</span>
+<a name="line-3816"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: The gradient.</span>
+<a name="line-3817"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr__: Scaling factor. Must be a scalar.</span>
+<a name="line-3818"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l1__: L1 regularization. Must be a scalar.</span>
+<a name="line-3819"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l2__: L2 regularization. Must be a scalar.</span>
+<a name="line-3820"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v8</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __global_step__: Training step number. Must be a scalar.</span>
+<a name="line-3821"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-3822"></a><span class='hs-definition'>applyAdagradDA</span> <span class='hs-varid'>var</span> <span class='hs-varid'>gradient_accumulator</span> <span class='hs-varid'>gradient_squared_accumulator</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>l1</span>
+<a name="line-3823"></a>               <span class='hs-varid'>l2</span> <span class='hs-varid'>global_step</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3824"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ApplyAdagradDA"</span>
+<a name="line-3825"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-3826"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>gradient_accumulator</span> <span class='hs-varid'>gradient_squared_accumulator</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>l1</span> <span class='hs-varid'>l2</span>
+<a name="line-3827"></a>        <span class='hs-varid'>global_step</span>
+<a name="line-3828"></a><span class='hs-comment'>{-
+<a name="line-3829"></a>attr {
+<a name="line-3830"></a>  allowed_values {
+<a name="line-3831"></a>    list {
+<a name="line-3832"></a>      type: DT_FLOAT
+<a name="line-3833"></a>      type: DT_DOUBLE
+<a name="line-3834"></a>      type: DT_INT64
+<a name="line-3835"></a>      type: DT_INT32
+<a name="line-3836"></a>      type: DT_UINT8
+<a name="line-3837"></a>      type: DT_UINT16
+<a name="line-3838"></a>      type: DT_INT16
+<a name="line-3839"></a>      type: DT_INT8
+<a name="line-3840"></a>      type: DT_COMPLEX64
+<a name="line-3841"></a>      type: DT_COMPLEX128
+<a name="line-3842"></a>      type: DT_QINT8
+<a name="line-3843"></a>      type: DT_QUINT8
+<a name="line-3844"></a>      type: DT_QINT32
+<a name="line-3845"></a>      type: DT_HALF
+<a name="line-3846"></a>    }
+<a name="line-3847"></a>  }
+<a name="line-3848"></a>  name: "T"
+<a name="line-3849"></a>  type: "type"
+<a name="line-3850"></a>}
+<a name="line-3851"></a>attr {
+<a name="line-3852"></a>  default_value { b: false }
+<a name="line-3853"></a>  description: "If True, updating of the var and accum tensors will be protected by\na lock; otherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-3854"></a>  name: "use_locking"
+<a name="line-3855"></a>  type: "bool"
+<a name="line-3856"></a>}
+<a name="line-3857"></a>input_arg {
+<a name="line-3858"></a>  description: "Should be from a Variable()."
+<a name="line-3859"></a>  is_ref: true
+<a name="line-3860"></a>  name: "var"
+<a name="line-3861"></a>  type_attr: "T"
+<a name="line-3862"></a>}
+<a name="line-3863"></a>input_arg {
+<a name="line-3864"></a>  description: "Should be from a Variable()."
+<a name="line-3865"></a>  is_ref: true
+<a name="line-3866"></a>  name: "gradient_accumulator"
+<a name="line-3867"></a>  type_attr: "T"
+<a name="line-3868"></a>}
+<a name="line-3869"></a>input_arg {
+<a name="line-3870"></a>  description: "Should be from a Variable()."
+<a name="line-3871"></a>  is_ref: true
+<a name="line-3872"></a>  name: "gradient_squared_accumulator"
+<a name="line-3873"></a>  type_attr: "T"
+<a name="line-3874"></a>}
+<a name="line-3875"></a>input_arg {
+<a name="line-3876"></a>  description: "The gradient." name: "grad" type_attr: "T"
+<a name="line-3877"></a>}
+<a name="line-3878"></a>input_arg {
+<a name="line-3879"></a>  description: "Scaling factor. Must be a scalar."
+<a name="line-3880"></a>  name: "lr"
+<a name="line-3881"></a>  type_attr: "T"
+<a name="line-3882"></a>}
+<a name="line-3883"></a>input_arg {
+<a name="line-3884"></a>  description: "L1 regularization. Must be a scalar."
+<a name="line-3885"></a>  name: "l1"
+<a name="line-3886"></a>  type_attr: "T"
+<a name="line-3887"></a>}
+<a name="line-3888"></a>input_arg {
+<a name="line-3889"></a>  description: "L2 regularization. Must be a scalar."
+<a name="line-3890"></a>  name: "l2"
+<a name="line-3891"></a>  type_attr: "T"
+<a name="line-3892"></a>}
+<a name="line-3893"></a>input_arg {
+<a name="line-3894"></a>  description: "Training step number. Must be a scalar."
+<a name="line-3895"></a>  name: "global_step"
+<a name="line-3896"></a>  type: DT_INT64
+<a name="line-3897"></a>}
+<a name="line-3898"></a>output_arg {
+<a name="line-3899"></a>  description: "Same as \"var\"."
+<a name="line-3900"></a>  is_ref: true
+<a name="line-3901"></a>  name: "out"
+<a name="line-3902"></a>  type_attr: "T"
+<a name="line-3903"></a>}
+<a name="line-3904"></a>-}</span>
+<a name="line-3905"></a>
+<a name="line-3906"></a><a name="stringToHashBucket"></a><span class='hs-comment'>-- | Converts each string in the input Tensor to its hash mod by a number of buckets.</span>
+<a name="line-3907"></a><span class='hs-comment'>--</span>
+<a name="line-3908"></a><span class='hs-comment'>-- The hash function is deterministic on the content of the string within the</span>
+<a name="line-3909"></a><span class='hs-comment'>-- process.</span>
+<a name="line-3910"></a><span class='hs-comment'>-- </span>
+<a name="line-3911"></a><span class='hs-comment'>-- Note that the hash function may change from time to time.</span>
+<a name="line-3912"></a><span class='hs-comment'>-- This functionality will be deprecated and it's recommended to use</span>
+<a name="line-3913"></a><span class='hs-comment'>-- `tf.string_to_hash_bucket_fast()` or `tf.string_to_hash_bucket_strong()`.</span>
+<a name="line-3914"></a><span class='hs-definition'>stringToHashBucket</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_buckets__: The number of buckets.</span>
+<a name="line-3915"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __string_tensor__</span>
+<a name="line-3916"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __output__: A Tensor of the same shape as the input `string_tensor`.</span>
+<a name="line-3917"></a><span class='hs-definition'>stringToHashBucket</span> <span class='hs-varid'>num_buckets</span> <span class='hs-varid'>string_tensor</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3918"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"StringToHashBucket"</span>
+<a name="line-3919"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_buckets"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_buckets</span><span class='hs-layout'>)</span>
+<a name="line-3920"></a>        <span class='hs-varid'>string_tensor</span>
+<a name="line-3921"></a><span class='hs-comment'>{-
+<a name="line-3922"></a>attr {
+<a name="line-3923"></a>  description: "The number of buckets."
+<a name="line-3924"></a>  has_minimum: true
+<a name="line-3925"></a>  minimum: 1
+<a name="line-3926"></a>  name: "num_buckets"
+<a name="line-3927"></a>  type: "int"
+<a name="line-3928"></a>}
+<a name="line-3929"></a>input_arg { name: "string_tensor" type: DT_STRING }
+<a name="line-3930"></a>output_arg {
+<a name="line-3931"></a>  description: "A Tensor of the same shape as the input `string_tensor`."
+<a name="line-3932"></a>  name: "output"
+<a name="line-3933"></a>  type: DT_INT64
+<a name="line-3934"></a>}
+<a name="line-3935"></a>-}</span>
+<a name="line-3936"></a>
+<a name="line-3937"></a><span class='hs-comment'>-- | Computes gradients for the exponential linear (Elu) operation.</span>
+<a name="line-3938"></a>
+<a name="line-3939"></a><a name="eluGrad"></a><span class='hs-definition'>eluGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-3940"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-3941"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-3942"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-3943"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-3944"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-3945"></a>                                                   <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-3946"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __gradients__: The backpropagated gradients to the corresponding Elu operation.</span>
+<a name="line-3947"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __outputs__: The outputs of the corresponding Elu operation.</span>
+<a name="line-3948"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __backprops__: The gradients: `gradients * (outputs + 1)` if outputs &lt; 0,</span>
+<a name="line-3949"></a>           <span class='hs-comment'>-- `gradients` otherwise.</span>
+<a name="line-3950"></a><span class='hs-definition'>eluGrad</span> <span class='hs-varid'>gradients</span> <span class='hs-varid'>outputs</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-3951"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"EluGrad"</span>
+<a name="line-3952"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-3953"></a>        <span class='hs-varid'>gradients</span> <span class='hs-varid'>outputs</span>
+<a name="line-3954"></a><span class='hs-comment'>{-
+<a name="line-3955"></a>attr {
+<a name="line-3956"></a>  allowed_values {
+<a name="line-3957"></a>    list {
+<a name="line-3958"></a>      type: DT_FLOAT
+<a name="line-3959"></a>      type: DT_DOUBLE
+<a name="line-3960"></a>      type: DT_INT32
+<a name="line-3961"></a>      type: DT_INT64
+<a name="line-3962"></a>      type: DT_UINT8
+<a name="line-3963"></a>      type: DT_INT16
+<a name="line-3964"></a>      type: DT_INT8
+<a name="line-3965"></a>      type: DT_UINT16
+<a name="line-3966"></a>      type: DT_HALF
+<a name="line-3967"></a>    }
+<a name="line-3968"></a>  }
+<a name="line-3969"></a>  name: "T"
+<a name="line-3970"></a>  type: "type"
+<a name="line-3971"></a>}
+<a name="line-3972"></a>input_arg {
+<a name="line-3973"></a>  description: "The backpropagated gradients to the corresponding Elu operation."
+<a name="line-3974"></a>  name: "gradients"
+<a name="line-3975"></a>  type_attr: "T"
+<a name="line-3976"></a>}
+<a name="line-3977"></a>input_arg {
+<a name="line-3978"></a>  description: "The outputs of the corresponding Elu operation."
+<a name="line-3979"></a>  name: "outputs"
+<a name="line-3980"></a>  type_attr: "T"
+<a name="line-3981"></a>}
+<a name="line-3982"></a>output_arg {
+<a name="line-3983"></a>  description: "The gradients: `gradients * (outputs + 1)` if outputs &lt; 0,\n`gradients` otherwise."
+<a name="line-3984"></a>  name: "backprops"
+<a name="line-3985"></a>  type_attr: "T"
+<a name="line-3986"></a>}
+<a name="line-3987"></a>-}</span>
+<a name="line-3988"></a>
+<a name="line-3989"></a><a name="fractionalAvgPoolGrad"></a><span class='hs-comment'>-- | Computes gradient of the FractionalAvgPool function.</span>
+<a name="line-3990"></a><span class='hs-comment'>--</span>
+<a name="line-3991"></a><span class='hs-comment'>-- Unlike FractionalMaxPoolGrad, we don't need to find arg_max for</span>
+<a name="line-3992"></a><span class='hs-comment'>-- FractionalAvgPoolGrad, we just need to evenly back-propagate each element of</span>
+<a name="line-3993"></a><span class='hs-comment'>-- out_backprop to those indices that form the same pooling cell. Therefore, we</span>
+<a name="line-3994"></a><span class='hs-comment'>-- just need to know the shape of original input tensor, instead of the whole</span>
+<a name="line-3995"></a><span class='hs-comment'>-- tensor.</span>
+<a name="line-3996"></a><span class='hs-definition'>fractionalAvgPoolGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-3997"></a>                                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-3998"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-3999"></a>                                                         <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-4000"></a>                         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __orig_input_tensor_shape__: Original input tensor shape for `fractional_avg_pool`</span>
+<a name="line-4001"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out_backprop__: 4-D with shape `[batch, height, width, channels]`.  Gradients</span>
+<a name="line-4002"></a>                                        <span class='hs-comment'>-- w.r.t. the output of `fractional_avg_pool`.</span>
+<a name="line-4003"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __row_pooling_sequence__: row pooling sequence, form pooling region with</span>
+<a name="line-4004"></a>                                                     <span class='hs-comment'>-- col_pooling_sequence.</span>
+<a name="line-4005"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __col_pooling_sequence__: column pooling sequence, form pooling region with</span>
+<a name="line-4006"></a>                                                     <span class='hs-comment'>-- row_pooling sequence.</span>
+<a name="line-4007"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 4-D.  Gradients w.r.t. the input of `fractional_avg_pool`.</span>
+<a name="line-4008"></a><span class='hs-definition'>fractionalAvgPoolGrad</span> <span class='hs-varid'>orig_input_tensor_shape</span> <span class='hs-varid'>out_backprop</span> <span class='hs-varid'>row_pooling_sequence</span>
+<a name="line-4009"></a>                      <span class='hs-varid'>col_pooling_sequence</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4010"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"FractionalAvgPoolGrad"</span>
+<a name="line-4011"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-4012"></a>        <span class='hs-varid'>orig_input_tensor_shape</span> <span class='hs-varid'>out_backprop</span> <span class='hs-varid'>row_pooling_sequence</span>
+<a name="line-4013"></a>        <span class='hs-varid'>col_pooling_sequence</span>
+<a name="line-4014"></a><span class='hs-comment'>{-
+<a name="line-4015"></a>attr {
+<a name="line-4016"></a>  default_value { b: false }
+<a name="line-4017"></a>  description: "When set to True, it means when pooling, the values at the boundary\nof adjacent pooling cells are used by both cells. For example:\n\n`index  0  1  2  3  4`\n\n`value  20 5  16 3  7`\n\nIf the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.\nThe result would be [41/3, 26/3] for fractional avg pooling."
+<a name="line-4018"></a>  name: "overlapping"
+<a name="line-4019"></a>  type: "bool"
+<a name="line-4020"></a>}
+<a name="line-4021"></a>attr {
+<a name="line-4022"></a>  allowed_values {
+<a name="line-4023"></a>    list {
+<a name="line-4024"></a>      type: DT_FLOAT type: DT_DOUBLE type: DT_INT32 type: DT_INT64
+<a name="line-4025"></a>    }
+<a name="line-4026"></a>  }
+<a name="line-4027"></a>  name: "T"
+<a name="line-4028"></a>  type: "type"
+<a name="line-4029"></a>}
+<a name="line-4030"></a>input_arg {
+<a name="line-4031"></a>  description: "Original input tensor shape for `fractional_avg_pool`"
+<a name="line-4032"></a>  name: "orig_input_tensor_shape"
+<a name="line-4033"></a>  type: DT_INT64
+<a name="line-4034"></a>}
+<a name="line-4035"></a>input_arg {
+<a name="line-4036"></a>  description: "4-D with shape `[batch, height, width, channels]`.  Gradients\nw.r.t. the output of `fractional_avg_pool`."
+<a name="line-4037"></a>  name: "out_backprop"
+<a name="line-4038"></a>  type_attr: "T"
+<a name="line-4039"></a>}
+<a name="line-4040"></a>input_arg {
+<a name="line-4041"></a>  description: "row pooling sequence, form pooling region with\ncol_pooling_sequence."
+<a name="line-4042"></a>  name: "row_pooling_sequence"
+<a name="line-4043"></a>  type: DT_INT64
+<a name="line-4044"></a>}
+<a name="line-4045"></a>input_arg {
+<a name="line-4046"></a>  description: "column pooling sequence, form pooling region with\nrow_pooling sequence."
+<a name="line-4047"></a>  name: "col_pooling_sequence"
+<a name="line-4048"></a>  type: DT_INT64
+<a name="line-4049"></a>}
+<a name="line-4050"></a>output_arg {
+<a name="line-4051"></a>  description: "4-D.  Gradients w.r.t. the input of `fractional_avg_pool`."
+<a name="line-4052"></a>  name: "output"
+<a name="line-4053"></a>  type_attr: "T"
+<a name="line-4054"></a>}
+<a name="line-4055"></a>-}</span>
+<a name="line-4056"></a>
+<a name="line-4057"></a><a name="matrixTriangularSolve"></a><span class='hs-comment'>-- | Solves systems of linear equations with upper or lower triangular matrices by</span>
+<a name="line-4058"></a><span class='hs-comment'>--</span>
+<a name="line-4059"></a><span class='hs-comment'>-- backsubstitution.</span>
+<a name="line-4060"></a><span class='hs-comment'>-- </span>
+<a name="line-4061"></a><span class='hs-comment'>-- `matrix` is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions form</span>
+<a name="line-4062"></a><span class='hs-comment'>-- square matrices. If `lower` is `True` then the strictly upper triangular part</span>
+<a name="line-4063"></a><span class='hs-comment'>-- of each inner-most matrix is assumed to be zero and not accessed.</span>
+<a name="line-4064"></a><span class='hs-comment'>-- If `lower` is False then the strictly lower triangular part of each inner-most</span>
+<a name="line-4065"></a><span class='hs-comment'>-- matrix is assumed to be zero and not accessed.</span>
+<a name="line-4066"></a><span class='hs-comment'>-- `rhs` is a tensor of shape `[..., M, K]`.</span>
+<a name="line-4067"></a><span class='hs-comment'>-- </span>
+<a name="line-4068"></a><span class='hs-comment'>-- The output is a tensor of shape `[..., M, K]`. If `adjoint` is</span>
+<a name="line-4069"></a><span class='hs-comment'>-- `True` then the innermost matrices in output` satisfy matrix equations</span>
+<a name="line-4070"></a><span class='hs-comment'>-- `matrix[..., :, :] * output[..., :, :] = rhs[..., :, :]`.</span>
+<a name="line-4071"></a><span class='hs-comment'>-- If `adjoint` is `False` then the strictly then the  innermost matrices in</span>
+<a name="line-4072"></a><span class='hs-comment'>-- `output` satisfy matrix equations</span>
+<a name="line-4073"></a><span class='hs-comment'>-- `adjoint(matrix[..., i, k]) * output[..., k, j] = rhs[..., i, j]`.</span>
+<a name="line-4074"></a><span class='hs-definition'>matrixTriangularSolve</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-4075"></a>                                                                 <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-4076"></a>                         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __matrix__: Shape is `[..., M, M]`.</span>
+<a name="line-4077"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __rhs__: Shape is `[..., M, K]`.</span>
+<a name="line-4078"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Shape is `[..., M, K]`.</span>
+<a name="line-4079"></a><span class='hs-definition'>matrixTriangularSolve</span> <span class='hs-varid'>matrix</span> <span class='hs-varid'>rhs</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4080"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MatrixTriangularSolve"</span>
+<a name="line-4081"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-4082"></a>        <span class='hs-varid'>matrix</span> <span class='hs-varid'>rhs</span>
+<a name="line-4083"></a><span class='hs-comment'>{-
+<a name="line-4084"></a>attr {
+<a name="line-4085"></a>  default_value { b: true }
+<a name="line-4086"></a>  description: "Boolean indicating whether the innermost matrices in `matrix` are\nlower or upper triangular."
+<a name="line-4087"></a>  name: "lower"
+<a name="line-4088"></a>  type: "bool"
+<a name="line-4089"></a>}
+<a name="line-4090"></a>attr {
+<a name="line-4091"></a>  default_value { b: false }
+<a name="line-4092"></a>  description: "Boolean indicating whether to solve with `matrix` or its (block-wise)\nadjoint."
+<a name="line-4093"></a>  name: "adjoint"
+<a name="line-4094"></a>  type: "bool"
+<a name="line-4095"></a>}
+<a name="line-4096"></a>attr {
+<a name="line-4097"></a>  allowed_values { list { type: DT_DOUBLE type: DT_FLOAT } }
+<a name="line-4098"></a>  name: "T"
+<a name="line-4099"></a>  type: "type"
+<a name="line-4100"></a>}
+<a name="line-4101"></a>input_arg {
+<a name="line-4102"></a>  description: "Shape is `[..., M, M]`."
+<a name="line-4103"></a>  name: "matrix"
+<a name="line-4104"></a>  type_attr: "T"
+<a name="line-4105"></a>}
+<a name="line-4106"></a>input_arg {
+<a name="line-4107"></a>  description: "Shape is `[..., M, K]`." name: "rhs" type_attr: "T"
+<a name="line-4108"></a>}
+<a name="line-4109"></a>output_arg {
+<a name="line-4110"></a>  description: "Shape is `[..., M, K]`."
+<a name="line-4111"></a>  name: "output"
+<a name="line-4112"></a>  type_attr: "T"
+<a name="line-4113"></a>}
+<a name="line-4114"></a>-}</span>
+<a name="line-4115"></a>
+<a name="line-4116"></a><a name="editDistance"></a><span class='hs-comment'>-- | Computes the (possibly normalized) Levenshtein Edit Distance.</span>
+<a name="line-4117"></a><span class='hs-comment'>--</span>
+<a name="line-4118"></a><span class='hs-comment'>-- The inputs are variable-length sequences provided by SparseTensors</span>
+<a name="line-4119"></a><span class='hs-comment'>--   (hypothesis_indices, hypothesis_values, hypothesis_shape)</span>
+<a name="line-4120"></a><span class='hs-comment'>-- and</span>
+<a name="line-4121"></a><span class='hs-comment'>--   (truth_indices, truth_values, truth_shape).</span>
+<a name="line-4122"></a><span class='hs-comment'>-- </span>
+<a name="line-4123"></a><span class='hs-comment'>-- The inputs are:</span>
+<a name="line-4124"></a><span class='hs-definition'>editDistance</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-4125"></a>                <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __hypothesis_indices__: The indices of the hypothesis list SparseTensor.</span>
+<a name="line-4126"></a>                                         <span class='hs-comment'>-- This is an N x R int64 matrix.</span>
+<a name="line-4127"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __hypothesis_values__: The values of the hypothesis list SparseTensor.</span>
+<a name="line-4128"></a>                               <span class='hs-comment'>-- This is an N-length vector.</span>
+<a name="line-4129"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __hypothesis_shape__: The shape of the hypothesis list SparseTensor.</span>
+<a name="line-4130"></a>                                            <span class='hs-comment'>-- This is an R-length vector.</span>
+<a name="line-4131"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __truth_indices__: The indices of the truth list SparseTensor.</span>
+<a name="line-4132"></a>                                            <span class='hs-comment'>-- This is an M x R int64 matrix.</span>
+<a name="line-4133"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __truth_values__: The values of the truth list SparseTensor.</span>
+<a name="line-4134"></a>                               <span class='hs-comment'>-- This is an M-length vector.</span>
+<a name="line-4135"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __truth_shape__: truth indices, vector.</span>
+<a name="line-4136"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __output__: A dense float tensor with rank R - 1.</span>
+<a name="line-4137"></a>                <span class='hs-comment'>-- </span>
+<a name="line-4138"></a>                <span class='hs-comment'>-- For the example input:</span>
+<a name="line-4139"></a>                <span class='hs-comment'>-- </span>
+<a name="line-4140"></a>                <span class='hs-comment'>--     // hypothesis represents a 2x1 matrix with variable-length values:</span>
+<a name="line-4141"></a>                <span class='hs-comment'>--     //   (0,0) = ["a"]</span>
+<a name="line-4142"></a>                <span class='hs-comment'>--     //   (1,0) = ["b"]</span>
+<a name="line-4143"></a>                <span class='hs-comment'>--     hypothesis_indices = [[0, 0, 0],</span>
+<a name="line-4144"></a>                <span class='hs-comment'>--                           [1, 0, 0]]</span>
+<a name="line-4145"></a>                <span class='hs-comment'>--     hypothesis_values = ["a", "b"]</span>
+<a name="line-4146"></a>                <span class='hs-comment'>--     hypothesis_shape = [2, 1, 1]</span>
+<a name="line-4147"></a>                <span class='hs-comment'>-- </span>
+<a name="line-4148"></a>                <span class='hs-comment'>--     // truth represents a 2x2 matrix with variable-length values:</span>
+<a name="line-4149"></a>                <span class='hs-comment'>--     //   (0,0) = []</span>
+<a name="line-4150"></a>                <span class='hs-comment'>--     //   (0,1) = ["a"]</span>
+<a name="line-4151"></a>                <span class='hs-comment'>--     //   (1,0) = ["b", "c"]</span>
+<a name="line-4152"></a>                <span class='hs-comment'>--     //   (1,1) = ["a"]</span>
+<a name="line-4153"></a>                <span class='hs-comment'>--     truth_indices = [[0, 1, 0],</span>
+<a name="line-4154"></a>                <span class='hs-comment'>--                      [1, 0, 0],</span>
+<a name="line-4155"></a>                <span class='hs-comment'>--                      [1, 0, 1],</span>
+<a name="line-4156"></a>                <span class='hs-comment'>--                      [1, 1, 0]]</span>
+<a name="line-4157"></a>                <span class='hs-comment'>--     truth_values = ["a", "b", "c", "a"]</span>
+<a name="line-4158"></a>                <span class='hs-comment'>--     truth_shape = [2, 2, 2]</span>
+<a name="line-4159"></a>                <span class='hs-comment'>--     normalize = true</span>
+<a name="line-4160"></a>                <span class='hs-comment'>-- </span>
+<a name="line-4161"></a>                <span class='hs-comment'>-- The output will be:</span>
+<a name="line-4162"></a>                <span class='hs-comment'>-- </span>
+<a name="line-4163"></a>                <span class='hs-comment'>--     // output is a 2x2 matrix with edit distances normalized by truth lengths.</span>
+<a name="line-4164"></a>                <span class='hs-comment'>--     output = [[inf, 1.0],  // (0,0): no truth, (0,1): no hypothesis</span>
+<a name="line-4165"></a>                <span class='hs-comment'>--               [0.5, 1.0]]  // (1,0): addition, (1,1): no hypothesis</span>
+<a name="line-4166"></a><span class='hs-definition'>editDistance</span> <span class='hs-varid'>hypothesis_indices</span> <span class='hs-varid'>hypothesis_values</span> <span class='hs-varid'>hypothesis_shape</span> <span class='hs-varid'>truth_indices</span>
+<a name="line-4167"></a>             <span class='hs-varid'>truth_values</span> <span class='hs-varid'>truth_shape</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4168"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"EditDistance"</span>
+<a name="line-4169"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-4170"></a>        <span class='hs-varid'>hypothesis_indices</span> <span class='hs-varid'>hypothesis_values</span> <span class='hs-varid'>hypothesis_shape</span> <span class='hs-varid'>truth_indices</span>
+<a name="line-4171"></a>        <span class='hs-varid'>truth_values</span> <span class='hs-varid'>truth_shape</span>
+<a name="line-4172"></a><span class='hs-comment'>{-
+<a name="line-4173"></a>attr {
+<a name="line-4174"></a>  default_value { b: true }
+<a name="line-4175"></a>  description: "boolean (if true, edit distances are normalized by length of truth).\n\nThe output is:"
+<a name="line-4176"></a>  name: "normalize"
+<a name="line-4177"></a>  type: "bool"
+<a name="line-4178"></a>}
+<a name="line-4179"></a>attr { name: "T" type: "type" }
+<a name="line-4180"></a>input_arg {
+<a name="line-4181"></a>  description: "The indices of the hypothesis list SparseTensor.\nThis is an N x R int64 matrix."
+<a name="line-4182"></a>  name: "hypothesis_indices"
+<a name="line-4183"></a>  type: DT_INT64
+<a name="line-4184"></a>}
+<a name="line-4185"></a>input_arg {
+<a name="line-4186"></a>  description: "The values of the hypothesis list SparseTensor.\nThis is an N-length vector."
+<a name="line-4187"></a>  name: "hypothesis_values"
+<a name="line-4188"></a>  type_attr: "T"
+<a name="line-4189"></a>}
+<a name="line-4190"></a>input_arg {
+<a name="line-4191"></a>  description: "The shape of the hypothesis list SparseTensor.\nThis is an R-length vector."
+<a name="line-4192"></a>  name: "hypothesis_shape"
+<a name="line-4193"></a>  type: DT_INT64
+<a name="line-4194"></a>}
+<a name="line-4195"></a>input_arg {
+<a name="line-4196"></a>  description: "The indices of the truth list SparseTensor.\nThis is an M x R int64 matrix."
+<a name="line-4197"></a>  name: "truth_indices"
+<a name="line-4198"></a>  type: DT_INT64
+<a name="line-4199"></a>}
+<a name="line-4200"></a>input_arg {
+<a name="line-4201"></a>  description: "The values of the truth list SparseTensor.\nThis is an M-length vector."
+<a name="line-4202"></a>  name: "truth_values"
+<a name="line-4203"></a>  type_attr: "T"
+<a name="line-4204"></a>}
+<a name="line-4205"></a>input_arg {
+<a name="line-4206"></a>  description: "truth indices, vector."
+<a name="line-4207"></a>  name: "truth_shape"
+<a name="line-4208"></a>  type: DT_INT64
+<a name="line-4209"></a>}
+<a name="line-4210"></a>output_arg {
+<a name="line-4211"></a>  description: "A dense float tensor with rank R - 1.\n\nFor the example input:\n\n    // hypothesis represents a 2x1 matrix with variable-length values:\n    //   (0,0) = [\"a\"]\n    //   (1,0) = [\"b\"]\n    hypothesis_indices = [[0, 0, 0],\n                          [1, 0, 0]]\n    hypothesis_values = [\"a\", \"b\"]\n    hypothesis_shape = [2, 1, 1]\n\n    // truth represents a 2x2 matrix with variable-length values:\n    //   (0,0) = []\n    //   (0,1) = [\"a\"]\n    //   (1,0) = [\"b\", \"c\"]\n    //   (1,1) = [\"a\"]\n    truth_indices = [[0, 1, 0],\n                     [1, 0, 0],\n                     [1, 0, 1],\n                     [1, 1, 0]]\n    truth_values = [\"a\", \"b\", \"c\", \"a\"]\n    truth_shape = [2, 2, 2]\n    normalize = true\n\nThe output will be:\n\n    // output is a 2x2 matrix with edit distances normalized by truth lengths.\n    output = [[inf, 1.0],  // (0,0): no truth, (0,1): no hypothesis\n              [0.5, 1.0]]  // (1,0): addition, (1,1): no hypothesis"
+<a name="line-4212"></a>  name: "output"
+<a name="line-4213"></a>  type: DT_FLOAT
+<a name="line-4214"></a>}
+<a name="line-4215"></a>-}</span>
+<a name="line-4216"></a>
+<a name="line-4217"></a><span class='hs-comment'>-- | Computes the number of incomplete elements in the given barrier.</span>
+<a name="line-4218"></a>
+<a name="line-4219"></a><a name="barrierIncompleteSize"></a><span class='hs-definition'>barrierIncompleteSize</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a barrier.</span>
+<a name="line-4220"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __size__: The number of incomplete elements (i.e. those with some of their value</span>
+<a name="line-4221"></a>                         <span class='hs-comment'>-- components not set) in the barrier.</span>
+<a name="line-4222"></a><span class='hs-definition'>barrierIncompleteSize</span> <span class='hs-varid'>handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4223"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BarrierIncompleteSize"</span><span class='hs-layout'>)</span>
+<a name="line-4224"></a>        <span class='hs-varid'>handle</span>
+<a name="line-4225"></a><span class='hs-comment'>{-
+<a name="line-4226"></a>input_arg {
+<a name="line-4227"></a>  description: "The handle to a barrier."
+<a name="line-4228"></a>  is_ref: true
+<a name="line-4229"></a>  name: "handle"
+<a name="line-4230"></a>  type: DT_STRING
+<a name="line-4231"></a>}
+<a name="line-4232"></a>output_arg {
+<a name="line-4233"></a>  description: "The number of incomplete elements (i.e. those with some of their value\ncomponents not set) in the barrier."
+<a name="line-4234"></a>  name: "size"
+<a name="line-4235"></a>  type: DT_INT32
+<a name="line-4236"></a>}
+<a name="line-4237"></a>-}</span>
+<a name="line-4238"></a>
+<a name="line-4239"></a><a name="threadUnsafeUnigramCandidateSampler"></a><span class='hs-comment'>-- | Generates labels for candidate sampling with a learned unigram distribution.</span>
+<a name="line-4240"></a><span class='hs-comment'>--</span>
+<a name="line-4241"></a><span class='hs-comment'>-- See explanations of candidate sampling and the data formats at</span>
+<a name="line-4242"></a><span class='hs-comment'>-- go/candidate-sampling.</span>
+<a name="line-4243"></a><span class='hs-comment'>-- </span>
+<a name="line-4244"></a><span class='hs-comment'>-- For each batch, this op picks a single set of sampled candidate labels.</span>
+<a name="line-4245"></a><span class='hs-comment'>-- </span>
+<a name="line-4246"></a><span class='hs-comment'>-- The advantages of sampling candidates per-batch are simplicity and the</span>
+<a name="line-4247"></a><span class='hs-comment'>-- possibility of efficient dense matrix multiplication. The disadvantage is that</span>
+<a name="line-4248"></a><span class='hs-comment'>-- the sampled candidates must be chosen independently of the context and of the</span>
+<a name="line-4249"></a><span class='hs-comment'>-- true labels.</span>
+<a name="line-4250"></a><span class='hs-definition'>threadUnsafeUnigramCandidateSampler</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_sampled__: Number of candidates to randomly sample per batch.</span>
+<a name="line-4251"></a>                                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_true__: Number of true labels per context.</span>
+<a name="line-4252"></a>                                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __range_max__: The sampler will sample integers from the interval [0, range_max).</span>
+<a name="line-4253"></a>                                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __unique__: If unique is true, we sample with rejection, so that all sampled</span>
+<a name="line-4254"></a>                                               <span class='hs-comment'>-- candidates in a batch are unique. This requires some approximation to</span>
+<a name="line-4255"></a>                                               <span class='hs-comment'>-- estimate the post-rejection sampling probabilities.</span>
+<a name="line-4256"></a>                                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __true_classes__: A batch_size * num_true matrix, in which each row contains the</span>
+<a name="line-4257"></a>                                                                   <span class='hs-comment'>-- IDs of the num_true target_classes in the corresponding original label.</span>
+<a name="line-4258"></a>                                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-4259"></a>                                           <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>,</span>
+<a name="line-4260"></a>                                           <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span>
+<a name="line-4261"></a>                                       <span class='hs-comment'>-- ^ (__sampled_candidates__, __true_expected_count__, __sampled_expected_count__)</span>
+<a name="line-4262"></a>                                       <span class='hs-comment'>--</span>
+<a name="line-4263"></a>                                       <span class='hs-comment'>-- * __sampled_candidates__: A vector of length num_sampled, in which each element is</span>
+<a name="line-4264"></a>                                       <span class='hs-comment'>-- the ID of a sampled candidate.</span>
+<a name="line-4265"></a>                                       <span class='hs-comment'>--</span>
+<a name="line-4266"></a>                                       <span class='hs-comment'>-- * __true_expected_count__: A batch_size * num_true matrix, representing</span>
+<a name="line-4267"></a>                                       <span class='hs-comment'>-- the number of times each candidate is expected to occur in a batch</span>
+<a name="line-4268"></a>                                       <span class='hs-comment'>-- of sampled candidates. If unique=true, then this is a probability.</span>
+<a name="line-4269"></a>                                       <span class='hs-comment'>--</span>
+<a name="line-4270"></a>                                       <span class='hs-comment'>-- * __sampled_expected_count__: A vector of length num_sampled, for each sampled</span>
+<a name="line-4271"></a>                                       <span class='hs-comment'>-- candidate representing the number of times the candidate is expected</span>
+<a name="line-4272"></a>                                       <span class='hs-comment'>-- to occur in a batch of sampled candidates.  If unique=true, then this is a</span>
+<a name="line-4273"></a>                                       <span class='hs-comment'>-- probability.</span>
+<a name="line-4274"></a><span class='hs-definition'>threadUnsafeUnigramCandidateSampler</span> <span class='hs-varid'>num_sampled</span> <span class='hs-varid'>num_true</span> <span class='hs-varid'>range_max</span> <span class='hs-varid'>unique</span>
+<a name="line-4275"></a>                                    <span class='hs-varid'>true_classes</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4276"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ThreadUnsafeUnigramCandidateSampler"</span>
+<a name="line-4277"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_sampled"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_sampled</span>
+<a name="line-4278"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_true"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_true</span>
+<a name="line-4279"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"range_max"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>range_max</span>
+<a name="line-4280"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"unique"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>unique</span><span class='hs-layout'>)</span>
+<a name="line-4281"></a>        <span class='hs-varid'>true_classes</span>
+<a name="line-4282"></a><span class='hs-comment'>{-
+<a name="line-4283"></a>attr {
+<a name="line-4284"></a>  description: "Number of true labels per context."
+<a name="line-4285"></a>  has_minimum: true
+<a name="line-4286"></a>  minimum: 1
+<a name="line-4287"></a>  name: "num_true"
+<a name="line-4288"></a>  type: "int"
+<a name="line-4289"></a>}
+<a name="line-4290"></a>attr {
+<a name="line-4291"></a>  description: "Number of candidates to randomly sample per batch."
+<a name="line-4292"></a>  has_minimum: true
+<a name="line-4293"></a>  minimum: 1
+<a name="line-4294"></a>  name: "num_sampled"
+<a name="line-4295"></a>  type: "int"
+<a name="line-4296"></a>}
+<a name="line-4297"></a>attr {
+<a name="line-4298"></a>  description: "If unique is true, we sample with rejection, so that all sampled\ncandidates in a batch are unique. This requires some approximation to\nestimate the post-rejection sampling probabilities."
+<a name="line-4299"></a>  name: "unique"
+<a name="line-4300"></a>  type: "bool"
+<a name="line-4301"></a>}
+<a name="line-4302"></a>attr {
+<a name="line-4303"></a>  description: "The sampler will sample integers from the interval [0, range_max)."
+<a name="line-4304"></a>  has_minimum: true
+<a name="line-4305"></a>  minimum: 1
+<a name="line-4306"></a>  name: "range_max"
+<a name="line-4307"></a>  type: "int"
+<a name="line-4308"></a>}
+<a name="line-4309"></a>attr {
+<a name="line-4310"></a>  default_value { i: 0 }
+<a name="line-4311"></a>  description: "If either seed or seed2 are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-4312"></a>  name: "seed"
+<a name="line-4313"></a>  type: "int"
+<a name="line-4314"></a>}
+<a name="line-4315"></a>attr {
+<a name="line-4316"></a>  default_value { i: 0 }
+<a name="line-4317"></a>  description: "An second seed to avoid seed collision."
+<a name="line-4318"></a>  name: "seed2"
+<a name="line-4319"></a>  type: "int"
+<a name="line-4320"></a>}
+<a name="line-4321"></a>input_arg {
+<a name="line-4322"></a>  description: "A batch_size * num_true matrix, in which each row contains the\nIDs of the num_true target_classes in the corresponding original label."
+<a name="line-4323"></a>  name: "true_classes"
+<a name="line-4324"></a>  type: DT_INT64
+<a name="line-4325"></a>}
+<a name="line-4326"></a>output_arg {
+<a name="line-4327"></a>  description: "A vector of length num_sampled, in which each element is\nthe ID of a sampled candidate."
+<a name="line-4328"></a>  name: "sampled_candidates"
+<a name="line-4329"></a>  type: DT_INT64
+<a name="line-4330"></a>}
+<a name="line-4331"></a>output_arg {
+<a name="line-4332"></a>  description: "A batch_size * num_true matrix, representing\nthe number of times each candidate is expected to occur in a batch\nof sampled candidates. If unique=true, then this is a probability."
+<a name="line-4333"></a>  name: "true_expected_count"
+<a name="line-4334"></a>  type: DT_FLOAT
+<a name="line-4335"></a>}
+<a name="line-4336"></a>output_arg {
+<a name="line-4337"></a>  description: "A vector of length num_sampled, for each sampled\ncandidate representing the number of times the candidate is expected\nto occur in a batch of sampled candidates.  If unique=true, then this is a\nprobability."
+<a name="line-4338"></a>  name: "sampled_expected_count"
+<a name="line-4339"></a>  type: DT_FLOAT
+<a name="line-4340"></a>}
+<a name="line-4341"></a>-}</span>
+<a name="line-4342"></a>
+<a name="line-4343"></a><span class='hs-comment'>-- | Computes the number of complete elements in the given barrier.</span>
+<a name="line-4344"></a>
+<a name="line-4345"></a><a name="barrierReadySize"></a><span class='hs-definition'>barrierReadySize</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a barrier.</span>
+<a name="line-4346"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __size__: The number of complete elements (i.e. those with all of their value</span>
+<a name="line-4347"></a>                    <span class='hs-comment'>-- components set) in the barrier.</span>
+<a name="line-4348"></a><span class='hs-definition'>barrierReadySize</span> <span class='hs-varid'>handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4349"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BarrierReadySize"</span><span class='hs-layout'>)</span>
+<a name="line-4350"></a>        <span class='hs-varid'>handle</span>
+<a name="line-4351"></a><span class='hs-comment'>{-
+<a name="line-4352"></a>input_arg {
+<a name="line-4353"></a>  description: "The handle to a barrier."
+<a name="line-4354"></a>  is_ref: true
+<a name="line-4355"></a>  name: "handle"
+<a name="line-4356"></a>  type: DT_STRING
+<a name="line-4357"></a>}
+<a name="line-4358"></a>output_arg {
+<a name="line-4359"></a>  description: "The number of complete elements (i.e. those with all of their value\ncomponents set) in the barrier."
+<a name="line-4360"></a>  name: "size"
+<a name="line-4361"></a>  type: DT_INT32
+<a name="line-4362"></a>}
+<a name="line-4363"></a>-}</span>
+<a name="line-4364"></a>
+<a name="line-4365"></a><a name="barrierClose"></a><span class='hs-comment'>-- | Closes the given barrier.</span>
+<a name="line-4366"></a><span class='hs-comment'>--</span>
+<a name="line-4367"></a><span class='hs-comment'>-- This operation signals that no more new elements will be inserted in the</span>
+<a name="line-4368"></a><span class='hs-comment'>-- given barrier. Subsequent InsertMany that try to introduce a new key will fail.</span>
+<a name="line-4369"></a><span class='hs-comment'>-- Subsequent InsertMany operations that just add missing components to already</span>
+<a name="line-4370"></a><span class='hs-comment'>-- existing elements will continue to succeed. Subsequent TakeMany operations will</span>
+<a name="line-4371"></a><span class='hs-comment'>-- continue to succeed if sufficient completed elements remain in the barrier.</span>
+<a name="line-4372"></a><span class='hs-comment'>-- Subsequent TakeMany operations that would block will fail immediately.</span>
+<a name="line-4373"></a><span class='hs-definition'>barrierClose</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a barrier.</span>
+<a name="line-4374"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-4375"></a><span class='hs-definition'>barrierClose</span> <span class='hs-varid'>handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4376"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BarrierClose"</span><span class='hs-layout'>)</span>
+<a name="line-4377"></a>        <span class='hs-varid'>handle</span>
+<a name="line-4378"></a><span class='hs-comment'>{-
+<a name="line-4379"></a>attr {
+<a name="line-4380"></a>  default_value { b: false }
+<a name="line-4381"></a>  description: "If true, all pending enqueue requests that are\nblocked on the barrier\'s queue will be cancelled. InsertMany will fail, even\nif no new key is introduced."
+<a name="line-4382"></a>  name: "cancel_pending_enqueues"
+<a name="line-4383"></a>  type: "bool"
+<a name="line-4384"></a>}
+<a name="line-4385"></a>input_arg {
+<a name="line-4386"></a>  description: "The handle to a barrier."
+<a name="line-4387"></a>  is_ref: true
+<a name="line-4388"></a>  name: "handle"
+<a name="line-4389"></a>  type: DT_STRING
+<a name="line-4390"></a>}
+<a name="line-4391"></a>-}</span>
+<a name="line-4392"></a>
+<a name="line-4393"></a><span class='hs-comment'>-- | A Reader that outputs the lines of a file delimited by '\n'.</span>
+<a name="line-4394"></a>
+<a name="line-4395"></a><a name="textLineReader"></a><span class='hs-definition'>textLineReader</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __reader_handle__: The handle to reference the Reader.</span>
+<a name="line-4396"></a><span class='hs-definition'>textLineReader</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4397"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TextLineReader"</span><span class='hs-layout'>)</span>
+<a name="line-4398"></a>        
+<a name="line-4399"></a><span class='hs-comment'>{-
+<a name="line-4400"></a>attr {
+<a name="line-4401"></a>  default_value { i: 0 }
+<a name="line-4402"></a>  description: "Number of lines to skip from the beginning of every file."
+<a name="line-4403"></a>  name: "skip_header_lines"
+<a name="line-4404"></a>  type: "int"
+<a name="line-4405"></a>}
+<a name="line-4406"></a>attr {
+<a name="line-4407"></a>  default_value { s: "" }
+<a name="line-4408"></a>  description: "If non-empty, this reader is placed in the given container.\nOtherwise, a default container is used."
+<a name="line-4409"></a>  name: "container"
+<a name="line-4410"></a>  type: "string"
+<a name="line-4411"></a>}
+<a name="line-4412"></a>attr {
+<a name="line-4413"></a>  default_value { s: "" }
+<a name="line-4414"></a>  description: "If non-empty, this reader is named in the given bucket\nwith this shared_name. Otherwise, the node name is used instead."
+<a name="line-4415"></a>  name: "shared_name"
+<a name="line-4416"></a>  type: "string"
+<a name="line-4417"></a>}
+<a name="line-4418"></a>output_arg {
+<a name="line-4419"></a>  description: "The handle to reference the Reader."
+<a name="line-4420"></a>  is_ref: true
+<a name="line-4421"></a>  name: "reader_handle"
+<a name="line-4422"></a>  type: DT_STRING
+<a name="line-4423"></a>}
+<a name="line-4424"></a>-}</span>
+<a name="line-4425"></a>
+<a name="line-4426"></a><a name="fFT3D"></a><span class='hs-comment'>-- | Compute the 3-dimensional discrete Fourier Transform over the inner-most 3</span>
+<a name="line-4427"></a><span class='hs-comment'>--</span>
+<a name="line-4428"></a><span class='hs-comment'>-- dimensions of `input`.</span>
+<a name="line-4429"></a><span class='hs-definition'>fFT3D</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __input__: A complex64 tensor.</span>
+<a name="line-4430"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __output__: A complex64 tensor of the same shape as `input`. The inner-most 3</span>
+<a name="line-4431"></a>         <span class='hs-comment'>-- dimensions of `input` are replaced with their 3D Fourier Transform.</span>
+<a name="line-4432"></a><span class='hs-definition'>fFT3D</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4433"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"FFT3D"</span><span class='hs-layout'>)</span>
+<a name="line-4434"></a>        <span class='hs-varid'>input</span>
+<a name="line-4435"></a><span class='hs-comment'>{-
+<a name="line-4436"></a>input_arg {
+<a name="line-4437"></a>  description: "A complex64 tensor." name: "input" type: DT_COMPLEX64
+<a name="line-4438"></a>}
+<a name="line-4439"></a>output_arg {
+<a name="line-4440"></a>  description: "A complex64 tensor of the same shape as `input`. The inner-most 3\ndimensions of `input` are replaced with their 3D Fourier Transform."
+<a name="line-4441"></a>  name: "output"
+<a name="line-4442"></a>  type: DT_COMPLEX64
+<a name="line-4443"></a>}
+<a name="line-4444"></a>-}</span>
+<a name="line-4445"></a>
+<a name="line-4446"></a><a name="refExit"></a><span class='hs-comment'>-- | Exits the current frame to its parent frame.</span>
+<a name="line-4447"></a><span class='hs-comment'>--</span>
+<a name="line-4448"></a><span class='hs-comment'>-- Exit makes its input `data` available to the parent frame.</span>
+<a name="line-4449"></a><span class='hs-definition'>refExit</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-4450"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__: The tensor to be made available to the parent frame.</span>
+<a name="line-4451"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The same tensor as `data`.</span>
+<a name="line-4452"></a><span class='hs-definition'>refExit</span> <span class='hs-varid'>data'</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4453"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RefExit"</span>
+<a name="line-4454"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-4455"></a>        <span class='hs-varid'>data'</span>
+<a name="line-4456"></a><span class='hs-comment'>{-
+<a name="line-4457"></a>attr { name: "T" type: "type" }
+<a name="line-4458"></a>input_arg {
+<a name="line-4459"></a>  description: "The tensor to be made available to the parent frame."
+<a name="line-4460"></a>  is_ref: true
+<a name="line-4461"></a>  name: "data"
+<a name="line-4462"></a>  type_attr: "T"
+<a name="line-4463"></a>}
+<a name="line-4464"></a>output_arg {
+<a name="line-4465"></a>  description: "The same tensor as `data`."
+<a name="line-4466"></a>  is_ref: true
+<a name="line-4467"></a>  name: "output"
+<a name="line-4468"></a>  type_attr: "T"
+<a name="line-4469"></a>}
+<a name="line-4470"></a>-}</span>
+<a name="line-4471"></a>
+<a name="line-4472"></a><span class='hs-comment'>-- | Computes exponential of x element-wise.  \\(y = e^x\\).</span>
+<a name="line-4473"></a>
+<a name="line-4474"></a><a name="exp"></a><span class='hs-definition'>exp</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-4475"></a>                                            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-4476"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-4477"></a>                                            <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-4478"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-4479"></a><span class='hs-definition'>exp</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4480"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Exp"</span>
+<a name="line-4481"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-4482"></a>        <span class='hs-varid'>x</span>
+<a name="line-4483"></a><span class='hs-comment'>{-
+<a name="line-4484"></a>attr {
+<a name="line-4485"></a>  allowed_values {
+<a name="line-4486"></a>    list {
+<a name="line-4487"></a>      type: DT_HALF
+<a name="line-4488"></a>      type: DT_FLOAT
+<a name="line-4489"></a>      type: DT_DOUBLE
+<a name="line-4490"></a>      type: DT_COMPLEX64
+<a name="line-4491"></a>      type: DT_COMPLEX128
+<a name="line-4492"></a>    }
+<a name="line-4493"></a>  }
+<a name="line-4494"></a>  name: "T"
+<a name="line-4495"></a>  type: "type"
+<a name="line-4496"></a>}
+<a name="line-4497"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-4498"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-4499"></a>-}</span>
+<a name="line-4500"></a>
+<a name="line-4501"></a><a name="restoreSlice"></a><span class='hs-comment'>-- | Restores a tensor from checkpoint files.</span>
+<a name="line-4502"></a><span class='hs-comment'>--</span>
+<a name="line-4503"></a><span class='hs-comment'>-- This is like `Restore` except that restored tensor can be listed as filling</span>
+<a name="line-4504"></a><span class='hs-comment'>-- only a slice of a larger tensor.  `shape_and_slice` specifies the shape of the</span>
+<a name="line-4505"></a><span class='hs-comment'>-- larger tensor and the slice that the restored tensor covers.</span>
+<a name="line-4506"></a><span class='hs-comment'>-- </span>
+<a name="line-4507"></a><span class='hs-comment'>-- The `shape_and_slice` input has the same format as the</span>
+<a name="line-4508"></a><span class='hs-comment'>-- elements of the `shapes_and_slices` input of the `SaveSlices` op.</span>
+<a name="line-4509"></a><span class='hs-definition'>restoreSlice</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>dt</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dt</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-4510"></a>                <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __file_pattern__: Must have a single element. The pattern of the files from</span>
+<a name="line-4511"></a>                                                     <span class='hs-comment'>-- which we read the tensor.</span>
+<a name="line-4512"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __tensor_name__: Must have a single element. The name of the tensor to be</span>
+<a name="line-4513"></a>                                                        <span class='hs-comment'>-- restored.</span>
+<a name="line-4514"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __shape_and_slice__: Scalar. The shapes and slice specifications to use when</span>
+<a name="line-4515"></a>                                                        <span class='hs-comment'>-- restoring a tensors.</span>
+<a name="line-4516"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dt</span> <span class='hs-comment'>-- ^ __tensor__: The restored tensor.</span>
+<a name="line-4517"></a><span class='hs-definition'>restoreSlice</span> <span class='hs-varid'>file_pattern</span> <span class='hs-varid'>tensor_name</span> <span class='hs-varid'>shape_and_slice</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4518"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RestoreSlice"</span>
+<a name="line-4519"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dt"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dt</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-4520"></a>        <span class='hs-varid'>file_pattern</span> <span class='hs-varid'>tensor_name</span> <span class='hs-varid'>shape_and_slice</span>
+<a name="line-4521"></a><span class='hs-comment'>{-
+<a name="line-4522"></a>attr {
+<a name="line-4523"></a>  description: "The type of the tensor to be restored."
+<a name="line-4524"></a>  name: "dt"
+<a name="line-4525"></a>  type: "type"
+<a name="line-4526"></a>}
+<a name="line-4527"></a>attr {
+<a name="line-4528"></a>  default_value { i: -1 }
+<a name="line-4529"></a>  description: "Index of file to open first if multiple files match\n`file_pattern`. See the documentation for `Restore`."
+<a name="line-4530"></a>  name: "preferred_shard"
+<a name="line-4531"></a>  type: "int"
+<a name="line-4532"></a>}
+<a name="line-4533"></a>input_arg {
+<a name="line-4534"></a>  description: "Must have a single element. The pattern of the files from\nwhich we read the tensor."
+<a name="line-4535"></a>  name: "file_pattern"
+<a name="line-4536"></a>  type: DT_STRING
+<a name="line-4537"></a>}
+<a name="line-4538"></a>input_arg {
+<a name="line-4539"></a>  description: "Must have a single element. The name of the tensor to be\nrestored."
+<a name="line-4540"></a>  name: "tensor_name"
+<a name="line-4541"></a>  type: DT_STRING
+<a name="line-4542"></a>}
+<a name="line-4543"></a>input_arg {
+<a name="line-4544"></a>  description: "Scalar. The shapes and slice specifications to use when\nrestoring a tensors."
+<a name="line-4545"></a>  name: "shape_and_slice"
+<a name="line-4546"></a>  type: DT_STRING
+<a name="line-4547"></a>}
+<a name="line-4548"></a>output_arg {
+<a name="line-4549"></a>  description: "The restored tensor." name: "tensor" type_attr: "dt"
+<a name="line-4550"></a>}
+<a name="line-4551"></a>-}</span>
+<a name="line-4552"></a>
+<a name="line-4553"></a><a name="conj"></a><span class='hs-comment'>-- | Returns the complex conjugate of a complex number.</span>
+<a name="line-4554"></a><span class='hs-comment'>--</span>
+<a name="line-4555"></a><span class='hs-comment'>-- Given a tensor `input` of complex numbers, this operation returns a tensor of</span>
+<a name="line-4556"></a><span class='hs-comment'>-- complex numbers that are the complex conjugate of each element in `input`. The</span>
+<a name="line-4557"></a><span class='hs-comment'>-- complex numbers in `input` must be of the form \\(a + bj\\), where *a* is the</span>
+<a name="line-4558"></a><span class='hs-comment'>-- real part and *b* is the imaginary part.</span>
+<a name="line-4559"></a><span class='hs-comment'>-- </span>
+<a name="line-4560"></a><span class='hs-comment'>-- The complex conjugate returned by this operation is of the form \\(a - bj\\).</span>
+<a name="line-4561"></a><span class='hs-comment'>-- </span>
+<a name="line-4562"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-4563"></a><span class='hs-comment'>-- </span>
+<a name="line-4564"></a><span class='hs-comment'>-- ```</span>
+<a name="line-4565"></a><span class='hs-comment'>-- # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]</span>
+<a name="line-4566"></a><span class='hs-comment'>-- tf.conj(input) ==&gt; [-2.25 - 4.75j, 3.25 - 5.75j]</span>
+<a name="line-4567"></a><span class='hs-comment'>-- ```</span>
+<a name="line-4568"></a><span class='hs-definition'>conj</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-4569"></a>                                             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-4570"></a>        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-4571"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-4572"></a><span class='hs-definition'>conj</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4573"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Conj"</span>
+<a name="line-4574"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-4575"></a>        <span class='hs-varid'>input</span>
+<a name="line-4576"></a><span class='hs-comment'>{-
+<a name="line-4577"></a>attr {
+<a name="line-4578"></a>  allowed_values { list { type: DT_COMPLEX64 type: DT_COMPLEX128 } }
+<a name="line-4579"></a>  default_value { type: DT_COMPLEX64 }
+<a name="line-4580"></a>  name: "T"
+<a name="line-4581"></a>  type: "type"
+<a name="line-4582"></a>}
+<a name="line-4583"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-4584"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-4585"></a>-}</span>
+<a name="line-4586"></a>
+<a name="line-4587"></a><span class='hs-comment'>-- | Computes the gradient of nearest neighbor interpolation.</span>
+<a name="line-4588"></a>
+<a name="line-4589"></a><a name="resizeNearestNeighborGrad"></a><span class='hs-definition'>resizeNearestNeighborGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-4590"></a>                                               <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-4591"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-4592"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-4593"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-4594"></a>                                                       <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-4595"></a>                             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grads__: 4-D with shape `[batch, height, width, channels]`.</span>
+<a name="line-4596"></a>                             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __size__: = A 1-D int32 Tensor of 2 elements: `orig_height, orig_width`. The</span>
+<a name="line-4597"></a>                                                         <span class='hs-comment'>-- original input size.</span>
+<a name="line-4598"></a>                             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 4-D with shape `[batch, orig_height, orig_width, channels]`. Gradients</span>
+<a name="line-4599"></a>                             <span class='hs-comment'>-- with respect to the input image.</span>
+<a name="line-4600"></a><span class='hs-definition'>resizeNearestNeighborGrad</span> <span class='hs-varid'>grads</span> <span class='hs-varid'>size</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4601"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ResizeNearestNeighborGrad"</span>
+<a name="line-4602"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-4603"></a>        <span class='hs-varid'>grads</span> <span class='hs-varid'>size</span>
+<a name="line-4604"></a><span class='hs-comment'>{-
+<a name="line-4605"></a>attr {
+<a name="line-4606"></a>  allowed_values {
+<a name="line-4607"></a>    list {
+<a name="line-4608"></a>      type: DT_UINT8
+<a name="line-4609"></a>      type: DT_INT8
+<a name="line-4610"></a>      type: DT_INT32
+<a name="line-4611"></a>      type: DT_HALF
+<a name="line-4612"></a>      type: DT_FLOAT
+<a name="line-4613"></a>      type: DT_DOUBLE
+<a name="line-4614"></a>    }
+<a name="line-4615"></a>  }
+<a name="line-4616"></a>  name: "T"
+<a name="line-4617"></a>  type: "type"
+<a name="line-4618"></a>}
+<a name="line-4619"></a>attr {
+<a name="line-4620"></a>  default_value { b: false }
+<a name="line-4621"></a>  description: "If true, rescale grads by (orig_height - 1) / (height - 1), which\nexactly aligns the 4 corners of grads and original_image. If false, rescale by\norig_height / height. Treat similarly the width dimension."
+<a name="line-4622"></a>  name: "align_corners"
+<a name="line-4623"></a>  type: "bool"
+<a name="line-4624"></a>}
+<a name="line-4625"></a>input_arg {
+<a name="line-4626"></a>  description: "4-D with shape `[batch, height, width, channels]`."
+<a name="line-4627"></a>  name: "grads"
+<a name="line-4628"></a>  type_attr: "T"
+<a name="line-4629"></a>}
+<a name="line-4630"></a>input_arg {
+<a name="line-4631"></a>  description: "= A 1-D int32 Tensor of 2 elements: `orig_height, orig_width`. The\noriginal input size."
+<a name="line-4632"></a>  name: "size"
+<a name="line-4633"></a>  type: DT_INT32
+<a name="line-4634"></a>}
+<a name="line-4635"></a>output_arg {
+<a name="line-4636"></a>  description: "4-D with shape `[batch, orig_height, orig_width, channels]`. Gradients\nwith respect to the input image."
+<a name="line-4637"></a>  name: "output"
+<a name="line-4638"></a>  type_attr: "T"
+<a name="line-4639"></a>}
+<a name="line-4640"></a>-}</span>
+<a name="line-4641"></a>
+<a name="line-4642"></a><a name="tensorArrayClose"></a><span class='hs-comment'>-- | Delete the TensorArray from its resource container.  This enables</span>
+<a name="line-4643"></a><span class='hs-comment'>--</span>
+<a name="line-4644"></a><span class='hs-comment'>-- the user to close and release the resource in the middle of a step/run.</span>
+<a name="line-4645"></a><span class='hs-definition'>tensorArrayClose</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a TensorArray (output of TensorArray or TensorArrayGrad).</span>
+<a name="line-4646"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-4647"></a><span class='hs-definition'>tensorArrayClose</span> <span class='hs-varid'>handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4648"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TensorArrayClose"</span><span class='hs-layout'>)</span>
+<a name="line-4649"></a>        <span class='hs-varid'>handle</span>
+<a name="line-4650"></a><span class='hs-comment'>{-
+<a name="line-4651"></a>input_arg {
+<a name="line-4652"></a>  description: "The handle to a TensorArray (output of TensorArray or TensorArrayGrad)."
+<a name="line-4653"></a>  is_ref: true
+<a name="line-4654"></a>  name: "handle"
+<a name="line-4655"></a>  type: DT_STRING
+<a name="line-4656"></a>}
+<a name="line-4657"></a>-}</span>
+<a name="line-4658"></a>
+<a name="line-4659"></a><span class='hs-comment'>-- | Computes atan of x element-wise.</span>
+<a name="line-4660"></a>
+<a name="line-4661"></a><a name="atan"></a><span class='hs-definition'>atan</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-4662"></a>                                             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-4663"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-4664"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-4665"></a>                                             <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-4666"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-4667"></a><span class='hs-definition'>atan</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4668"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Atan"</span>
+<a name="line-4669"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-4670"></a>        <span class='hs-varid'>x</span>
+<a name="line-4671"></a><span class='hs-comment'>{-
+<a name="line-4672"></a>attr {
+<a name="line-4673"></a>  allowed_values {
+<a name="line-4674"></a>    list {
+<a name="line-4675"></a>      type: DT_HALF
+<a name="line-4676"></a>      type: DT_FLOAT
+<a name="line-4677"></a>      type: DT_DOUBLE
+<a name="line-4678"></a>      type: DT_INT32
+<a name="line-4679"></a>      type: DT_INT64
+<a name="line-4680"></a>      type: DT_COMPLEX64
+<a name="line-4681"></a>      type: DT_COMPLEX128
+<a name="line-4682"></a>    }
+<a name="line-4683"></a>  }
+<a name="line-4684"></a>  name: "T"
+<a name="line-4685"></a>  type: "type"
+<a name="line-4686"></a>}
+<a name="line-4687"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-4688"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-4689"></a>-}</span>
+<a name="line-4690"></a>
+<a name="line-4691"></a><span class='hs-comment'>-- | Get the current size of the TensorArray.</span>
+<a name="line-4692"></a>
+<a name="line-4693"></a><a name="tensorArraySize"></a><span class='hs-definition'>tensorArraySize</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a TensorArray (output of TensorArray or TensorArrayGrad).</span>
+<a name="line-4694"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __flow_in__: A float scalar that enforces proper chaining of operations.</span>
+<a name="line-4695"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __size__: The current size of the TensorArray.</span>
+<a name="line-4696"></a><span class='hs-definition'>tensorArraySize</span> <span class='hs-varid'>handle</span> <span class='hs-varid'>flow_in</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4697"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TensorArraySize"</span><span class='hs-layout'>)</span>
+<a name="line-4698"></a>        <span class='hs-varid'>handle</span> <span class='hs-varid'>flow_in</span>
+<a name="line-4699"></a><span class='hs-comment'>{-
+<a name="line-4700"></a>input_arg {
+<a name="line-4701"></a>  description: "The handle to a TensorArray (output of TensorArray or TensorArrayGrad)."
+<a name="line-4702"></a>  is_ref: true
+<a name="line-4703"></a>  name: "handle"
+<a name="line-4704"></a>  type: DT_STRING
+<a name="line-4705"></a>}
+<a name="line-4706"></a>input_arg {
+<a name="line-4707"></a>  description: "A float scalar that enforces proper chaining of operations."
+<a name="line-4708"></a>  name: "flow_in"
+<a name="line-4709"></a>  type: DT_FLOAT
+<a name="line-4710"></a>}
+<a name="line-4711"></a>output_arg {
+<a name="line-4712"></a>  description: "The current size of the TensorArray."
+<a name="line-4713"></a>  name: "size"
+<a name="line-4714"></a>  type: DT_INT32
+<a name="line-4715"></a>}
+<a name="line-4716"></a>-}</span>
+<a name="line-4717"></a>
+<a name="line-4718"></a><a name="tensorArrayConcat"></a><span class='hs-comment'>-- | Concat the elements from the TensorArray into value `value`.</span>
+<a name="line-4719"></a><span class='hs-comment'>--</span>
+<a name="line-4720"></a><span class='hs-comment'>-- Takes `T` elements of shapes</span>
+<a name="line-4721"></a><span class='hs-comment'>-- </span>
+<a name="line-4722"></a><span class='hs-comment'>--   ```</span>
+<a name="line-4723"></a><span class='hs-comment'>--   (n0 x d0 x d1 x ...), (n1 x d0 x d1 x ...), ..., (n(T-1) x d0 x d1 x ...)</span>
+<a name="line-4724"></a><span class='hs-comment'>--   ```</span>
+<a name="line-4725"></a><span class='hs-comment'>-- </span>
+<a name="line-4726"></a><span class='hs-comment'>-- and concatenates them into a Tensor of shape:</span>
+<a name="line-4727"></a><span class='hs-comment'>-- </span>
+<a name="line-4728"></a><span class='hs-comment'>--   ```(n0 + n1 + ... + n(T-1) x d0 x d1 x ...)```</span>
+<a name="line-4729"></a><span class='hs-comment'>-- </span>
+<a name="line-4730"></a><span class='hs-comment'>-- All elements must have the same shape (excepting the first dimension).</span>
+<a name="line-4731"></a><span class='hs-definition'>tensorArrayConcat</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-4732"></a>                     <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a TensorArray.</span>
+<a name="line-4733"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __flow_in__: A float scalar that enforces proper chaining of operations.</span>
+<a name="line-4734"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-4735"></a>                     <span class='hs-comment'>-- ^ (__value__, __lengths__)</span>
+<a name="line-4736"></a>                     <span class='hs-comment'>--</span>
+<a name="line-4737"></a>                     <span class='hs-comment'>-- * __value__: All of the elements in the TensorArray, concatenated along the first</span>
+<a name="line-4738"></a>                     <span class='hs-comment'>-- axis.</span>
+<a name="line-4739"></a>                     <span class='hs-comment'>--</span>
+<a name="line-4740"></a>                     <span class='hs-comment'>-- * __lengths__: A vector of the row sizes of the original T elements in the</span>
+<a name="line-4741"></a>                     <span class='hs-comment'>-- value output.  In the example above, this would be the values:</span>
+<a name="line-4742"></a>                     <span class='hs-comment'>-- `(n1, n2, ..., n(T-1))`.</span>
+<a name="line-4743"></a><span class='hs-definition'>tensorArrayConcat</span> <span class='hs-varid'>handle</span> <span class='hs-varid'>flow_in</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4744"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TensorArrayConcat"</span>
+<a name="line-4745"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-4746"></a>        <span class='hs-varid'>handle</span> <span class='hs-varid'>flow_in</span>
+<a name="line-4747"></a><span class='hs-comment'>{-
+<a name="line-4748"></a>attr {
+<a name="line-4749"></a>  description: "The type of the elem that is returned."
+<a name="line-4750"></a>  name: "dtype"
+<a name="line-4751"></a>  type: "type"
+<a name="line-4752"></a>}
+<a name="line-4753"></a>attr {
+<a name="line-4754"></a>  default_value { shape { unknown_rank: true } }
+<a name="line-4755"></a>  description: "The expected shape of an element, if known,\nexcluding the first dimension. Used to validate the shapes of\nTensorArray elements. If this shape is not fully specified, concatenating\nzero-size TensorArrays is an error."
+<a name="line-4756"></a>  name: "element_shape_except0"
+<a name="line-4757"></a>  type: "shape"
+<a name="line-4758"></a>}
+<a name="line-4759"></a>input_arg {
+<a name="line-4760"></a>  description: "The handle to a TensorArray."
+<a name="line-4761"></a>  is_ref: true
+<a name="line-4762"></a>  name: "handle"
+<a name="line-4763"></a>  type: DT_STRING
+<a name="line-4764"></a>}
+<a name="line-4765"></a>input_arg {
+<a name="line-4766"></a>  description: "A float scalar that enforces proper chaining of operations."
+<a name="line-4767"></a>  name: "flow_in"
+<a name="line-4768"></a>  type: DT_FLOAT
+<a name="line-4769"></a>}
+<a name="line-4770"></a>output_arg {
+<a name="line-4771"></a>  description: "All of the elements in the TensorArray, concatenated along the first\naxis."
+<a name="line-4772"></a>  name: "value"
+<a name="line-4773"></a>  type_attr: "dtype"
+<a name="line-4774"></a>}
+<a name="line-4775"></a>output_arg {
+<a name="line-4776"></a>  description: "A vector of the row sizes of the original T elements in the\nvalue output.  In the example above, this would be the values:\n`(n1, n2, ..., n(T-1))`."
+<a name="line-4777"></a>  name: "lengths"
+<a name="line-4778"></a>  type: DT_INT64
+<a name="line-4779"></a>}
+<a name="line-4780"></a>-}</span>
+<a name="line-4781"></a>
+<a name="line-4782"></a><a name="lRN"></a><span class='hs-comment'>-- | Local Response Normalization.</span>
+<a name="line-4783"></a><span class='hs-comment'>--</span>
+<a name="line-4784"></a><span class='hs-comment'>-- The 4-D `input` tensor is treated as a 3-D array of 1-D vectors (along the last</span>
+<a name="line-4785"></a><span class='hs-comment'>-- dimension), and each vector is normalized independently.  Within a given vector,</span>
+<a name="line-4786"></a><span class='hs-comment'>-- each component is divided by the weighted, squared sum of inputs within</span>
+<a name="line-4787"></a><span class='hs-comment'>-- `depth_radius`.  In detail,</span>
+<a name="line-4788"></a><span class='hs-comment'>-- </span>
+<a name="line-4789"></a><span class='hs-comment'>--     sqr_sum[a, b, c, d] =</span>
+<a name="line-4790"></a><span class='hs-comment'>--         sum(input[a, b, c, d - depth_radius : d + depth_radius + 1] ** 2)</span>
+<a name="line-4791"></a><span class='hs-comment'>--     output = input / (bias + alpha * sqr_sum) ** beta</span>
+<a name="line-4792"></a><span class='hs-comment'>-- </span>
+<a name="line-4793"></a><span class='hs-comment'>-- For details, see [Krizhevsky et al., ImageNet classification with deep</span>
+<a name="line-4794"></a><span class='hs-comment'>-- convolutional neural networks (NIPS 2012)]</span>
+<a name="line-4795"></a><span class='hs-comment'>-- (<a href="http://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks).">http://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks).</a></span>
+<a name="line-4796"></a><span class='hs-definition'>lRN</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-4797"></a>       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: 4-D.</span>
+<a name="line-4798"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-4799"></a><span class='hs-definition'>lRN</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4800"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"LRN"</span>
+<a name="line-4801"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-4802"></a>        <span class='hs-varid'>input</span>
+<a name="line-4803"></a><span class='hs-comment'>{-
+<a name="line-4804"></a>attr {
+<a name="line-4805"></a>  default_value { i: 5 }
+<a name="line-4806"></a>  description: "0-D.  Half-width of the 1-D normalization window."
+<a name="line-4807"></a>  name: "depth_radius"
+<a name="line-4808"></a>  type: "int"
+<a name="line-4809"></a>}
+<a name="line-4810"></a>attr {
+<a name="line-4811"></a>  default_value { f: 1.0 }
+<a name="line-4812"></a>  description: "An offset (usually positive to avoid dividing by 0)."
+<a name="line-4813"></a>  name: "bias"
+<a name="line-4814"></a>  type: "float"
+<a name="line-4815"></a>}
+<a name="line-4816"></a>attr {
+<a name="line-4817"></a>  default_value { f: 1.0 }
+<a name="line-4818"></a>  description: "A scale factor, usually positive."
+<a name="line-4819"></a>  name: "alpha"
+<a name="line-4820"></a>  type: "float"
+<a name="line-4821"></a>}
+<a name="line-4822"></a>attr {
+<a name="line-4823"></a>  default_value { f: 0.5 }
+<a name="line-4824"></a>  description: "An exponent."
+<a name="line-4825"></a>  name: "beta"
+<a name="line-4826"></a>  type: "float"
+<a name="line-4827"></a>}
+<a name="line-4828"></a>attr {
+<a name="line-4829"></a>  allowed_values { list { type: DT_FLOAT type: DT_HALF } }
+<a name="line-4830"></a>  default_value { type: DT_FLOAT }
+<a name="line-4831"></a>  name: "T"
+<a name="line-4832"></a>  type: "type"
+<a name="line-4833"></a>}
+<a name="line-4834"></a>input_arg { description: "4-D." name: "input" type_attr: "T" }
+<a name="line-4835"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-4836"></a>-}</span>
+<a name="line-4837"></a>
+<a name="line-4838"></a><a name="stringToHashBucketFast"></a><span class='hs-comment'>-- | Converts each string in the input Tensor to its hash mod by a number of buckets.</span>
+<a name="line-4839"></a><span class='hs-comment'>--</span>
+<a name="line-4840"></a><span class='hs-comment'>-- The hash function is deterministic on the content of the string within the</span>
+<a name="line-4841"></a><span class='hs-comment'>-- process and will never change. However, it is not suitable for cryptography.</span>
+<a name="line-4842"></a><span class='hs-comment'>-- This function may be used when CPU time is scarce and inputs are trusted or</span>
+<a name="line-4843"></a><span class='hs-comment'>-- unimportant. There is a risk of adversaries constructing inputs that all hash</span>
+<a name="line-4844"></a><span class='hs-comment'>-- to the same bucket. To prevent this problem, use a strong hash function with</span>
+<a name="line-4845"></a><span class='hs-comment'>-- `tf.string_to_hash_bucket_strong`.</span>
+<a name="line-4846"></a><span class='hs-definition'>stringToHashBucketFast</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_buckets__: The number of buckets.</span>
+<a name="line-4847"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __input__: The strings to assign a hash bucket.</span>
+<a name="line-4848"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __output__: A Tensor of the same shape as the input `string_tensor`.</span>
+<a name="line-4849"></a><span class='hs-definition'>stringToHashBucketFast</span> <span class='hs-varid'>num_buckets</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4850"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"StringToHashBucketFast"</span>
+<a name="line-4851"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_buckets"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_buckets</span><span class='hs-layout'>)</span>
+<a name="line-4852"></a>        <span class='hs-varid'>input</span>
+<a name="line-4853"></a><span class='hs-comment'>{-
+<a name="line-4854"></a>attr {
+<a name="line-4855"></a>  description: "The number of buckets."
+<a name="line-4856"></a>  has_minimum: true
+<a name="line-4857"></a>  minimum: 1
+<a name="line-4858"></a>  name: "num_buckets"
+<a name="line-4859"></a>  type: "int"
+<a name="line-4860"></a>}
+<a name="line-4861"></a>input_arg {
+<a name="line-4862"></a>  description: "The strings to assign a hash bucket."
+<a name="line-4863"></a>  name: "input"
+<a name="line-4864"></a>  type: DT_STRING
+<a name="line-4865"></a>}
+<a name="line-4866"></a>output_arg {
+<a name="line-4867"></a>  description: "A Tensor of the same shape as the input `string_tensor`."
+<a name="line-4868"></a>  name: "output"
+<a name="line-4869"></a>  type: DT_INT64
+<a name="line-4870"></a>}
+<a name="line-4871"></a>-}</span>
+<a name="line-4872"></a>
+<a name="line-4873"></a><a name="tensorArrayPack"></a><span class='hs-comment'>-- | Pack the elements from the TensorArray into output `value`.</span>
+<a name="line-4874"></a><span class='hs-comment'>--</span>
+<a name="line-4875"></a><span class='hs-comment'>-- **WARNING: This op is deprecated.**</span>
+<a name="line-4876"></a><span class='hs-comment'>-- </span>
+<a name="line-4877"></a><span class='hs-comment'>-- Instead of this op, use `TensorArrayGather` with</span>
+<a name="line-4878"></a><span class='hs-comment'>-- `indices = RangeOp(0, TensorArraySizeOp)`.</span>
+<a name="line-4879"></a><span class='hs-comment'>-- </span>
+<a name="line-4880"></a><span class='hs-comment'>-- All elements must have the same shape.</span>
+<a name="line-4881"></a><span class='hs-definition'>tensorArrayPack</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-4882"></a>                   <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a TensorArray.</span>
+<a name="line-4883"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __flow_in__: A float scalar that enforces proper chaining of operations.</span>
+<a name="line-4884"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __value__: All of the elements in the TensorArray, concatenated along a new</span>
+<a name="line-4885"></a>                   <span class='hs-comment'>-- axis (the new dimension 0).</span>
+<a name="line-4886"></a><span class='hs-definition'>tensorArrayPack</span> <span class='hs-varid'>handle</span> <span class='hs-varid'>flow_in</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4887"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TensorArrayPack"</span>
+<a name="line-4888"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-4889"></a>        <span class='hs-varid'>handle</span> <span class='hs-varid'>flow_in</span>
+<a name="line-4890"></a><span class='hs-comment'>{-
+<a name="line-4891"></a>attr {
+<a name="line-4892"></a>  description: "The type of the elem that is returned."
+<a name="line-4893"></a>  name: "dtype"
+<a name="line-4894"></a>  type: "type"
+<a name="line-4895"></a>}
+<a name="line-4896"></a>attr {
+<a name="line-4897"></a>  default_value { shape { unknown_rank: true } }
+<a name="line-4898"></a>  description: "The expected shape of an element, if known. Used to\nvalidate the shapes of TensorArray elements. If this shape is not\nfully specified, packing zero-size TensorArrays is an error."
+<a name="line-4899"></a>  name: "element_shape"
+<a name="line-4900"></a>  type: "shape"
+<a name="line-4901"></a>}
+<a name="line-4902"></a>input_arg {
+<a name="line-4903"></a>  description: "The handle to a TensorArray."
+<a name="line-4904"></a>  is_ref: true
+<a name="line-4905"></a>  name: "handle"
+<a name="line-4906"></a>  type: DT_STRING
+<a name="line-4907"></a>}
+<a name="line-4908"></a>input_arg {
+<a name="line-4909"></a>  description: "A float scalar that enforces proper chaining of operations."
+<a name="line-4910"></a>  name: "flow_in"
+<a name="line-4911"></a>  type: DT_FLOAT
+<a name="line-4912"></a>}
+<a name="line-4913"></a>output_arg {
+<a name="line-4914"></a>  description: "All of the elements in the TensorArray, concatenated along a new\naxis (the new dimension 0)."
+<a name="line-4915"></a>  name: "value"
+<a name="line-4916"></a>  type_attr: "dtype"
+<a name="line-4917"></a>}
+<a name="line-4918"></a>-}</span>
+<a name="line-4919"></a>
+<a name="line-4920"></a><a name="concatOffset"></a><span class='hs-comment'>-- | Computes offsets of concat inputs within its output.</span>
+<a name="line-4921"></a><span class='hs-comment'>--</span>
+<a name="line-4922"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-4923"></a><span class='hs-comment'>-- </span>
+<a name="line-4924"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-4925"></a><span class='hs-comment'>-- # 'x' is [2, 2, 7]</span>
+<a name="line-4926"></a><span class='hs-comment'>-- # 'y' is [2, 3, 7]</span>
+<a name="line-4927"></a><span class='hs-comment'>-- # 'z' is [2, 5, 7]</span>
+<a name="line-4928"></a><span class='hs-comment'>-- concat_offset(2, [x, y, z]) =&gt; [0, 0, 0], [0, 2, 0], [0, 5, 0]</span>
+<a name="line-4929"></a><span class='hs-comment'>-- ```</span>
+<a name="line-4930"></a><span class='hs-definition'>concatOffset</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __concat_dim__: The dimension along which to concatenate.</span>
+<a name="line-4931"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __shape__: The `N` int32 vectors representing shape of tensors being concatenated.</span>
+<a name="line-4932"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __offset__: The `N` int32 vectors representing the starting offset</span>
+<a name="line-4933"></a>                <span class='hs-comment'>--         of input tensors within the concatenated output.</span>
+<a name="line-4934"></a>                <span class='hs-comment'>-- </span>
+<a name="line-4935"></a>                <span class='hs-comment'>-- This is typically used by gradient computations for a concat operation.</span>
+<a name="line-4936"></a><span class='hs-definition'>concatOffset</span> <span class='hs-varid'>concat_dim</span>
+<a name="line-4937"></a>             <span class='hs-varid'>shape</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"N"</span><span class='hs-layout'>,</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"shape"</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>shape</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4938"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ConcatOffset"</span>
+<a name="line-4939"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"N"</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>shape</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-4940"></a>        <span class='hs-varid'>concat_dim</span> <span class='hs-varid'>shape</span>
+<a name="line-4941"></a><span class='hs-comment'>{-
+<a name="line-4942"></a>attr { has_minimum: true minimum: 2 name: "N" type: "int" }
+<a name="line-4943"></a>input_arg {
+<a name="line-4944"></a>  description: "The dimension along which to concatenate."
+<a name="line-4945"></a>  name: "concat_dim"
+<a name="line-4946"></a>  type: DT_INT32
+<a name="line-4947"></a>}
+<a name="line-4948"></a>input_arg {
+<a name="line-4949"></a>  description: "The `N` int32 vectors representing shape of tensors being concatenated."
+<a name="line-4950"></a>  name: "shape"
+<a name="line-4951"></a>  number_attr: "N"
+<a name="line-4952"></a>  type: DT_INT32
+<a name="line-4953"></a>}
+<a name="line-4954"></a>output_arg {
+<a name="line-4955"></a>  description: "The `N` int32 vectors representing the starting offset\n        of input tensors within the concatenated output.\n\nThis is typically used by gradient computations for a concat operation."
+<a name="line-4956"></a>  name: "offset"
+<a name="line-4957"></a>  number_attr: "N"
+<a name="line-4958"></a>  type: DT_INT32
+<a name="line-4959"></a>}
+<a name="line-4960"></a>-}</span>
+<a name="line-4961"></a>
+<a name="line-4962"></a><a name="refEnter"></a><span class='hs-comment'>-- | Creates or finds a child frame, and makes `data` available to the child frame.</span>
+<a name="line-4963"></a><span class='hs-comment'>--</span>
+<a name="line-4964"></a><span class='hs-comment'>-- The unique `frame_name` is used by the `Executor` to identify frames. If</span>
+<a name="line-4965"></a><span class='hs-comment'>-- `is_constant` is true, `output` is a constant in the child frame; otherwise</span>
+<a name="line-4966"></a><span class='hs-comment'>-- it may be changed in the child frame. At most `parallel_iterations` iterations</span>
+<a name="line-4967"></a><span class='hs-comment'>-- are run in parallel in the child frame.</span>
+<a name="line-4968"></a><span class='hs-definition'>refEnter</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-4969"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__: The tensor to be made available to the child frame.</span>
+<a name="line-4970"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The same tensor as `data`.</span>
+<a name="line-4971"></a><span class='hs-definition'>refEnter</span> <span class='hs-varid'>data'</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-4972"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RefEnter"</span>
+<a name="line-4973"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-4974"></a>        <span class='hs-varid'>data'</span>
+<a name="line-4975"></a><span class='hs-comment'>{-
+<a name="line-4976"></a>attr { name: "T" type: "type" }
+<a name="line-4977"></a>attr {
+<a name="line-4978"></a>  description: "The name of the child frame."
+<a name="line-4979"></a>  name: "frame_name"
+<a name="line-4980"></a>  type: "string"
+<a name="line-4981"></a>}
+<a name="line-4982"></a>attr {
+<a name="line-4983"></a>  default_value { b: false }
+<a name="line-4984"></a>  description: "If true, the output is constant within the child frame."
+<a name="line-4985"></a>  name: "is_constant"
+<a name="line-4986"></a>  type: "bool"
+<a name="line-4987"></a>}
+<a name="line-4988"></a>attr {
+<a name="line-4989"></a>  default_value { i: 10 }
+<a name="line-4990"></a>  description: "The number of iterations allowed to run in parallel."
+<a name="line-4991"></a>  name: "parallel_iterations"
+<a name="line-4992"></a>  type: "int"
+<a name="line-4993"></a>}
+<a name="line-4994"></a>input_arg {
+<a name="line-4995"></a>  description: "The tensor to be made available to the child frame."
+<a name="line-4996"></a>  is_ref: true
+<a name="line-4997"></a>  name: "data"
+<a name="line-4998"></a>  type_attr: "T"
+<a name="line-4999"></a>}
+<a name="line-5000"></a>output_arg {
+<a name="line-5001"></a>  description: "The same tensor as `data`."
+<a name="line-5002"></a>  is_ref: true
+<a name="line-5003"></a>  name: "output"
+<a name="line-5004"></a>  type_attr: "T"
+<a name="line-5005"></a>}
+<a name="line-5006"></a>-}</span>
+<a name="line-5007"></a>
+<a name="line-5008"></a><span class='hs-comment'>-- | Computes softsign: `features / (abs(features) + 1)`.</span>
+<a name="line-5009"></a>
+<a name="line-5010"></a><a name="softsign"></a><span class='hs-definition'>softsign</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-5011"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-5012"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-5013"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-5014"></a>                                                 <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-5015"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __features__</span>
+<a name="line-5016"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __activations__</span>
+<a name="line-5017"></a><span class='hs-definition'>softsign</span> <span class='hs-varid'>features</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5018"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Softsign"</span>
+<a name="line-5019"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-5020"></a>        <span class='hs-varid'>features</span>
+<a name="line-5021"></a><span class='hs-comment'>{-
+<a name="line-5022"></a>attr {
+<a name="line-5023"></a>  allowed_values {
+<a name="line-5024"></a>    list {
+<a name="line-5025"></a>      type: DT_FLOAT
+<a name="line-5026"></a>      type: DT_DOUBLE
+<a name="line-5027"></a>      type: DT_INT32
+<a name="line-5028"></a>      type: DT_INT64
+<a name="line-5029"></a>      type: DT_UINT8
+<a name="line-5030"></a>      type: DT_INT16
+<a name="line-5031"></a>      type: DT_INT8
+<a name="line-5032"></a>      type: DT_UINT16
+<a name="line-5033"></a>      type: DT_HALF
+<a name="line-5034"></a>    }
+<a name="line-5035"></a>  }
+<a name="line-5036"></a>  name: "T"
+<a name="line-5037"></a>  type: "type"
+<a name="line-5038"></a>}
+<a name="line-5039"></a>input_arg { name: "features" type_attr: "T" }
+<a name="line-5040"></a>output_arg { name: "activations" type_attr: "T" }
+<a name="line-5041"></a>-}</span>
+<a name="line-5042"></a>
+<a name="line-5043"></a><span class='hs-comment'>-- | Push an element onto the tensor_array.</span>
+<a name="line-5044"></a>
+<a name="line-5045"></a><a name="tensorArrayWrite"></a><span class='hs-definition'>tensorArrayWrite</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-5046"></a>                    <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a TensorArray.</span>
+<a name="line-5047"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __index__: The position to write to inside the TensorArray.</span>
+<a name="line-5048"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: The tensor to write to the TensorArray.</span>
+<a name="line-5049"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __flow_in__: A float scalar that enforces proper chaining of operations.</span>
+<a name="line-5050"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __flow_out__: A float scalar that enforces proper chaining of operations.</span>
+<a name="line-5051"></a><span class='hs-definition'>tensorArrayWrite</span> <span class='hs-varid'>handle</span> <span class='hs-varid'>index</span> <span class='hs-varid'>value</span> <span class='hs-varid'>flow_in</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5052"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TensorArrayWrite"</span>
+<a name="line-5053"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-5054"></a>        <span class='hs-varid'>handle</span> <span class='hs-varid'>index</span> <span class='hs-varid'>value</span> <span class='hs-varid'>flow_in</span>
+<a name="line-5055"></a><span class='hs-comment'>{-
+<a name="line-5056"></a>attr { name: "T" type: "type" }
+<a name="line-5057"></a>input_arg {
+<a name="line-5058"></a>  description: "The handle to a TensorArray."
+<a name="line-5059"></a>  is_ref: true
+<a name="line-5060"></a>  name: "handle"
+<a name="line-5061"></a>  type: DT_STRING
+<a name="line-5062"></a>}
+<a name="line-5063"></a>input_arg {
+<a name="line-5064"></a>  description: "The position to write to inside the TensorArray."
+<a name="line-5065"></a>  name: "index"
+<a name="line-5066"></a>  type: DT_INT32
+<a name="line-5067"></a>}
+<a name="line-5068"></a>input_arg {
+<a name="line-5069"></a>  description: "The tensor to write to the TensorArray."
+<a name="line-5070"></a>  name: "value"
+<a name="line-5071"></a>  type_attr: "T"
+<a name="line-5072"></a>}
+<a name="line-5073"></a>input_arg {
+<a name="line-5074"></a>  description: "A float scalar that enforces proper chaining of operations."
+<a name="line-5075"></a>  name: "flow_in"
+<a name="line-5076"></a>  type: DT_FLOAT
+<a name="line-5077"></a>}
+<a name="line-5078"></a>output_arg {
+<a name="line-5079"></a>  description: "A float scalar that enforces proper chaining of operations."
+<a name="line-5080"></a>  name: "flow_out"
+<a name="line-5081"></a>  type: DT_FLOAT
+<a name="line-5082"></a>}
+<a name="line-5083"></a>-}</span>
+<a name="line-5084"></a>
+<a name="line-5085"></a><a name="diag"></a><span class='hs-comment'>-- | Returns a diagonal tensor with a given diagonal values.</span>
+<a name="line-5086"></a><span class='hs-comment'>--</span>
+<a name="line-5087"></a><span class='hs-comment'>-- Given a `diagonal`, this operation returns a tensor with the `diagonal` and</span>
+<a name="line-5088"></a><span class='hs-comment'>-- everything else padded with zeros. The diagonal is computed as follows:</span>
+<a name="line-5089"></a><span class='hs-comment'>-- </span>
+<a name="line-5090"></a><span class='hs-comment'>-- Assume `diagonal` has dimensions [D1,..., Dk], then the output is a tensor of</span>
+<a name="line-5091"></a><span class='hs-comment'>-- rank 2k with dimensions [D1,..., Dk, D1,..., Dk] where:</span>
+<a name="line-5092"></a><span class='hs-comment'>-- </span>
+<a name="line-5093"></a><span class='hs-comment'>-- `output[i1,..., ik, i1,..., ik] = diagonal[i1, ..., ik]` and 0 everywhere else.</span>
+<a name="line-5094"></a><span class='hs-comment'>-- </span>
+<a name="line-5095"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-5096"></a><span class='hs-comment'>-- </span>
+<a name="line-5097"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-5098"></a><span class='hs-comment'>-- # 'diagonal' is [1, 2, 3, 4]</span>
+<a name="line-5099"></a><span class='hs-comment'>-- tf.diag(diagonal) ==&gt; [[1, 0, 0, 0]</span>
+<a name="line-5100"></a><span class='hs-comment'>--                        [0, 2, 0, 0]</span>
+<a name="line-5101"></a><span class='hs-comment'>--                        [0, 0, 3, 0]</span>
+<a name="line-5102"></a><span class='hs-comment'>--                        [0, 0, 0, 4]]</span>
+<a name="line-5103"></a><span class='hs-comment'>-- ```</span>
+<a name="line-5104"></a><span class='hs-definition'>diag</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-5105"></a>                                             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-5106"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-5107"></a>                                             <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-5108"></a>        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __diagonal__: Rank k tensor where k is at most 3.</span>
+<a name="line-5109"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-5110"></a><span class='hs-definition'>diag</span> <span class='hs-varid'>diagonal</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5111"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Diag"</span>
+<a name="line-5112"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-5113"></a>        <span class='hs-varid'>diagonal</span>
+<a name="line-5114"></a><span class='hs-comment'>{-
+<a name="line-5115"></a>attr {
+<a name="line-5116"></a>  allowed_values {
+<a name="line-5117"></a>    list {
+<a name="line-5118"></a>      type: DT_FLOAT
+<a name="line-5119"></a>      type: DT_DOUBLE
+<a name="line-5120"></a>      type: DT_INT32
+<a name="line-5121"></a>      type: DT_INT64
+<a name="line-5122"></a>      type: DT_COMPLEX64
+<a name="line-5123"></a>      type: DT_COMPLEX128
+<a name="line-5124"></a>    }
+<a name="line-5125"></a>  }
+<a name="line-5126"></a>  name: "T"
+<a name="line-5127"></a>  type: "type"
+<a name="line-5128"></a>}
+<a name="line-5129"></a>input_arg {
+<a name="line-5130"></a>  description: "Rank k tensor where k is at most 3."
+<a name="line-5131"></a>  name: "diagonal"
+<a name="line-5132"></a>  type_attr: "T"
+<a name="line-5133"></a>}
+<a name="line-5134"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-5135"></a>-}</span>
+<a name="line-5136"></a>
+<a name="line-5137"></a><a name="matrixDiagPart"></a><span class='hs-comment'>-- | Returns the batched diagonal part of a batched tensor.</span>
+<a name="line-5138"></a><span class='hs-comment'>--</span>
+<a name="line-5139"></a><span class='hs-comment'>-- This operation returns a tensor with the `diagonal` part</span>
+<a name="line-5140"></a><span class='hs-comment'>-- of the batched `input`. The `diagonal` part is computed as follows:</span>
+<a name="line-5141"></a><span class='hs-comment'>-- </span>
+<a name="line-5142"></a><span class='hs-comment'>-- Assume `input` has `k` dimensions `[I, J, K, ..., N, N]`, then the output is a</span>
+<a name="line-5143"></a><span class='hs-comment'>-- tensor of rank `k - 1` with dimensions `[I, J, K, ..., N]` where:</span>
+<a name="line-5144"></a><span class='hs-comment'>-- </span>
+<a name="line-5145"></a><span class='hs-comment'>-- `diagonal[i, j, k, ..., n] = input[i, j, k, ..., n, n]`.</span>
+<a name="line-5146"></a><span class='hs-comment'>-- </span>
+<a name="line-5147"></a><span class='hs-comment'>-- The input must be at least a matrix.</span>
+<a name="line-5148"></a><span class='hs-comment'>-- </span>
+<a name="line-5149"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-5150"></a><span class='hs-comment'>-- </span>
+<a name="line-5151"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-5152"></a><span class='hs-comment'>-- # 'input' is [[[1, 0, 0, 0]</span>
+<a name="line-5153"></a><span class='hs-comment'>--                [0, 2, 0, 0]</span>
+<a name="line-5154"></a><span class='hs-comment'>--                [0, 0, 3, 0]</span>
+<a name="line-5155"></a><span class='hs-comment'>--                [0, 0, 0, 4]],</span>
+<a name="line-5156"></a><span class='hs-comment'>--               [[5, 0, 0, 0]</span>
+<a name="line-5157"></a><span class='hs-comment'>--                [0, 6, 0, 0]</span>
+<a name="line-5158"></a><span class='hs-comment'>--                [0, 0, 7, 0]</span>
+<a name="line-5159"></a><span class='hs-comment'>--                [0, 0, 0, 8]]]</span>
+<a name="line-5160"></a><span class='hs-comment'>-- </span>
+<a name="line-5161"></a><span class='hs-comment'>-- and input.shape = (2, 4, 4)</span>
+<a name="line-5162"></a><span class='hs-comment'>-- </span>
+<a name="line-5163"></a><span class='hs-comment'>-- tf.matrix_diag_part(input) ==&gt; [[1, 2, 3, 4], [5, 6, 7, 8]]</span>
+<a name="line-5164"></a><span class='hs-comment'>-- </span>
+<a name="line-5165"></a><span class='hs-comment'>-- which has shape (2, 4)</span>
+<a name="line-5166"></a><span class='hs-comment'>-- ```</span>
+<a name="line-5167"></a><span class='hs-definition'>matrixDiagPart</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-5168"></a>                  <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Rank `k` tensor where `k &gt;= 2` and the last two dimensions are equal.</span>
+<a name="line-5169"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __diagonal__: The extracted diagonal(s) having shape</span>
+<a name="line-5170"></a>                  <span class='hs-comment'>-- `diagonal.shape = input.shape[:-1]`.</span>
+<a name="line-5171"></a><span class='hs-definition'>matrixDiagPart</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5172"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MatrixDiagPart"</span>
+<a name="line-5173"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-5174"></a>        <span class='hs-varid'>input</span>
+<a name="line-5175"></a><span class='hs-comment'>{-
+<a name="line-5176"></a>attr { name: "T" type: "type" }
+<a name="line-5177"></a>input_arg {
+<a name="line-5178"></a>  description: "Rank `k` tensor where `k &gt;= 2` and the last two dimensions are equal."
+<a name="line-5179"></a>  name: "input"
+<a name="line-5180"></a>  type_attr: "T"
+<a name="line-5181"></a>}
+<a name="line-5182"></a>output_arg {
+<a name="line-5183"></a>  description: "The extracted diagonal(s) having shape\n`diagonal.shape = input.shape[:-1]`."
+<a name="line-5184"></a>  name: "diagonal"
+<a name="line-5185"></a>  type_attr: "T"
+<a name="line-5186"></a>}
+<a name="line-5187"></a>-}</span>
+<a name="line-5188"></a>
+<a name="line-5189"></a><span class='hs-comment'>-- | Computes the number of elements in the given queue.</span>
+<a name="line-5190"></a>
+<a name="line-5191"></a><a name="queueSize"></a><span class='hs-definition'>queueSize</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a queue.</span>
+<a name="line-5192"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __size__: The number of elements in the given queue.</span>
+<a name="line-5193"></a><span class='hs-definition'>queueSize</span> <span class='hs-varid'>handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5194"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"QueueSize"</span><span class='hs-layout'>)</span>
+<a name="line-5195"></a>        <span class='hs-varid'>handle</span>
+<a name="line-5196"></a><span class='hs-comment'>{-
+<a name="line-5197"></a>input_arg {
+<a name="line-5198"></a>  description: "The handle to a queue."
+<a name="line-5199"></a>  is_ref: true
+<a name="line-5200"></a>  name: "handle"
+<a name="line-5201"></a>  type: DT_STRING
+<a name="line-5202"></a>}
+<a name="line-5203"></a>output_arg {
+<a name="line-5204"></a>  description: "The number of elements in the given queue."
+<a name="line-5205"></a>  name: "size"
+<a name="line-5206"></a>  type: DT_INT32
+<a name="line-5207"></a>}
+<a name="line-5208"></a>-}</span>
+<a name="line-5209"></a>
+<a name="line-5210"></a><a name="decodePng"></a><span class='hs-comment'>-- | Decode a PNG-encoded image to a uint8 or uint16 tensor.</span>
+<a name="line-5211"></a><span class='hs-comment'>--</span>
+<a name="line-5212"></a><span class='hs-comment'>-- The attr `channels` indicates the desired number of color channels for the</span>
+<a name="line-5213"></a><span class='hs-comment'>-- decoded image.</span>
+<a name="line-5214"></a><span class='hs-comment'>-- </span>
+<a name="line-5215"></a><span class='hs-comment'>-- Accepted values are:</span>
+<a name="line-5216"></a><span class='hs-comment'>-- </span>
+<a name="line-5217"></a><span class='hs-comment'>-- *   0: Use the number of channels in the PNG-encoded image.</span>
+<a name="line-5218"></a><span class='hs-comment'>-- *   1: output a grayscale image.</span>
+<a name="line-5219"></a><span class='hs-comment'>-- *   3: output an RGB image.</span>
+<a name="line-5220"></a><span class='hs-comment'>-- *   4: output an RGBA image.</span>
+<a name="line-5221"></a><span class='hs-comment'>-- </span>
+<a name="line-5222"></a><span class='hs-comment'>-- If needed, the PNG-encoded image is transformed to match the requested number</span>
+<a name="line-5223"></a><span class='hs-comment'>-- of color channels.</span>
+<a name="line-5224"></a><span class='hs-definition'>decodePng</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-5225"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-5226"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __contents__: 0-D.  The PNG-encoded image.</span>
+<a name="line-5227"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __image__: 3-D with shape `[height, width, channels]`.</span>
+<a name="line-5228"></a><span class='hs-definition'>decodePng</span> <span class='hs-varid'>contents</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5229"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DecodePng"</span>
+<a name="line-5230"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-5231"></a>        <span class='hs-varid'>contents</span>
+<a name="line-5232"></a><span class='hs-comment'>{-
+<a name="line-5233"></a>attr {
+<a name="line-5234"></a>  default_value { i: 0 }
+<a name="line-5235"></a>  description: "Number of color channels for the decoded image."
+<a name="line-5236"></a>  name: "channels"
+<a name="line-5237"></a>  type: "int"
+<a name="line-5238"></a>}
+<a name="line-5239"></a>attr {
+<a name="line-5240"></a>  allowed_values { list { type: DT_UINT8 type: DT_UINT16 } }
+<a name="line-5241"></a>  default_value { type: DT_UINT8 }
+<a name="line-5242"></a>  name: "dtype"
+<a name="line-5243"></a>  type: "type"
+<a name="line-5244"></a>}
+<a name="line-5245"></a>input_arg {
+<a name="line-5246"></a>  description: "0-D.  The PNG-encoded image."
+<a name="line-5247"></a>  name: "contents"
+<a name="line-5248"></a>  type: DT_STRING
+<a name="line-5249"></a>}
+<a name="line-5250"></a>output_arg {
+<a name="line-5251"></a>  description: "3-D with shape `[height, width, channels]`."
+<a name="line-5252"></a>  name: "image"
+<a name="line-5253"></a>  type_attr: "dtype"
+<a name="line-5254"></a>}
+<a name="line-5255"></a>-}</span>
+<a name="line-5256"></a>
+<a name="line-5257"></a><span class='hs-comment'>-- | Returns element-wise smallest integer in not less than x.</span>
+<a name="line-5258"></a>
+<a name="line-5259"></a><a name="ceil"></a><span class='hs-definition'>ceil</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-5260"></a>                                             <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-5261"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-5262"></a><span class='hs-definition'>ceil</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5263"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Ceil"</span>
+<a name="line-5264"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-5265"></a>        <span class='hs-varid'>x</span>
+<a name="line-5266"></a><span class='hs-comment'>{-
+<a name="line-5267"></a>attr {
+<a name="line-5268"></a>  allowed_values {
+<a name="line-5269"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-5270"></a>  }
+<a name="line-5271"></a>  name: "T"
+<a name="line-5272"></a>  type: "type"
+<a name="line-5273"></a>}
+<a name="line-5274"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-5275"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-5276"></a>-}</span>
+<a name="line-5277"></a>
+<a name="line-5278"></a><a name="priorityQueue"></a><span class='hs-comment'>-- | A queue that produces elements sorted by the first component value.</span>
+<a name="line-5279"></a><span class='hs-comment'>--</span>
+<a name="line-5280"></a><span class='hs-comment'>-- Note that the PriorityQueue requires the first component of any element</span>
+<a name="line-5281"></a><span class='hs-comment'>-- to be a scalar int64, in addition to the other elements declared by</span>
+<a name="line-5282"></a><span class='hs-comment'>-- component_types.  Therefore calls to Enqueue and EnqueueMany (resp. Dequeue</span>
+<a name="line-5283"></a><span class='hs-comment'>-- and DequeueMany) on a PriorityQueue will all require (resp. output) one extra</span>
+<a name="line-5284"></a><span class='hs-comment'>-- entry in their input (resp. output) lists.</span>
+<a name="line-5285"></a><span class='hs-definition'>priorityQueue</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to the queue.</span>
+<a name="line-5286"></a><span class='hs-definition'>priorityQueue</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5287"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"PriorityQueue"</span><span class='hs-layout'>)</span>
+<a name="line-5288"></a>        
+<a name="line-5289"></a><span class='hs-comment'>{-
+<a name="line-5290"></a>attr {
+<a name="line-5291"></a>  default_value { list { } }
+<a name="line-5292"></a>  description: "The type of each component in a value."
+<a name="line-5293"></a>  has_minimum: true
+<a name="line-5294"></a>  name: "component_types"
+<a name="line-5295"></a>  type: "list(type)"
+<a name="line-5296"></a>}
+<a name="line-5297"></a>attr {
+<a name="line-5298"></a>  description: "The shape of each component in a value. The length of this attr must\nbe either 0 or the same as the length of component_types. If the length of\nthis attr is 0, the shapes of queue elements are not constrained, and\nonly one element may be dequeued at a time."
+<a name="line-5299"></a>  has_minimum: true
+<a name="line-5300"></a>  name: "shapes"
+<a name="line-5301"></a>  type: "list(shape)"
+<a name="line-5302"></a>}
+<a name="line-5303"></a>attr {
+<a name="line-5304"></a>  default_value { i: -1 }
+<a name="line-5305"></a>  description: "The upper bound on the number of elements in this queue.\nNegative numbers mean no limit."
+<a name="line-5306"></a>  name: "capacity"
+<a name="line-5307"></a>  type: "int"
+<a name="line-5308"></a>}
+<a name="line-5309"></a>attr {
+<a name="line-5310"></a>  default_value { s: "" }
+<a name="line-5311"></a>  description: "If non-empty, this queue is placed in the given container.\nOtherwise, a default container is used."
+<a name="line-5312"></a>  name: "container"
+<a name="line-5313"></a>  type: "string"
+<a name="line-5314"></a>}
+<a name="line-5315"></a>attr {
+<a name="line-5316"></a>  default_value { s: "" }
+<a name="line-5317"></a>  description: "If non-empty, this queue will be shared under the given name\nacross multiple sessions."
+<a name="line-5318"></a>  name: "shared_name"
+<a name="line-5319"></a>  type: "string"
+<a name="line-5320"></a>}
+<a name="line-5321"></a>output_arg {
+<a name="line-5322"></a>  description: "The handle to the queue."
+<a name="line-5323"></a>  is_ref: true
+<a name="line-5324"></a>  name: "handle"
+<a name="line-5325"></a>  type: DT_STRING
+<a name="line-5326"></a>}
+<a name="line-5327"></a>-}</span>
+<a name="line-5328"></a>
+<a name="line-5329"></a><span class='hs-comment'>-- | A placeholder op that passes though `input` when its output is not fed.</span>
+<a name="line-5330"></a>
+<a name="line-5331"></a><a name="placeholderWithDefault"></a><span class='hs-definition'>placeholderWithDefault</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-5332"></a>                          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __input__: The default value to produce when `output` is not fed.</span>
+<a name="line-5333"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __output__: A placeholder tensor that defaults to `input` if it is not fed.</span>
+<a name="line-5334"></a><span class='hs-definition'>placeholderWithDefault</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5335"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"PlaceholderWithDefault"</span>
+<a name="line-5336"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-5337"></a>        <span class='hs-varid'>input</span>
+<a name="line-5338"></a><span class='hs-comment'>{-
+<a name="line-5339"></a>attr {
+<a name="line-5340"></a>  description: "The type of elements in the tensor."
+<a name="line-5341"></a>  name: "dtype"
+<a name="line-5342"></a>  type: "type"
+<a name="line-5343"></a>}
+<a name="line-5344"></a>attr {
+<a name="line-5345"></a>  description: "The (possibly partial) shape of the tensor."
+<a name="line-5346"></a>  name: "shape"
+<a name="line-5347"></a>  type: "shape"
+<a name="line-5348"></a>}
+<a name="line-5349"></a>input_arg {
+<a name="line-5350"></a>  description: "The default value to produce when `output` is not fed."
+<a name="line-5351"></a>  name: "input"
+<a name="line-5352"></a>  type_attr: "dtype"
+<a name="line-5353"></a>}
+<a name="line-5354"></a>output_arg {
+<a name="line-5355"></a>  description: "A placeholder tensor that defaults to `input` if it is not fed."
+<a name="line-5356"></a>  name: "output"
+<a name="line-5357"></a>  type_attr: "dtype"
+<a name="line-5358"></a>}
+<a name="line-5359"></a>-}</span>
+<a name="line-5360"></a>
+<a name="line-5361"></a><span class='hs-comment'>-- | Computes the gradient of the crop_and_resize op wrt the input image tensor.</span>
+<a name="line-5362"></a>
+<a name="line-5363"></a><a name="cropAndResizeGradImage"></a><span class='hs-definition'>cropAndResizeGradImage</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-5364"></a>                                                  <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-5365"></a>                                                          <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-5366"></a>                          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __grads__: A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`.</span>
+<a name="line-5367"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __boxes__: A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor</span>
+<a name="line-5368"></a>                                             <span class='hs-comment'>-- specifies the coordinates of a box in the `box_ind[i]` image and is specified</span>
+<a name="line-5369"></a>                                             <span class='hs-comment'>-- in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of</span>
+<a name="line-5370"></a>                                             <span class='hs-comment'>-- `y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the</span>
+<a name="line-5371"></a>                                             <span class='hs-comment'>-- `[0, 1]` interval of normalized image height is mapped to</span>
+<a name="line-5372"></a>                                             <span class='hs-comment'>-- `[0, image_height - 1] in image height coordinates. We do allow y1 &gt; y2, in</span>
+<a name="line-5373"></a>                                             <span class='hs-comment'>-- which case the sampled crop is an up-down flipped version of the original</span>
+<a name="line-5374"></a>                                             <span class='hs-comment'>-- image. The width dimension is treated similarly. Normalized coordinates</span>
+<a name="line-5375"></a>                                             <span class='hs-comment'>-- outside the `[0, 1]` range are allowed, in which case we use</span>
+<a name="line-5376"></a>                                             <span class='hs-comment'>-- `extrapolation_value` to extrapolate the input image values.</span>
+<a name="line-5377"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __box_ind__: A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.</span>
+<a name="line-5378"></a>                                                      <span class='hs-comment'>-- The value of `box_ind[i]` specifies the image that the `i`-th box refers to.</span>
+<a name="line-5379"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __image_size__: A 1-D tensor with value `[batch, image_height, image_width, depth]`</span>
+<a name="line-5380"></a>                                                      <span class='hs-comment'>-- containing the original image size. Both `image_height` and `image_width` need</span>
+<a name="line-5381"></a>                                                      <span class='hs-comment'>-- to be positive.</span>
+<a name="line-5382"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: A 4-D tensor of shape `[batch, image_height, image_width, depth]`.</span>
+<a name="line-5383"></a><span class='hs-definition'>cropAndResizeGradImage</span> <span class='hs-varid'>grads</span> <span class='hs-varid'>boxes</span> <span class='hs-varid'>box_ind</span> <span class='hs-varid'>image_size</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5384"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"CropAndResizeGradImage"</span>
+<a name="line-5385"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-5386"></a>        <span class='hs-varid'>grads</span> <span class='hs-varid'>boxes</span> <span class='hs-varid'>box_ind</span> <span class='hs-varid'>image_size</span>
+<a name="line-5387"></a><span class='hs-comment'>{-
+<a name="line-5388"></a>attr {
+<a name="line-5389"></a>  allowed_values {
+<a name="line-5390"></a>    list { type: DT_FLOAT type: DT_HALF type: DT_DOUBLE }
+<a name="line-5391"></a>  }
+<a name="line-5392"></a>  name: "T"
+<a name="line-5393"></a>  type: "type"
+<a name="line-5394"></a>}
+<a name="line-5395"></a>attr {
+<a name="line-5396"></a>  allowed_values { list { s: "bilinear" } }
+<a name="line-5397"></a>  default_value { s: "bilinear" }
+<a name="line-5398"></a>  description: "A string specifying the interpolation method. Only \'bilinear\' is\nsupported for now."
+<a name="line-5399"></a>  name: "method"
+<a name="line-5400"></a>  type: "string"
+<a name="line-5401"></a>}
+<a name="line-5402"></a>input_arg {
+<a name="line-5403"></a>  description: "A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`."
+<a name="line-5404"></a>  name: "grads"
+<a name="line-5405"></a>  type: DT_FLOAT
+<a name="line-5406"></a>}
+<a name="line-5407"></a>input_arg {
+<a name="line-5408"></a>  description: "A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor\nspecifies the coordinates of a box in the `box_ind[i]` image and is specified\nin normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of\n`y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the\n`[0, 1]` interval of normalized image height is mapped to\n`[0, image_height - 1] in image height coordinates. We do allow y1 &gt; y2, in\nwhich case the sampled crop is an up-down flipped version of the original\nimage. The width dimension is treated similarly. Normalized coordinates\noutside the `[0, 1]` range are allowed, in which case we use\n`extrapolation_value` to extrapolate the input image values."
+<a name="line-5409"></a>  name: "boxes"
+<a name="line-5410"></a>  type: DT_FLOAT
+<a name="line-5411"></a>}
+<a name="line-5412"></a>input_arg {
+<a name="line-5413"></a>  description: "A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.\nThe value of `box_ind[i]` specifies the image that the `i`-th box refers to."
+<a name="line-5414"></a>  name: "box_ind"
+<a name="line-5415"></a>  type: DT_INT32
+<a name="line-5416"></a>}
+<a name="line-5417"></a>input_arg {
+<a name="line-5418"></a>  description: "A 1-D tensor with value `[batch, image_height, image_width, depth]`\ncontaining the original image size. Both `image_height` and `image_width` need\nto be positive."
+<a name="line-5419"></a>  name: "image_size"
+<a name="line-5420"></a>  type: DT_INT32
+<a name="line-5421"></a>}
+<a name="line-5422"></a>output_arg {
+<a name="line-5423"></a>  description: "A 4-D tensor of shape `[batch, image_height, image_width, depth]`."
+<a name="line-5424"></a>  name: "output"
+<a name="line-5425"></a>  type_attr: "T"
+<a name="line-5426"></a>}
+<a name="line-5427"></a>-}</span>
+<a name="line-5428"></a>
+<a name="line-5429"></a><span class='hs-comment'>-- | Restore a Reader to its initial clean state.</span>
+<a name="line-5430"></a>
+<a name="line-5431"></a><a name="readerReset"></a><span class='hs-definition'>readerReset</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __reader_handle__: Handle to a Reader.</span>
+<a name="line-5432"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-5433"></a><span class='hs-definition'>readerReset</span> <span class='hs-varid'>reader_handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5434"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ReaderReset"</span><span class='hs-layout'>)</span>
+<a name="line-5435"></a>        <span class='hs-varid'>reader_handle</span>
+<a name="line-5436"></a><span class='hs-comment'>{-
+<a name="line-5437"></a>input_arg {
+<a name="line-5438"></a>  description: "Handle to a Reader."
+<a name="line-5439"></a>  is_ref: true
+<a name="line-5440"></a>  name: "reader_handle"
+<a name="line-5441"></a>  type: DT_STRING
+<a name="line-5442"></a>}
+<a name="line-5443"></a>-}</span>
+<a name="line-5444"></a>
+<a name="line-5445"></a><span class='hs-comment'>-- | Extract `patches` from `images` and put them in the "depth" output dimension.</span>
+<a name="line-5446"></a>
+<a name="line-5447"></a><a name="extractImagePatches"></a><span class='hs-definition'>extractImagePatches</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-5448"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-5449"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-5450"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-5451"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-5452"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-5453"></a>                                                            <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-5454"></a>                       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __images__: 4-D Tensor with shape `[batch, in_rows, in_cols, depth]`.</span>
+<a name="line-5455"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __patches__: 4-D Tensor with shape `[batch, out_rows, out_cols, ksize_rows *</span>
+<a name="line-5456"></a>                       <span class='hs-comment'>-- ksize_cols * depth]` containing image patches with size</span>
+<a name="line-5457"></a>                       <span class='hs-comment'>-- `ksize_rows x ksize_cols x depth` vectorized in the "depth" dimension.</span>
+<a name="line-5458"></a><span class='hs-definition'>extractImagePatches</span> <span class='hs-varid'>images</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5459"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ExtractImagePatches"</span>
+<a name="line-5460"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-5461"></a>        <span class='hs-varid'>images</span>
+<a name="line-5462"></a><span class='hs-comment'>{-
+<a name="line-5463"></a>attr {
+<a name="line-5464"></a>  description: "The size of the sliding window for each dimension of `images`."
+<a name="line-5465"></a>  has_minimum: true
+<a name="line-5466"></a>  minimum: 4
+<a name="line-5467"></a>  name: "ksizes"
+<a name="line-5468"></a>  type: "list(int)"
+<a name="line-5469"></a>}
+<a name="line-5470"></a>attr {
+<a name="line-5471"></a>  description: "1-D of length 4. How far the centers of two consecutive patches are in\nthe images. Must be: `[1, stride_rows, stride_cols, 1]`."
+<a name="line-5472"></a>  has_minimum: true
+<a name="line-5473"></a>  minimum: 4
+<a name="line-5474"></a>  name: "strides"
+<a name="line-5475"></a>  type: "list(int)"
+<a name="line-5476"></a>}
+<a name="line-5477"></a>attr {
+<a name="line-5478"></a>  description: "1-D of length 4. Must be: `[1, rate_rows, rate_cols, 1]`. This is the\ninput stride, specifying how far two consecutive patch samples are in the\ninput. Equivalent to extracting patches with\n`patch_sizes_eff = patch_sizes + (patch_sizes - 1) * (rates - 1), followed by\nsubsampling them spatially by a factor of `rates`."
+<a name="line-5479"></a>  has_minimum: true
+<a name="line-5480"></a>  minimum: 4
+<a name="line-5481"></a>  name: "rates"
+<a name="line-5482"></a>  type: "list(int)"
+<a name="line-5483"></a>}
+<a name="line-5484"></a>attr {
+<a name="line-5485"></a>  allowed_values {
+<a name="line-5486"></a>    list {
+<a name="line-5487"></a>      type: DT_FLOAT
+<a name="line-5488"></a>      type: DT_DOUBLE
+<a name="line-5489"></a>      type: DT_INT32
+<a name="line-5490"></a>      type: DT_INT64
+<a name="line-5491"></a>      type: DT_UINT8
+<a name="line-5492"></a>      type: DT_INT16
+<a name="line-5493"></a>      type: DT_INT8
+<a name="line-5494"></a>      type: DT_UINT16
+<a name="line-5495"></a>      type: DT_HALF
+<a name="line-5496"></a>    }
+<a name="line-5497"></a>  }
+<a name="line-5498"></a>  name: "T"
+<a name="line-5499"></a>  type: "type"
+<a name="line-5500"></a>}
+<a name="line-5501"></a>attr {
+<a name="line-5502"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-5503"></a>  description: "The type of padding algorithm to use.\n\nWe specify the size-related attributes as:\n\n      ksizes = [1, ksize_rows, ksize_cols, 1]\n      strides = [1, strides_rows, strides_cols, 1]\n      rates = [1, rates_rows, rates_cols, 1]"
+<a name="line-5504"></a>  name: "padding"
+<a name="line-5505"></a>  type: "string"
+<a name="line-5506"></a>}
+<a name="line-5507"></a>input_arg {
+<a name="line-5508"></a>  description: "4-D Tensor with shape `[batch, in_rows, in_cols, depth]`."
+<a name="line-5509"></a>  name: "images"
+<a name="line-5510"></a>  type_attr: "T"
+<a name="line-5511"></a>}
+<a name="line-5512"></a>output_arg {
+<a name="line-5513"></a>  description: "4-D Tensor with shape `[batch, out_rows, out_cols, ksize_rows *\nksize_cols * depth]` containing image patches with size\n`ksize_rows x ksize_cols x depth` vectorized in the \"depth\" dimension."
+<a name="line-5514"></a>  name: "patches"
+<a name="line-5515"></a>  type_attr: "T"
+<a name="line-5516"></a>}
+<a name="line-5517"></a>-}</span>
+<a name="line-5518"></a>
+<a name="line-5519"></a><span class='hs-comment'>-- | </span>
+<a name="line-5520"></a>
+<a name="line-5521"></a><a name="batchMatrixSetDiag"></a><span class='hs-definition'>batchMatrixSetDiag</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-5522"></a>                      <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-5523"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __diagonal__</span>
+<a name="line-5524"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-5525"></a><span class='hs-definition'>batchMatrixSetDiag</span> <span class='hs-varid'>input</span> <span class='hs-varid'>diagonal</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5526"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchMatrixSetDiag"</span>
+<a name="line-5527"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-5528"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>diagonal</span>
+<a name="line-5529"></a><span class='hs-comment'>{-
+<a name="line-5530"></a>attr { name: "T" type: "type" }
+<a name="line-5531"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-5532"></a>input_arg { name: "diagonal" type_attr: "T" }
+<a name="line-5533"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-5534"></a>-}</span>
+<a name="line-5535"></a>
+<a name="line-5536"></a><span class='hs-comment'>-- | Delete the stack from its resource container.</span>
+<a name="line-5537"></a>
+<a name="line-5538"></a><a name="stackClose"></a><span class='hs-definition'>stackClose</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a stack.</span>
+<a name="line-5539"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-5540"></a><span class='hs-definition'>stackClose</span> <span class='hs-varid'>handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5541"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"StackClose"</span><span class='hs-layout'>)</span>
+<a name="line-5542"></a>        <span class='hs-varid'>handle</span>
+<a name="line-5543"></a><span class='hs-comment'>{-
+<a name="line-5544"></a>input_arg {
+<a name="line-5545"></a>  description: "The handle to a stack."
+<a name="line-5546"></a>  is_ref: true
+<a name="line-5547"></a>  name: "handle"
+<a name="line-5548"></a>  type: DT_STRING
+<a name="line-5549"></a>}
+<a name="line-5550"></a>-}</span>
+<a name="line-5551"></a>
+<a name="line-5552"></a><a name="quantizeAndDequantize"></a><span class='hs-comment'>-- | Quantizes then dequantizes a tensor.</span>
+<a name="line-5553"></a><span class='hs-comment'>--</span>
+<a name="line-5554"></a><span class='hs-comment'>-- This op simulates the precision loss from the quantized forward pass by:</span>
+<a name="line-5555"></a><span class='hs-comment'>-- 1. Quantizing the tensor to fixed point numbers, which should match the target</span>
+<a name="line-5556"></a><span class='hs-comment'>--    quantization method when it is used in inference.</span>
+<a name="line-5557"></a><span class='hs-comment'>-- 2. Dequantizing it back to floating point numbers for the following ops, most</span>
+<a name="line-5558"></a><span class='hs-comment'>--    likely matmul.</span>
+<a name="line-5559"></a><span class='hs-comment'>-- </span>
+<a name="line-5560"></a><span class='hs-comment'>-- There are different ways to quantize. This version does not use the full range</span>
+<a name="line-5561"></a><span class='hs-comment'>-- of the output type, choosing to elide the lowest possible value for symmetry</span>
+<a name="line-5562"></a><span class='hs-comment'>-- (e.g., output range is -127 to 127, not -128 to 127 for signed 8 bit</span>
+<a name="line-5563"></a><span class='hs-comment'>-- quantization), so that 0.0 maps to 0.</span>
+<a name="line-5564"></a><span class='hs-comment'>-- </span>
+<a name="line-5565"></a><span class='hs-comment'>-- To perform this op, we first find the range of values in our tensor. The range</span>
+<a name="line-5566"></a><span class='hs-comment'>-- we use is always centered on 0, so we find m such that</span>
+<a name="line-5567"></a><span class='hs-comment'>-- </span>
+<a name="line-5568"></a><span class='hs-comment'>-- 1. m = max(abs(input_min), abs(input_max)) if range_given is true,</span>
+<a name="line-5569"></a><span class='hs-comment'>-- 2. m = max(max(abs(min_elem(input)), abs(max_elem(input))) otherwise.</span>
+<a name="line-5570"></a><span class='hs-comment'>-- </span>
+<a name="line-5571"></a><span class='hs-comment'>-- Our input tensor range is then [-m, m].</span>
+<a name="line-5572"></a><span class='hs-comment'>-- </span>
+<a name="line-5573"></a><span class='hs-comment'>-- Next, we choose our fixed-point quantization buckets, [min_fixed, max_fixed].</span>
+<a name="line-5574"></a><span class='hs-comment'>-- If signed_input is true, this is</span>
+<a name="line-5575"></a><span class='hs-comment'>-- </span>
+<a name="line-5576"></a><span class='hs-comment'>--   [min_fixed, max_fixed ] =</span>
+<a name="line-5577"></a><span class='hs-comment'>--       [-(1 &lt;&lt; (num_bits - 1) - 1), (1 &lt;&lt; (num_bits - 1)) - 1].</span>
+<a name="line-5578"></a><span class='hs-comment'>-- </span>
+<a name="line-5579"></a><span class='hs-comment'>-- Otherwise, if signed_input is false, the fixed-point range is</span>
+<a name="line-5580"></a><span class='hs-comment'>-- </span>
+<a name="line-5581"></a><span class='hs-comment'>--   [min_fixed, max_fixed] = [0, (1 &lt;&lt; num_bits) - 1].</span>
+<a name="line-5582"></a><span class='hs-comment'>-- </span>
+<a name="line-5583"></a><span class='hs-comment'>-- From this we compute our scaling factor, s:</span>
+<a name="line-5584"></a><span class='hs-comment'>-- </span>
+<a name="line-5585"></a><span class='hs-comment'>--   s = (max_fixed - min_fixed) / (2 * m).</span>
+<a name="line-5586"></a><span class='hs-comment'>-- </span>
+<a name="line-5587"></a><span class='hs-comment'>-- Now we can quantize and dequantize the elements of our tensor.  An element e</span>
+<a name="line-5588"></a><span class='hs-comment'>-- is transformed into e':</span>
+<a name="line-5589"></a><span class='hs-comment'>-- </span>
+<a name="line-5590"></a><span class='hs-comment'>--   e' = (e * s).round_to_nearest() / s.</span>
+<a name="line-5591"></a><span class='hs-comment'>-- </span>
+<a name="line-5592"></a><span class='hs-comment'>-- Note that we have a different number of buckets in the signed vs. unsigned</span>
+<a name="line-5593"></a><span class='hs-comment'>-- cases.  For example, if num_bits == 8, we get 254 buckets in the signed case</span>
+<a name="line-5594"></a><span class='hs-comment'>-- vs. 255 in the unsigned case.</span>
+<a name="line-5595"></a><span class='hs-comment'>-- </span>
+<a name="line-5596"></a><span class='hs-comment'>-- For example, suppose num_bits = 8 and m = 1.  Then</span>
+<a name="line-5597"></a><span class='hs-comment'>-- </span>
+<a name="line-5598"></a><span class='hs-comment'>--   [min_fixed, max_fixed] = [-127, 127], and</span>
+<a name="line-5599"></a><span class='hs-comment'>--   s = (127 + 127) / 2 = 127.</span>
+<a name="line-5600"></a><span class='hs-comment'>-- </span>
+<a name="line-5601"></a><span class='hs-comment'>-- Given the vector {-1, -0.5, 0, 0.3}, this is quantized to</span>
+<a name="line-5602"></a><span class='hs-comment'>-- {-127, -63, 0, 38}, and dequantized to {-1, -63.0/127, 0, 38.0/127}.</span>
+<a name="line-5603"></a><span class='hs-definition'>quantizeAndDequantize</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-5604"></a>                                                              <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-5605"></a>                         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Tensor to quantize and then dequantize.</span>
+<a name="line-5606"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-5607"></a><span class='hs-definition'>quantizeAndDequantize</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5608"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"QuantizeAndDequantize"</span>
+<a name="line-5609"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-5610"></a>        <span class='hs-varid'>input</span>
+<a name="line-5611"></a><span class='hs-comment'>{-
+<a name="line-5612"></a>attr {
+<a name="line-5613"></a>  default_value { b: true }
+<a name="line-5614"></a>  description: "If the quantization is signed or unsigned."
+<a name="line-5615"></a>  name: "signed_input"
+<a name="line-5616"></a>  type: "bool"
+<a name="line-5617"></a>}
+<a name="line-5618"></a>attr {
+<a name="line-5619"></a>  default_value { i: 8 }
+<a name="line-5620"></a>  description: "The bitwidth of the quantization."
+<a name="line-5621"></a>  name: "num_bits"
+<a name="line-5622"></a>  type: "int"
+<a name="line-5623"></a>}
+<a name="line-5624"></a>attr {
+<a name="line-5625"></a>  default_value { b: false }
+<a name="line-5626"></a>  description: "If the range is given or should be computed from the tensor."
+<a name="line-5627"></a>  name: "range_given"
+<a name="line-5628"></a>  type: "bool"
+<a name="line-5629"></a>}
+<a name="line-5630"></a>attr {
+<a name="line-5631"></a>  default_value { f: 0.0 }
+<a name="line-5632"></a>  description: "If range is given, this is the min of the range."
+<a name="line-5633"></a>  name: "input_min"
+<a name="line-5634"></a>  type: "float"
+<a name="line-5635"></a>}
+<a name="line-5636"></a>attr {
+<a name="line-5637"></a>  default_value { f: 0.0 }
+<a name="line-5638"></a>  description: "If range is given, this is the max of the range."
+<a name="line-5639"></a>  name: "input_max"
+<a name="line-5640"></a>  type: "float"
+<a name="line-5641"></a>}
+<a name="line-5642"></a>attr {
+<a name="line-5643"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-5644"></a>  name: "T"
+<a name="line-5645"></a>  type: "type"
+<a name="line-5646"></a>}
+<a name="line-5647"></a>input_arg {
+<a name="line-5648"></a>  description: "Tensor to quantize and then dequantize."
+<a name="line-5649"></a>  name: "input"
+<a name="line-5650"></a>  type_attr: "T"
+<a name="line-5651"></a>}
+<a name="line-5652"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-5653"></a>-}</span>
+<a name="line-5654"></a>
+<a name="line-5655"></a><span class='hs-comment'>-- | Returns which elements of x are NaN.</span>
+<a name="line-5656"></a>
+<a name="line-5657"></a><a name="isNan"></a><span class='hs-definition'>isNan</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-5658"></a>                                              <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-5659"></a>         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-5660"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-5661"></a><span class='hs-definition'>isNan</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5662"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"IsNan"</span>
+<a name="line-5663"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-5664"></a>        <span class='hs-varid'>x</span>
+<a name="line-5665"></a><span class='hs-comment'>{-
+<a name="line-5666"></a>attr {
+<a name="line-5667"></a>  allowed_values {
+<a name="line-5668"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-5669"></a>  }
+<a name="line-5670"></a>  name: "T"
+<a name="line-5671"></a>  type: "type"
+<a name="line-5672"></a>}
+<a name="line-5673"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-5674"></a>output_arg { name: "y" type: DT_BOOL }
+<a name="line-5675"></a>-}</span>
+<a name="line-5676"></a>
+<a name="line-5677"></a><a name="where'"></a><span class='hs-comment'>-- | Returns locations of true values in a boolean tensor.</span>
+<a name="line-5678"></a><span class='hs-comment'>--</span>
+<a name="line-5679"></a><span class='hs-comment'>-- This operation returns the coordinates of true elements in `input`. The</span>
+<a name="line-5680"></a><span class='hs-comment'>-- coordinates are returned in a 2-D tensor where the first dimension (rows)</span>
+<a name="line-5681"></a><span class='hs-comment'>-- represents the number of true elements, and the second dimension (columns)</span>
+<a name="line-5682"></a><span class='hs-comment'>-- represents the coordinates of the true elements. Keep in mind, the shape of</span>
+<a name="line-5683"></a><span class='hs-comment'>-- the output tensor can vary depending on how many true values there are in</span>
+<a name="line-5684"></a><span class='hs-comment'>-- `input`. Indices are output in row-major order.</span>
+<a name="line-5685"></a><span class='hs-comment'>-- </span>
+<a name="line-5686"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-5687"></a><span class='hs-comment'>-- </span>
+<a name="line-5688"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-5689"></a><span class='hs-comment'>-- # 'input' tensor is [[True, False]</span>
+<a name="line-5690"></a><span class='hs-comment'>-- #                    [True, False]]</span>
+<a name="line-5691"></a><span class='hs-comment'>-- # 'input' has two true values, so output has two coordinates.</span>
+<a name="line-5692"></a><span class='hs-comment'>-- # 'input' has rank of 2, so coordinates have two indices.</span>
+<a name="line-5693"></a><span class='hs-comment'>-- where(input) ==&gt; [[0, 0],</span>
+<a name="line-5694"></a><span class='hs-comment'>--                   [1, 0]]</span>
+<a name="line-5695"></a><span class='hs-comment'>-- </span>
+<a name="line-5696"></a><span class='hs-comment'>-- # `input` tensor is [[[True, False]</span>
+<a name="line-5697"></a><span class='hs-comment'>-- #                     [True, False]]</span>
+<a name="line-5698"></a><span class='hs-comment'>-- #                    [[False, True]</span>
+<a name="line-5699"></a><span class='hs-comment'>-- #                     [False, True]]</span>
+<a name="line-5700"></a><span class='hs-comment'>-- #                    [[False, False]</span>
+<a name="line-5701"></a><span class='hs-comment'>-- #                     [False, True]]]</span>
+<a name="line-5702"></a><span class='hs-comment'>-- # 'input' has 5 true values, so output has 5 coordinates.</span>
+<a name="line-5703"></a><span class='hs-comment'>-- # 'input' has rank of 3, so coordinates have three indices.</span>
+<a name="line-5704"></a><span class='hs-comment'>-- where(input) ==&gt; [[0, 0, 0],</span>
+<a name="line-5705"></a><span class='hs-comment'>--                   [0, 1, 0],</span>
+<a name="line-5706"></a><span class='hs-comment'>--                   [1, 0, 1],</span>
+<a name="line-5707"></a><span class='hs-comment'>--                   [1, 1, 1],</span>
+<a name="line-5708"></a><span class='hs-comment'>--                   [2, 1, 1]]</span>
+<a name="line-5709"></a><span class='hs-comment'>-- ```</span>
+<a name="line-5710"></a><span class='hs-definition'>where'</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-5711"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __index__</span>
+<a name="line-5712"></a><span class='hs-definition'>where'</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5713"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Where"</span><span class='hs-layout'>)</span>
+<a name="line-5714"></a>        <span class='hs-varid'>input</span>
+<a name="line-5715"></a><span class='hs-comment'>{-
+<a name="line-5716"></a>input_arg { name: "input" type: DT_BOOL }
+<a name="line-5717"></a>output_arg { name: "index" type: DT_INT64 }
+<a name="line-5718"></a>-}</span>
+<a name="line-5719"></a>
+<a name="line-5720"></a><a name="listDiff"></a><span class='hs-comment'>-- | Computes the difference between two lists of numbers or strings.</span>
+<a name="line-5721"></a><span class='hs-comment'>--</span>
+<a name="line-5722"></a><span class='hs-comment'>-- Given a list `x` and a list `y`, this operation returns a list `out` that</span>
+<a name="line-5723"></a><span class='hs-comment'>-- represents all values that are in `x` but not in `y`. The returned list `out`</span>
+<a name="line-5724"></a><span class='hs-comment'>-- is sorted in the same order that the numbers appear in `x` (duplicates are</span>
+<a name="line-5725"></a><span class='hs-comment'>-- preserved). This operation also returns a list `idx` that represents the</span>
+<a name="line-5726"></a><span class='hs-comment'>-- position of each `out` element in `x`. In other words:</span>
+<a name="line-5727"></a><span class='hs-comment'>-- </span>
+<a name="line-5728"></a><span class='hs-comment'>-- `out[i] = x[idx[i]] for i in [0, 1, ..., len(out) - 1]`</span>
+<a name="line-5729"></a><span class='hs-comment'>-- </span>
+<a name="line-5730"></a><span class='hs-comment'>-- For example, given this input:</span>
+<a name="line-5731"></a><span class='hs-comment'>-- </span>
+<a name="line-5732"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-5733"></a><span class='hs-comment'>-- x = [1, 2, 3, 4, 5, 6]</span>
+<a name="line-5734"></a><span class='hs-comment'>-- y = [1, 3, 5]</span>
+<a name="line-5735"></a><span class='hs-comment'>-- ```</span>
+<a name="line-5736"></a><span class='hs-comment'>-- </span>
+<a name="line-5737"></a><span class='hs-comment'>-- This operation would return:</span>
+<a name="line-5738"></a><span class='hs-comment'>-- </span>
+<a name="line-5739"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-5740"></a><span class='hs-comment'>-- out ==&gt; [2, 4, 6]</span>
+<a name="line-5741"></a><span class='hs-comment'>-- idx ==&gt; [1, 3, 5]</span>
+<a name="line-5742"></a><span class='hs-comment'>-- ```</span>
+<a name="line-5743"></a><span class='hs-definition'>listDiff</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>out_idx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>out_idx</span><span class='hs-layout'>,</span>
+<a name="line-5744"></a>                                      <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-5745"></a>                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>out_idx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-5746"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__: 1-D. Values to keep.</span>
+<a name="line-5747"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__: 1-D. Values to remove.</span>
+<a name="line-5748"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>out_idx</span><span class='hs-layout'>)</span>
+<a name="line-5749"></a>            <span class='hs-comment'>-- ^ (__out__, __idx__)</span>
+<a name="line-5750"></a>            <span class='hs-comment'>--</span>
+<a name="line-5751"></a>            <span class='hs-comment'>-- * __out__: 1-D. Values present in `x` but not in `y`.</span>
+<a name="line-5752"></a>            <span class='hs-comment'>--</span>
+<a name="line-5753"></a>            <span class='hs-comment'>-- * __idx__: 1-D. Positions of `x` values preserved in `out`.</span>
+<a name="line-5754"></a><span class='hs-definition'>listDiff</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5755"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ListDiff"</span>
+<a name="line-5756"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-5757"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"out_idx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>out_idx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-5758"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-5759"></a><span class='hs-comment'>{-
+<a name="line-5760"></a>attr { name: "T" type: "type" }
+<a name="line-5761"></a>attr {
+<a name="line-5762"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-5763"></a>  default_value { type: DT_INT32 }
+<a name="line-5764"></a>  name: "out_idx"
+<a name="line-5765"></a>  type: "type"
+<a name="line-5766"></a>}
+<a name="line-5767"></a>input_arg {
+<a name="line-5768"></a>  description: "1-D. Values to keep." name: "x" type_attr: "T"
+<a name="line-5769"></a>}
+<a name="line-5770"></a>input_arg {
+<a name="line-5771"></a>  description: "1-D. Values to remove." name: "y" type_attr: "T"
+<a name="line-5772"></a>}
+<a name="line-5773"></a>output_arg {
+<a name="line-5774"></a>  description: "1-D. Values present in `x` but not in `y`."
+<a name="line-5775"></a>  name: "out"
+<a name="line-5776"></a>  type_attr: "T"
+<a name="line-5777"></a>}
+<a name="line-5778"></a>output_arg {
+<a name="line-5779"></a>  description: "1-D. Positions of `x` values preserved in `out`."
+<a name="line-5780"></a>  name: "idx"
+<a name="line-5781"></a>  type_attr: "out_idx"
+<a name="line-5782"></a>}
+<a name="line-5783"></a>-}</span>
+<a name="line-5784"></a>
+<a name="line-5785"></a><a name="stridedSlice"></a><span class='hs-comment'>-- | Return a strided slice from `input`.</span>
+<a name="line-5786"></a><span class='hs-comment'>--</span>
+<a name="line-5787"></a><span class='hs-comment'>-- The output tensor is a tensor with dimensions implied by `begin`,</span>
+<a name="line-5788"></a><span class='hs-comment'>-- `end`, and `strides`, whose values are extracted from `begin`.</span>
+<a name="line-5789"></a><span class='hs-comment'>-- </span>
+<a name="line-5790"></a><span class='hs-comment'>-- Specifically, the result tensor at index `(i[0], i[1], ..., i[n-1])`</span>
+<a name="line-5791"></a><span class='hs-comment'>-- will obtain the value `input[begin[0] + i[0] * stride[0], ..., `</span>
+<a name="line-5792"></a><span class='hs-comment'>--                             `begin[n-1] + i[n-1] * stride[n-1])]`.</span>
+<a name="line-5793"></a><span class='hs-comment'>-- </span>
+<a name="line-5794"></a><span class='hs-comment'>-- *Requirements*:</span>
+<a name="line-5795"></a><span class='hs-comment'>--   `0 != strides[i] for i in [0, n)`</span>
+<a name="line-5796"></a><span class='hs-definition'>stridedSlice</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>index</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>index</span><span class='hs-layout'>,</span>
+<a name="line-5797"></a>                                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-5798"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>index</span><span class='hs-layout'>,</span>
+<a name="line-5799"></a>                                              <span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-5800"></a>                <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-5801"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>index</span> <span class='hs-comment'>-- ^ __begin__: `begin[i]` specifies the offset into the `i`th dimension of</span>
+<a name="line-5802"></a>                                   <span class='hs-comment'>-- `input` to slice from.</span>
+<a name="line-5803"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>index</span> <span class='hs-comment'>-- ^ __end__: `end[i]` specifies the first offset into the `i`th dimension of</span>
+<a name="line-5804"></a>                                   <span class='hs-comment'>-- `input` that will not be extracted. Out or range values are</span>
+<a name="line-5805"></a>                                   <span class='hs-comment'>-- clamped to `[0,dim[i]) if slice[i] &gt; 0` or `[-1,dim[i]-1]`</span>
+<a name="line-5806"></a>                                   <span class='hs-comment'>-- `if slice[i] &lt; 0`</span>
+<a name="line-5807"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>index</span> <span class='hs-comment'>-- ^ __strides__: `strides[i]` specifies the increment in the `i`th dimension</span>
+<a name="line-5808"></a>                                   <span class='hs-comment'>-- after extracting a given element. Negative indices will reverse</span>
+<a name="line-5809"></a>                                   <span class='hs-comment'>-- the original order. Out or range values are</span>
+<a name="line-5810"></a>                                   <span class='hs-comment'>-- clamped to `[0,dim[i]) if slice[i]&gt;0` or `[-1,dim[i]-1] if slice[i] &lt; 0`</span>
+<a name="line-5811"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-5812"></a><span class='hs-definition'>stridedSlice</span> <span class='hs-varid'>input</span> <span class='hs-varid'>begin</span> <span class='hs-varid'>end</span> <span class='hs-varid'>strides</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5813"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"StridedSlice"</span>
+<a name="line-5814"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Index"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>index</span><span class='hs-layout'>)</span>
+<a name="line-5815"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-5816"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>begin</span> <span class='hs-varid'>end</span> <span class='hs-varid'>strides</span>
+<a name="line-5817"></a><span class='hs-comment'>{-
+<a name="line-5818"></a>attr { name: "T" type: "type" }
+<a name="line-5819"></a>attr {
+<a name="line-5820"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-5821"></a>  name: "Index"
+<a name="line-5822"></a>  type: "type"
+<a name="line-5823"></a>}
+<a name="line-5824"></a>attr {
+<a name="line-5825"></a>  default_value { i: 0 }
+<a name="line-5826"></a>  description: "a bitmask where a bit i being 1 means to ignore the begin\nvalue and instead use the largest interval possible. At runtime\nbegin[i] will be replaced with `[0, n-1) if `stride[i] &gt; 0` or\n`[-1, n-1]` if `stride[i] &lt; 0`"
+<a name="line-5827"></a>  name: "begin_mask"
+<a name="line-5828"></a>  type: "int"
+<a name="line-5829"></a>}
+<a name="line-5830"></a>attr {
+<a name="line-5831"></a>  default_value { i: 0 }
+<a name="line-5832"></a>  description: "analogous to `begin_mask`"
+<a name="line-5833"></a>  name: "end_mask"
+<a name="line-5834"></a>  type: "int"
+<a name="line-5835"></a>}
+<a name="line-5836"></a>attr {
+<a name="line-5837"></a>  default_value { i: 0 }
+<a name="line-5838"></a>  description: "a bitmask where bit `i` being 1 means the `i`th\nposition is actually an ellipsis. One bit at most can be 1."
+<a name="line-5839"></a>  name: "ellipsis_mask"
+<a name="line-5840"></a>  type: "int"
+<a name="line-5841"></a>}
+<a name="line-5842"></a>attr {
+<a name="line-5843"></a>  default_value { i: 0 }
+<a name="line-5844"></a>  description: "a bitmask where bit `i` being 1 means the `i`th\nposition creates a dimension in the tensor of length 1. Thus\nthe total number of elements remain unchanged but the shape\ngets a 1 in the appropriate position."
+<a name="line-5845"></a>  name: "new_axis_mask"
+<a name="line-5846"></a>  type: "int"
+<a name="line-5847"></a>}
+<a name="line-5848"></a>attr {
+<a name="line-5849"></a>  default_value { i: 0 }
+<a name="line-5850"></a>  description: "a bitmask where bit `i` implies that the `i`th\nposition should shrink the dimensionality. begin and end\nmust imply a slice of size 1 in the dimension. For example in\npython one might do `foo[:,3,:]` which would result in\n`shrink_axis_mask` being 2."
+<a name="line-5851"></a>  name: "shrink_axis_mask"
+<a name="line-5852"></a>  type: "int"
+<a name="line-5853"></a>}
+<a name="line-5854"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-5855"></a>input_arg {
+<a name="line-5856"></a>  description: "`begin[i]` specifies the offset into the `i`th dimension of\n`input` to slice from."
+<a name="line-5857"></a>  name: "begin"
+<a name="line-5858"></a>  type_attr: "Index"
+<a name="line-5859"></a>}
+<a name="line-5860"></a>input_arg {
+<a name="line-5861"></a>  description: "`end[i]` specifies the first offset into the `i`th dimension of\n`input` that will not be extracted. Out or range values are\nclamped to `[0,dim[i]) if slice[i] &gt; 0` or `[-1,dim[i]-1]`\n`if slice[i] &lt; 0`"
+<a name="line-5862"></a>  name: "end"
+<a name="line-5863"></a>  type_attr: "Index"
+<a name="line-5864"></a>}
+<a name="line-5865"></a>input_arg {
+<a name="line-5866"></a>  description: "`strides[i]` specifies the increment in the `i`th dimension\nafter extracting a given element. Negative indices will reverse\nthe original order. Out or range values are\nclamped to `[0,dim[i]) if slice[i]&gt;0` or `[-1,dim[i]-1] if slice[i] &lt; 0`"
+<a name="line-5867"></a>  name: "strides"
+<a name="line-5868"></a>  type_attr: "Index"
+<a name="line-5869"></a>}
+<a name="line-5870"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-5871"></a>-}</span>
+<a name="line-5872"></a>
+<a name="line-5873"></a><span class='hs-comment'>-- | A queue that randomizes the order of elements.</span>
+<a name="line-5874"></a>
+<a name="line-5875"></a><a name="randomShuffleQueue"></a><span class='hs-definition'>randomShuffleQueue</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to the queue.</span>
+<a name="line-5876"></a><span class='hs-definition'>randomShuffleQueue</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5877"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RandomShuffleQueue"</span><span class='hs-layout'>)</span>
+<a name="line-5878"></a>        
+<a name="line-5879"></a><span class='hs-comment'>{-
+<a name="line-5880"></a>attr {
+<a name="line-5881"></a>  description: "The type of each component in a value."
+<a name="line-5882"></a>  has_minimum: true
+<a name="line-5883"></a>  minimum: 1
+<a name="line-5884"></a>  name: "component_types"
+<a name="line-5885"></a>  type: "list(type)"
+<a name="line-5886"></a>}
+<a name="line-5887"></a>attr {
+<a name="line-5888"></a>  default_value { list { } }
+<a name="line-5889"></a>  description: "The shape of each component in a value. The length of this attr must\nbe either 0 or the same as the length of component_types. If the length of\nthis attr is 0, the shapes of queue elements are not constrained, and\nonly one element may be dequeued at a time."
+<a name="line-5890"></a>  has_minimum: true
+<a name="line-5891"></a>  name: "shapes"
+<a name="line-5892"></a>  type: "list(shape)"
+<a name="line-5893"></a>}
+<a name="line-5894"></a>attr {
+<a name="line-5895"></a>  default_value { i: -1 }
+<a name="line-5896"></a>  description: "The upper bound on the number of elements in this queue.\nNegative numbers mean no limit."
+<a name="line-5897"></a>  name: "capacity"
+<a name="line-5898"></a>  type: "int"
+<a name="line-5899"></a>}
+<a name="line-5900"></a>attr {
+<a name="line-5901"></a>  default_value { i: 0 }
+<a name="line-5902"></a>  description: "Dequeue will block unless there would be this\nmany elements after the dequeue or the queue is closed. This\nensures a minimum level of mixing of elements."
+<a name="line-5903"></a>  name: "min_after_dequeue"
+<a name="line-5904"></a>  type: "int"
+<a name="line-5905"></a>}
+<a name="line-5906"></a>attr {
+<a name="line-5907"></a>  default_value { i: 0 }
+<a name="line-5908"></a>  description: "If either seed or seed2 is set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, a random seed is used."
+<a name="line-5909"></a>  name: "seed"
+<a name="line-5910"></a>  type: "int"
+<a name="line-5911"></a>}
+<a name="line-5912"></a>attr {
+<a name="line-5913"></a>  default_value { i: 0 }
+<a name="line-5914"></a>  description: "A second seed to avoid seed collision."
+<a name="line-5915"></a>  name: "seed2"
+<a name="line-5916"></a>  type: "int"
+<a name="line-5917"></a>}
+<a name="line-5918"></a>attr {
+<a name="line-5919"></a>  default_value { s: "" }
+<a name="line-5920"></a>  description: "If non-empty, this queue is placed in the given container.\nOtherwise, a default container is used."
+<a name="line-5921"></a>  name: "container"
+<a name="line-5922"></a>  type: "string"
+<a name="line-5923"></a>}
+<a name="line-5924"></a>attr {
+<a name="line-5925"></a>  default_value { s: "" }
+<a name="line-5926"></a>  description: "If non-empty, this queue will be shared under the given name\nacross multiple sessions."
+<a name="line-5927"></a>  name: "shared_name"
+<a name="line-5928"></a>  type: "string"
+<a name="line-5929"></a>}
+<a name="line-5930"></a>output_arg {
+<a name="line-5931"></a>  description: "The handle to the queue."
+<a name="line-5932"></a>  is_ref: true
+<a name="line-5933"></a>  name: "handle"
+<a name="line-5934"></a>  type: DT_STRING
+<a name="line-5935"></a>}
+<a name="line-5936"></a>-}</span>
+<a name="line-5937"></a>
+<a name="line-5938"></a><a name="tileGrad"></a><span class='hs-comment'>-- | Returns the gradient of `Tile`.</span>
+<a name="line-5939"></a><span class='hs-comment'>--</span>
+<a name="line-5940"></a><span class='hs-comment'>-- Since `Tile` takes an input and repeats the input `multiples` times</span>
+<a name="line-5941"></a><span class='hs-comment'>-- along each dimension, `TileGrad` takes in `multiples` and aggregates</span>
+<a name="line-5942"></a><span class='hs-comment'>-- each repeated tile of `input` into `output`.</span>
+<a name="line-5943"></a><span class='hs-definition'>tileGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-5944"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __multiples__</span>
+<a name="line-5945"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-5946"></a><span class='hs-definition'>tileGrad</span> <span class='hs-varid'>input</span> <span class='hs-varid'>multiples</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5947"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TileGrad"</span>
+<a name="line-5948"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-5949"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>multiples</span>
+<a name="line-5950"></a><span class='hs-comment'>{-
+<a name="line-5951"></a>attr { name: "T" type: "type" }
+<a name="line-5952"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-5953"></a>input_arg { name: "multiples" type: DT_INT32 }
+<a name="line-5954"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-5955"></a>-}</span>
+<a name="line-5956"></a>
+<a name="line-5957"></a><a name="stridedSliceAssign"></a><span class='hs-comment'>-- | Assign `value` to the sliced l-value reference of `ref`.</span>
+<a name="line-5958"></a><span class='hs-comment'>--</span>
+<a name="line-5959"></a><span class='hs-comment'>-- The values of `value` are assigned to the positions in the variable</span>
+<a name="line-5960"></a><span class='hs-comment'>-- `ref` that are selected by the slice parameters. The slice parameters</span>
+<a name="line-5961"></a><span class='hs-comment'>-- `begin, `end`, `strides`, etc. work exactly as in `StridedSlice`.</span>
+<a name="line-5962"></a><span class='hs-comment'>-- </span>
+<a name="line-5963"></a><span class='hs-comment'>-- NOTE this op currently does not support broadcasting and so `value`'s</span>
+<a name="line-5964"></a><span class='hs-comment'>-- shape must be exactly the shape produced by the slice of `ref`.</span>
+<a name="line-5965"></a><span class='hs-definition'>stridedSliceAssign</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>index</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>index</span><span class='hs-layout'>,</span>
+<a name="line-5966"></a>                                                       <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-5967"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>index</span><span class='hs-layout'>,</span>
+<a name="line-5968"></a>                                                       <span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-5969"></a>                      <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __ref__</span>
+<a name="line-5970"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>index</span> <span class='hs-comment'>-- ^ __begin__</span>
+<a name="line-5971"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>index</span> <span class='hs-comment'>-- ^ __end__</span>
+<a name="line-5972"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>index</span> <span class='hs-comment'>-- ^ __strides__</span>
+<a name="line-5973"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__</span>
+<a name="line-5974"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output_ref__</span>
+<a name="line-5975"></a><span class='hs-definition'>stridedSliceAssign</span> <span class='hs-varid'>ref</span> <span class='hs-varid'>begin</span> <span class='hs-varid'>end</span> <span class='hs-varid'>strides</span> <span class='hs-varid'>value</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-5976"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"StridedSliceAssign"</span>
+<a name="line-5977"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Index"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>index</span><span class='hs-layout'>)</span>
+<a name="line-5978"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-5979"></a>        <span class='hs-varid'>ref</span> <span class='hs-varid'>begin</span> <span class='hs-varid'>end</span> <span class='hs-varid'>strides</span> <span class='hs-varid'>value</span>
+<a name="line-5980"></a><span class='hs-comment'>{-
+<a name="line-5981"></a>attr { name: "T" type: "type" }
+<a name="line-5982"></a>attr {
+<a name="line-5983"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-5984"></a>  name: "Index"
+<a name="line-5985"></a>  type: "type"
+<a name="line-5986"></a>}
+<a name="line-5987"></a>attr { default_value { i: 0 } name: "begin_mask" type: "int" }
+<a name="line-5988"></a>attr { default_value { i: 0 } name: "end_mask" type: "int" }
+<a name="line-5989"></a>attr { default_value { i: 0 } name: "ellipsis_mask" type: "int" }
+<a name="line-5990"></a>attr { default_value { i: 0 } name: "new_axis_mask" type: "int" }
+<a name="line-5991"></a>attr {
+<a name="line-5992"></a>  default_value { i: 0 } name: "shrink_axis_mask" type: "int"
+<a name="line-5993"></a>}
+<a name="line-5994"></a>input_arg { is_ref: true name: "ref" type_attr: "T" }
+<a name="line-5995"></a>input_arg { name: "begin" type_attr: "Index" }
+<a name="line-5996"></a>input_arg { name: "end" type_attr: "Index" }
+<a name="line-5997"></a>input_arg { name: "strides" type_attr: "Index" }
+<a name="line-5998"></a>input_arg { name: "value" type_attr: "T" }
+<a name="line-5999"></a>output_arg { is_ref: true name: "output_ref" type_attr: "T" }
+<a name="line-6000"></a>-}</span>
+<a name="line-6001"></a>
+<a name="line-6002"></a><a name="reshape"></a><span class='hs-comment'>-- | Reshapes a tensor.</span>
+<a name="line-6003"></a><span class='hs-comment'>--</span>
+<a name="line-6004"></a><span class='hs-comment'>-- Given `tensor`, this operation returns a tensor that has the same values</span>
+<a name="line-6005"></a><span class='hs-comment'>-- as `tensor` with shape `shape`.</span>
+<a name="line-6006"></a><span class='hs-comment'>-- </span>
+<a name="line-6007"></a><span class='hs-comment'>-- If one component of `shape` is the special value -1, the size of that dimension</span>
+<a name="line-6008"></a><span class='hs-comment'>-- is computed so that the total size remains constant.  In particular, a `shape`</span>
+<a name="line-6009"></a><span class='hs-comment'>-- of `[-1]` flattens into 1-D.  At most one component of `shape` can be -1.</span>
+<a name="line-6010"></a><span class='hs-comment'>-- </span>
+<a name="line-6011"></a><span class='hs-comment'>-- If `shape` is 1-D or higher, then the operation returns a tensor with shape</span>
+<a name="line-6012"></a><span class='hs-comment'>-- `shape` filled with the values of `tensor`. In this case, the number of elements</span>
+<a name="line-6013"></a><span class='hs-comment'>-- implied by `shape` must be the same as the number of elements in `tensor`.</span>
+<a name="line-6014"></a><span class='hs-comment'>-- </span>
+<a name="line-6015"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-6016"></a><span class='hs-comment'>-- </span>
+<a name="line-6017"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-6018"></a><span class='hs-comment'>-- # tensor 't' is [1, 2, 3, 4, 5, 6, 7, 8, 9]</span>
+<a name="line-6019"></a><span class='hs-comment'>-- # tensor 't' has shape [9]</span>
+<a name="line-6020"></a><span class='hs-comment'>-- reshape(t, [3, 3]) ==&gt; [[1, 2, 3],</span>
+<a name="line-6021"></a><span class='hs-comment'>--                         [4, 5, 6],</span>
+<a name="line-6022"></a><span class='hs-comment'>--                         [7, 8, 9]]</span>
+<a name="line-6023"></a><span class='hs-comment'>-- </span>
+<a name="line-6024"></a><span class='hs-comment'>-- # tensor 't' is [[[1, 1], [2, 2]],</span>
+<a name="line-6025"></a><span class='hs-comment'>-- #                [[3, 3], [4, 4]]]</span>
+<a name="line-6026"></a><span class='hs-comment'>-- # tensor 't' has shape [2, 2, 2]</span>
+<a name="line-6027"></a><span class='hs-comment'>-- reshape(t, [2, 4]) ==&gt; [[1, 1, 2, 2],</span>
+<a name="line-6028"></a><span class='hs-comment'>--                         [3, 3, 4, 4]]</span>
+<a name="line-6029"></a><span class='hs-comment'>-- </span>
+<a name="line-6030"></a><span class='hs-comment'>-- # tensor 't' is [[[1, 1, 1],</span>
+<a name="line-6031"></a><span class='hs-comment'>-- #                 [2, 2, 2]],</span>
+<a name="line-6032"></a><span class='hs-comment'>-- #                [[3, 3, 3],</span>
+<a name="line-6033"></a><span class='hs-comment'>-- #                 [4, 4, 4]],</span>
+<a name="line-6034"></a><span class='hs-comment'>-- #                [[5, 5, 5],</span>
+<a name="line-6035"></a><span class='hs-comment'>-- #                 [6, 6, 6]]]</span>
+<a name="line-6036"></a><span class='hs-comment'>-- # tensor 't' has shape [3, 2, 3]</span>
+<a name="line-6037"></a><span class='hs-comment'>-- # pass '[-1]' to flatten 't'</span>
+<a name="line-6038"></a><span class='hs-comment'>-- reshape(t, [-1]) ==&gt; [1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6, 6, 6]</span>
+<a name="line-6039"></a><span class='hs-comment'>-- </span>
+<a name="line-6040"></a><span class='hs-comment'>-- # -1 can also be used to infer the shape</span>
+<a name="line-6041"></a><span class='hs-comment'>-- </span>
+<a name="line-6042"></a><span class='hs-comment'>-- # -1 is inferred to be 9:</span>
+<a name="line-6043"></a><span class='hs-comment'>-- reshape(t, [2, -1]) ==&gt; [[1, 1, 1, 2, 2, 2, 3, 3, 3],</span>
+<a name="line-6044"></a><span class='hs-comment'>--                          [4, 4, 4, 5, 5, 5, 6, 6, 6]]</span>
+<a name="line-6045"></a><span class='hs-comment'>-- # -1 is inferred to be 2:</span>
+<a name="line-6046"></a><span class='hs-comment'>-- reshape(t, [-1, 9]) ==&gt; [[1, 1, 1, 2, 2, 2, 3, 3, 3],</span>
+<a name="line-6047"></a><span class='hs-comment'>--                          [4, 4, 4, 5, 5, 5, 6, 6, 6]]</span>
+<a name="line-6048"></a><span class='hs-comment'>-- # -1 is inferred to be 3:</span>
+<a name="line-6049"></a><span class='hs-comment'>-- reshape(t, [ 2, -1, 3]) ==&gt; [[[1, 1, 1],</span>
+<a name="line-6050"></a><span class='hs-comment'>--                               [2, 2, 2],</span>
+<a name="line-6051"></a><span class='hs-comment'>--                               [3, 3, 3]],</span>
+<a name="line-6052"></a><span class='hs-comment'>--                              [[4, 4, 4],</span>
+<a name="line-6053"></a><span class='hs-comment'>--                               [5, 5, 5],</span>
+<a name="line-6054"></a><span class='hs-comment'>--                               [6, 6, 6]]]</span>
+<a name="line-6055"></a><span class='hs-comment'>-- </span>
+<a name="line-6056"></a><span class='hs-comment'>-- # tensor 't' is [7]</span>
+<a name="line-6057"></a><span class='hs-comment'>-- # shape `[]` reshapes to a scalar</span>
+<a name="line-6058"></a><span class='hs-comment'>-- reshape(t, []) ==&gt; 7</span>
+<a name="line-6059"></a><span class='hs-comment'>-- ```</span>
+<a name="line-6060"></a><span class='hs-definition'>reshape</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tshape</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tshape</span><span class='hs-layout'>,</span>
+<a name="line-6061"></a>                                    <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-6062"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tshape</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-6063"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __tensor__</span>
+<a name="line-6064"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tshape</span> <span class='hs-comment'>-- ^ __shape__: Defines the shape of the output tensor.</span>
+<a name="line-6065"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-6066"></a><span class='hs-definition'>reshape</span> <span class='hs-varid'>tensor</span> <span class='hs-varid'>shape</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6067"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Reshape"</span>
+<a name="line-6068"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-6069"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tshape"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tshape</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-6070"></a>        <span class='hs-varid'>tensor</span> <span class='hs-varid'>shape</span>
+<a name="line-6071"></a><span class='hs-comment'>{-
+<a name="line-6072"></a>attr { name: "T" type: "type" }
+<a name="line-6073"></a>attr {
+<a name="line-6074"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-6075"></a>  default_value { type: DT_INT32 }
+<a name="line-6076"></a>  name: "Tshape"
+<a name="line-6077"></a>  type: "type"
+<a name="line-6078"></a>}
+<a name="line-6079"></a>input_arg { name: "tensor" type_attr: "T" }
+<a name="line-6080"></a>input_arg {
+<a name="line-6081"></a>  description: "Defines the shape of the output tensor."
+<a name="line-6082"></a>  name: "shape"
+<a name="line-6083"></a>  type_attr: "Tshape"
+<a name="line-6084"></a>}
+<a name="line-6085"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-6086"></a>-}</span>
+<a name="line-6087"></a>
+<a name="line-6088"></a><span class='hs-comment'>-- | A queue that produces elements in first-in first-out order.</span>
+<a name="line-6089"></a>
+<a name="line-6090"></a><a name="fIFOQueue"></a><span class='hs-definition'>fIFOQueue</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to the queue.</span>
+<a name="line-6091"></a><span class='hs-definition'>fIFOQueue</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6092"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"FIFOQueue"</span><span class='hs-layout'>)</span>
+<a name="line-6093"></a>        
+<a name="line-6094"></a><span class='hs-comment'>{-
+<a name="line-6095"></a>attr {
+<a name="line-6096"></a>  description: "The type of each component in a value."
+<a name="line-6097"></a>  has_minimum: true
+<a name="line-6098"></a>  minimum: 1
+<a name="line-6099"></a>  name: "component_types"
+<a name="line-6100"></a>  type: "list(type)"
+<a name="line-6101"></a>}
+<a name="line-6102"></a>attr {
+<a name="line-6103"></a>  default_value { list { } }
+<a name="line-6104"></a>  description: "The shape of each component in a value. The length of this attr must\nbe either 0 or the same as the length of component_types. If the length of\nthis attr is 0, the shapes of queue elements are not constrained, and\nonly one element may be dequeued at a time."
+<a name="line-6105"></a>  has_minimum: true
+<a name="line-6106"></a>  name: "shapes"
+<a name="line-6107"></a>  type: "list(shape)"
+<a name="line-6108"></a>}
+<a name="line-6109"></a>attr {
+<a name="line-6110"></a>  default_value { i: -1 }
+<a name="line-6111"></a>  description: "The upper bound on the number of elements in this queue.\nNegative numbers mean no limit."
+<a name="line-6112"></a>  name: "capacity"
+<a name="line-6113"></a>  type: "int"
+<a name="line-6114"></a>}
+<a name="line-6115"></a>attr {
+<a name="line-6116"></a>  default_value { s: "" }
+<a name="line-6117"></a>  description: "If non-empty, this queue is placed in the given container.\nOtherwise, a default container is used."
+<a name="line-6118"></a>  name: "container"
+<a name="line-6119"></a>  type: "string"
+<a name="line-6120"></a>}
+<a name="line-6121"></a>attr {
+<a name="line-6122"></a>  default_value { s: "" }
+<a name="line-6123"></a>  description: "If non-empty, this queue will be shared under the given name\nacross multiple sessions."
+<a name="line-6124"></a>  name: "shared_name"
+<a name="line-6125"></a>  type: "string"
+<a name="line-6126"></a>}
+<a name="line-6127"></a>output_arg {
+<a name="line-6128"></a>  description: "The handle to the queue."
+<a name="line-6129"></a>  is_ref: true
+<a name="line-6130"></a>  name: "handle"
+<a name="line-6131"></a>  type: DT_STRING
+<a name="line-6132"></a>}
+<a name="line-6133"></a>-}</span>
+<a name="line-6134"></a>
+<a name="line-6135"></a><a name="learnedUnigramCandidateSampler"></a><span class='hs-comment'>-- | Generates labels for candidate sampling with a learned unigram distribution.</span>
+<a name="line-6136"></a><span class='hs-comment'>--</span>
+<a name="line-6137"></a><span class='hs-comment'>-- See explanations of candidate sampling and the data formats at</span>
+<a name="line-6138"></a><span class='hs-comment'>-- go/candidate-sampling.</span>
+<a name="line-6139"></a><span class='hs-comment'>-- </span>
+<a name="line-6140"></a><span class='hs-comment'>-- For each batch, this op picks a single set of sampled candidate labels.</span>
+<a name="line-6141"></a><span class='hs-comment'>-- </span>
+<a name="line-6142"></a><span class='hs-comment'>-- The advantages of sampling candidates per-batch are simplicity and the</span>
+<a name="line-6143"></a><span class='hs-comment'>-- possibility of efficient dense matrix multiplication. The disadvantage is that</span>
+<a name="line-6144"></a><span class='hs-comment'>-- the sampled candidates must be chosen independently of the context and of the</span>
+<a name="line-6145"></a><span class='hs-comment'>-- true labels.</span>
+<a name="line-6146"></a><span class='hs-definition'>learnedUnigramCandidateSampler</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_sampled__: Number of candidates to randomly sample per batch.</span>
+<a name="line-6147"></a>                                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_true__: Number of true labels per context.</span>
+<a name="line-6148"></a>                                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __range_max__: The sampler will sample integers from the interval [0, range_max).</span>
+<a name="line-6149"></a>                                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __unique__: If unique is true, we sample with rejection, so that all sampled</span>
+<a name="line-6150"></a>                                          <span class='hs-comment'>-- candidates in a batch are unique. This requires some approximation to</span>
+<a name="line-6151"></a>                                          <span class='hs-comment'>-- estimate the post-rejection sampling probabilities.</span>
+<a name="line-6152"></a>                                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __true_classes__: A batch_size * num_true matrix, in which each row contains the</span>
+<a name="line-6153"></a>                                                              <span class='hs-comment'>-- IDs of the num_true target_classes in the corresponding original label.</span>
+<a name="line-6154"></a>                                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-6155"></a>                                      <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span>
+<a name="line-6156"></a>                                  <span class='hs-comment'>-- ^ (__sampled_candidates__, __true_expected_count__, __sampled_expected_count__)</span>
+<a name="line-6157"></a>                                  <span class='hs-comment'>--</span>
+<a name="line-6158"></a>                                  <span class='hs-comment'>-- * __sampled_candidates__: A vector of length num_sampled, in which each element is</span>
+<a name="line-6159"></a>                                  <span class='hs-comment'>-- the ID of a sampled candidate.</span>
+<a name="line-6160"></a>                                  <span class='hs-comment'>--</span>
+<a name="line-6161"></a>                                  <span class='hs-comment'>-- * __true_expected_count__: A batch_size * num_true matrix, representing</span>
+<a name="line-6162"></a>                                  <span class='hs-comment'>-- the number of times each candidate is expected to occur in a batch</span>
+<a name="line-6163"></a>                                  <span class='hs-comment'>-- of sampled candidates. If unique=true, then this is a probability.</span>
+<a name="line-6164"></a>                                  <span class='hs-comment'>--</span>
+<a name="line-6165"></a>                                  <span class='hs-comment'>-- * __sampled_expected_count__: A vector of length num_sampled, for each sampled</span>
+<a name="line-6166"></a>                                  <span class='hs-comment'>-- candidate representing the number of times the candidate is expected</span>
+<a name="line-6167"></a>                                  <span class='hs-comment'>-- to occur in a batch of sampled candidates.  If unique=true, then this is a</span>
+<a name="line-6168"></a>                                  <span class='hs-comment'>-- probability.</span>
+<a name="line-6169"></a><span class='hs-definition'>learnedUnigramCandidateSampler</span> <span class='hs-varid'>num_sampled</span> <span class='hs-varid'>num_true</span> <span class='hs-varid'>range_max</span> <span class='hs-varid'>unique</span>
+<a name="line-6170"></a>                               <span class='hs-varid'>true_classes</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6171"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"LearnedUnigramCandidateSampler"</span>
+<a name="line-6172"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_sampled"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_sampled</span>
+<a name="line-6173"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_true"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_true</span>
+<a name="line-6174"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"range_max"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>range_max</span>
+<a name="line-6175"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"unique"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>unique</span><span class='hs-layout'>)</span>
+<a name="line-6176"></a>        <span class='hs-varid'>true_classes</span>
+<a name="line-6177"></a><span class='hs-comment'>{-
+<a name="line-6178"></a>attr {
+<a name="line-6179"></a>  description: "Number of true labels per context."
+<a name="line-6180"></a>  has_minimum: true
+<a name="line-6181"></a>  minimum: 1
+<a name="line-6182"></a>  name: "num_true"
+<a name="line-6183"></a>  type: "int"
+<a name="line-6184"></a>}
+<a name="line-6185"></a>attr {
+<a name="line-6186"></a>  description: "Number of candidates to randomly sample per batch."
+<a name="line-6187"></a>  has_minimum: true
+<a name="line-6188"></a>  minimum: 1
+<a name="line-6189"></a>  name: "num_sampled"
+<a name="line-6190"></a>  type: "int"
+<a name="line-6191"></a>}
+<a name="line-6192"></a>attr {
+<a name="line-6193"></a>  description: "If unique is true, we sample with rejection, so that all sampled\ncandidates in a batch are unique. This requires some approximation to\nestimate the post-rejection sampling probabilities."
+<a name="line-6194"></a>  name: "unique"
+<a name="line-6195"></a>  type: "bool"
+<a name="line-6196"></a>}
+<a name="line-6197"></a>attr {
+<a name="line-6198"></a>  description: "The sampler will sample integers from the interval [0, range_max)."
+<a name="line-6199"></a>  has_minimum: true
+<a name="line-6200"></a>  minimum: 1
+<a name="line-6201"></a>  name: "range_max"
+<a name="line-6202"></a>  type: "int"
+<a name="line-6203"></a>}
+<a name="line-6204"></a>attr {
+<a name="line-6205"></a>  default_value { i: 0 }
+<a name="line-6206"></a>  description: "If either seed or seed2 are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-6207"></a>  name: "seed"
+<a name="line-6208"></a>  type: "int"
+<a name="line-6209"></a>}
+<a name="line-6210"></a>attr {
+<a name="line-6211"></a>  default_value { i: 0 }
+<a name="line-6212"></a>  description: "An second seed to avoid seed collision."
+<a name="line-6213"></a>  name: "seed2"
+<a name="line-6214"></a>  type: "int"
+<a name="line-6215"></a>}
+<a name="line-6216"></a>input_arg {
+<a name="line-6217"></a>  description: "A batch_size * num_true matrix, in which each row contains the\nIDs of the num_true target_classes in the corresponding original label."
+<a name="line-6218"></a>  name: "true_classes"
+<a name="line-6219"></a>  type: DT_INT64
+<a name="line-6220"></a>}
+<a name="line-6221"></a>output_arg {
+<a name="line-6222"></a>  description: "A vector of length num_sampled, in which each element is\nthe ID of a sampled candidate."
+<a name="line-6223"></a>  name: "sampled_candidates"
+<a name="line-6224"></a>  type: DT_INT64
+<a name="line-6225"></a>}
+<a name="line-6226"></a>output_arg {
+<a name="line-6227"></a>  description: "A batch_size * num_true matrix, representing\nthe number of times each candidate is expected to occur in a batch\nof sampled candidates. If unique=true, then this is a probability."
+<a name="line-6228"></a>  name: "true_expected_count"
+<a name="line-6229"></a>  type: DT_FLOAT
+<a name="line-6230"></a>}
+<a name="line-6231"></a>output_arg {
+<a name="line-6232"></a>  description: "A vector of length num_sampled, for each sampled\ncandidate representing the number of times the candidate is expected\nto occur in a batch of sampled candidates.  If unique=true, then this is a\nprobability."
+<a name="line-6233"></a>  name: "sampled_expected_count"
+<a name="line-6234"></a>  type: DT_FLOAT
+<a name="line-6235"></a>}
+<a name="line-6236"></a>-}</span>
+<a name="line-6237"></a>
+<a name="line-6238"></a><a name="fractionalAvgPool"></a><span class='hs-comment'>-- | Performs fractional average pooling on the input.</span>
+<a name="line-6239"></a><span class='hs-comment'>--</span>
+<a name="line-6240"></a><span class='hs-comment'>-- Fractional average pooling is similar to Fractional max pooling in the pooling</span>
+<a name="line-6241"></a><span class='hs-comment'>-- region generation step. The only difference is that after pooling regions are</span>
+<a name="line-6242"></a><span class='hs-comment'>-- generated, a mean operation is performed instead of a max operation in each</span>
+<a name="line-6243"></a><span class='hs-comment'>-- pooling region.</span>
+<a name="line-6244"></a><span class='hs-definition'>fractionalAvgPool</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-6245"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-6246"></a>                                                          <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-6247"></a>                     <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: 4-D with shape `[batch, height, width, channels]`.</span>
+<a name="line-6248"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-6249"></a>                         <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-6250"></a>                     <span class='hs-comment'>-- ^ (__output__, __row_pooling_sequence__, __col_pooling_sequence__)</span>
+<a name="line-6251"></a>                     <span class='hs-comment'>--</span>
+<a name="line-6252"></a>                     <span class='hs-comment'>-- * __output__: output tensor after fractional avg pooling.</span>
+<a name="line-6253"></a>                     <span class='hs-comment'>--</span>
+<a name="line-6254"></a>                     <span class='hs-comment'>-- * __row_pooling_sequence__: row pooling sequence, needed to calculate gradient.</span>
+<a name="line-6255"></a>                     <span class='hs-comment'>--</span>
+<a name="line-6256"></a>                     <span class='hs-comment'>-- * __col_pooling_sequence__: column pooling sequence, needed to calculate gradient.</span>
+<a name="line-6257"></a><span class='hs-definition'>fractionalAvgPool</span> <span class='hs-varid'>value</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6258"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"FractionalAvgPool"</span>
+<a name="line-6259"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-6260"></a>        <span class='hs-varid'>value</span>
+<a name="line-6261"></a><span class='hs-comment'>{-
+<a name="line-6262"></a>attr {
+<a name="line-6263"></a>  description: "Pooling ratio for each dimension of `value`, currently only\nsupports row and col dimension and should be &gt;= 1.0. For example, a valid\npooling ratio looks like [1.0, 1.44, 1.73, 1.0]. The first and last elements\nmust be 1.0 because we don\'t allow pooling on batch and channels\ndimensions. 1.44 and 1.73 are pooling ratio on height and width dimensions\nrespectively."
+<a name="line-6264"></a>  has_minimum: true
+<a name="line-6265"></a>  minimum: 4
+<a name="line-6266"></a>  name: "pooling_ratio"
+<a name="line-6267"></a>  type: "list(float)"
+<a name="line-6268"></a>}
+<a name="line-6269"></a>attr {
+<a name="line-6270"></a>  default_value { b: false }
+<a name="line-6271"></a>  description: "When set to True, generates the pooling sequence in a\npseudorandom fashion, otherwise, in a random fashion. Check paper [Benjamin\nGraham, Fractional Max-Pooling] (<a href="http://arxiv.org/abs/1412.6071)">http://arxiv.org/abs/1412.6071)</a> for\ndifference between pseudorandom and random."
+<a name="line-6272"></a>  name: "pseudo_random"
+<a name="line-6273"></a>  type: "bool"
+<a name="line-6274"></a>}
+<a name="line-6275"></a>attr {
+<a name="line-6276"></a>  default_value { b: false }
+<a name="line-6277"></a>  description: "When set to True, it means when pooling, the values at the boundary\nof adjacent pooling cells are used by both cells. For example:\n\n`index  0  1  2  3  4`\n\n`value  20 5  16 3  7`\n\nIf the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.\nThe result would be [41/3, 26/3] for fractional avg pooling."
+<a name="line-6278"></a>  name: "overlapping"
+<a name="line-6279"></a>  type: "bool"
+<a name="line-6280"></a>}
+<a name="line-6281"></a>attr {
+<a name="line-6282"></a>  default_value { b: false }
+<a name="line-6283"></a>  description: "When set to True, a fixed pooling region will be used when\niterating over a FractionalAvgPool node in the computation graph. Mainly used\nin unit test to make FractionalAvgPool deterministic."
+<a name="line-6284"></a>  name: "deterministic"
+<a name="line-6285"></a>  type: "bool"
+<a name="line-6286"></a>}
+<a name="line-6287"></a>attr {
+<a name="line-6288"></a>  default_value { i: 0 }
+<a name="line-6289"></a>  description: "If either seed or seed2 are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-6290"></a>  name: "seed"
+<a name="line-6291"></a>  type: "int"
+<a name="line-6292"></a>}
+<a name="line-6293"></a>attr {
+<a name="line-6294"></a>  default_value { i: 0 }
+<a name="line-6295"></a>  description: "An second seed to avoid seed collision."
+<a name="line-6296"></a>  name: "seed2"
+<a name="line-6297"></a>  type: "int"
+<a name="line-6298"></a>}
+<a name="line-6299"></a>attr {
+<a name="line-6300"></a>  allowed_values {
+<a name="line-6301"></a>    list {
+<a name="line-6302"></a>      type: DT_FLOAT type: DT_DOUBLE type: DT_INT32 type: DT_INT64
+<a name="line-6303"></a>    }
+<a name="line-6304"></a>  }
+<a name="line-6305"></a>  name: "T"
+<a name="line-6306"></a>  type: "type"
+<a name="line-6307"></a>}
+<a name="line-6308"></a>input_arg {
+<a name="line-6309"></a>  description: "4-D with shape `[batch, height, width, channels]`."
+<a name="line-6310"></a>  name: "value"
+<a name="line-6311"></a>  type_attr: "T"
+<a name="line-6312"></a>}
+<a name="line-6313"></a>output_arg {
+<a name="line-6314"></a>  description: "output tensor after fractional avg pooling."
+<a name="line-6315"></a>  name: "output"
+<a name="line-6316"></a>  type_attr: "T"
+<a name="line-6317"></a>}
+<a name="line-6318"></a>output_arg {
+<a name="line-6319"></a>  description: "row pooling sequence, needed to calculate gradient."
+<a name="line-6320"></a>  name: "row_pooling_sequence"
+<a name="line-6321"></a>  type: DT_INT64
+<a name="line-6322"></a>}
+<a name="line-6323"></a>output_arg {
+<a name="line-6324"></a>  description: "column pooling sequence, needed to calculate gradient."
+<a name="line-6325"></a>  name: "col_pooling_sequence"
+<a name="line-6326"></a>  type: DT_INT64
+<a name="line-6327"></a>}
+<a name="line-6328"></a>-}</span>
+<a name="line-6329"></a>
+<a name="line-6330"></a><a name="randomCrop"></a><span class='hs-comment'>-- | Randomly crop `image`.</span>
+<a name="line-6331"></a><span class='hs-comment'>--</span>
+<a name="line-6332"></a><span class='hs-comment'>-- `size` is a 1-D int64 tensor with 2 elements representing the crop height and</span>
+<a name="line-6333"></a><span class='hs-comment'>-- width.  The values must be non negative.</span>
+<a name="line-6334"></a><span class='hs-comment'>-- </span>
+<a name="line-6335"></a><span class='hs-comment'>-- This Op picks a random location in `image` and crops a `height` by `width`</span>
+<a name="line-6336"></a><span class='hs-comment'>-- rectangle from that location.  The random location is picked so the cropped</span>
+<a name="line-6337"></a><span class='hs-comment'>-- area will fit inside the original image.</span>
+<a name="line-6338"></a><span class='hs-definition'>randomCrop</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-6339"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-6340"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-6341"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-6342"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-6343"></a>                                                      <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-6344"></a>              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __image__: 3-D of shape `[height, width, channels]`.</span>
+<a name="line-6345"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __size__: 1-D of length 2 containing: `crop_height`, `crop_width`..</span>
+<a name="line-6346"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 3-D of shape `[crop_height, crop_width, channels].`</span>
+<a name="line-6347"></a><span class='hs-definition'>randomCrop</span> <span class='hs-varid'>image</span> <span class='hs-varid'>size</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6348"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RandomCrop"</span>
+<a name="line-6349"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-6350"></a>        <span class='hs-varid'>image</span> <span class='hs-varid'>size</span>
+<a name="line-6351"></a><span class='hs-comment'>{-
+<a name="line-6352"></a>attr {
+<a name="line-6353"></a>  allowed_values {
+<a name="line-6354"></a>    list {
+<a name="line-6355"></a>      type: DT_UINT8
+<a name="line-6356"></a>      type: DT_INT8
+<a name="line-6357"></a>      type: DT_INT16
+<a name="line-6358"></a>      type: DT_INT32
+<a name="line-6359"></a>      type: DT_INT64
+<a name="line-6360"></a>      type: DT_FLOAT
+<a name="line-6361"></a>      type: DT_DOUBLE
+<a name="line-6362"></a>    }
+<a name="line-6363"></a>  }
+<a name="line-6364"></a>  name: "T"
+<a name="line-6365"></a>  type: "type"
+<a name="line-6366"></a>}
+<a name="line-6367"></a>attr {
+<a name="line-6368"></a>  default_value { i: 0 }
+<a name="line-6369"></a>  description: "If either seed or seed2 are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-6370"></a>  name: "seed"
+<a name="line-6371"></a>  type: "int"
+<a name="line-6372"></a>}
+<a name="line-6373"></a>attr {
+<a name="line-6374"></a>  default_value { i: 0 }
+<a name="line-6375"></a>  description: "An second seed to avoid seed collision."
+<a name="line-6376"></a>  name: "seed2"
+<a name="line-6377"></a>  type: "int"
+<a name="line-6378"></a>}
+<a name="line-6379"></a>input_arg {
+<a name="line-6380"></a>  description: "3-D of shape `[height, width, channels]`."
+<a name="line-6381"></a>  name: "image"
+<a name="line-6382"></a>  type_attr: "T"
+<a name="line-6383"></a>}
+<a name="line-6384"></a>input_arg {
+<a name="line-6385"></a>  description: "1-D of length 2 containing: `crop_height`, `crop_width`.."
+<a name="line-6386"></a>  name: "size"
+<a name="line-6387"></a>  type: DT_INT64
+<a name="line-6388"></a>}
+<a name="line-6389"></a>output_arg {
+<a name="line-6390"></a>  description: "3-D of shape `[crop_height, crop_width, channels].`"
+<a name="line-6391"></a>  name: "output"
+<a name="line-6392"></a>  type_attr: "T"
+<a name="line-6393"></a>}
+<a name="line-6394"></a>-}</span>
+<a name="line-6395"></a>
+<a name="line-6396"></a><span class='hs-comment'>-- | Cast x of type SrcT to y of DstT.</span>
+<a name="line-6397"></a><span class='hs-comment'>--</span>
+<a name="line-6398"></a><span class='hs-comment'>-- _HostCast requires its input and produces its output in host memory.</span>
+<a name="line-6399"></a><span class='hs-sel'>_HostCast</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>dstT</span> <span class='hs-varid'>srcT</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dstT</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>srcT</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-6400"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>srcT</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-6401"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dstT</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-6402"></a><span class='hs-sel'>_HostCast</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6403"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"_HostCast"</span>
+<a name="line-6404"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"DstT"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dstT</span><span class='hs-layout'>)</span>
+<a name="line-6405"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"SrcT"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>srcT</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-6406"></a>        <span class='hs-varid'>x</span>
+<a name="line-6407"></a><span class='hs-comment'>{-
+<a name="line-6408"></a>attr { name: "SrcT" type: "type" }
+<a name="line-6409"></a>attr { name: "DstT" type: "type" }
+<a name="line-6410"></a>input_arg { name: "x" type_attr: "SrcT" }
+<a name="line-6411"></a>output_arg { name: "y" type_attr: "DstT" }
+<a name="line-6412"></a>-}</span>
+<a name="line-6413"></a>
+<a name="line-6414"></a><a name="queueClose"></a><span class='hs-comment'>-- | Closes the given queue.</span>
+<a name="line-6415"></a><span class='hs-comment'>--</span>
+<a name="line-6416"></a><span class='hs-comment'>-- This operation signals that no more elements will be enqueued in the</span>
+<a name="line-6417"></a><span class='hs-comment'>-- given queue. Subsequent Enqueue(Many) operations will fail.</span>
+<a name="line-6418"></a><span class='hs-comment'>-- Subsequent Dequeue(Many) operations will continue to succeed if</span>
+<a name="line-6419"></a><span class='hs-comment'>-- sufficient elements remain in the queue. Subsequent Dequeue(Many)</span>
+<a name="line-6420"></a><span class='hs-comment'>-- operations that would block will fail immediately.</span>
+<a name="line-6421"></a><span class='hs-definition'>queueClose</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a queue.</span>
+<a name="line-6422"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-6423"></a><span class='hs-definition'>queueClose</span> <span class='hs-varid'>handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6424"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"QueueClose"</span><span class='hs-layout'>)</span>
+<a name="line-6425"></a>        <span class='hs-varid'>handle</span>
+<a name="line-6426"></a><span class='hs-comment'>{-
+<a name="line-6427"></a>attr {
+<a name="line-6428"></a>  default_value { b: false }
+<a name="line-6429"></a>  description: "If true, all pending enqueue requests that are\nblocked on the given queue will be cancelled."
+<a name="line-6430"></a>  name: "cancel_pending_enqueues"
+<a name="line-6431"></a>  type: "bool"
+<a name="line-6432"></a>}
+<a name="line-6433"></a>input_arg {
+<a name="line-6434"></a>  description: "The handle to a queue."
+<a name="line-6435"></a>  is_ref: true
+<a name="line-6436"></a>  name: "handle"
+<a name="line-6437"></a>  type: DT_STRING
+<a name="line-6438"></a>}
+<a name="line-6439"></a>-}</span>
+<a name="line-6440"></a>
+<a name="line-6441"></a><a name="slice"></a><span class='hs-comment'>-- | Return a slice from 'input'.</span>
+<a name="line-6442"></a><span class='hs-comment'>--</span>
+<a name="line-6443"></a><span class='hs-comment'>-- The output tensor is a tensor with dimensions described by 'size'</span>
+<a name="line-6444"></a><span class='hs-comment'>-- whose values are extracted from 'input' starting at the offsets in</span>
+<a name="line-6445"></a><span class='hs-comment'>-- 'begin'.</span>
+<a name="line-6446"></a><span class='hs-comment'>-- </span>
+<a name="line-6447"></a><span class='hs-comment'>-- *Requirements*:</span>
+<a name="line-6448"></a><span class='hs-comment'>--   0 &lt;= begin[i] &lt;= begin[i] + size[i] &lt;= Di  for i in [0, n)</span>
+<a name="line-6449"></a><span class='hs-definition'>slice</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>index</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>index</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-6450"></a>                                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>index</span><span class='hs-layout'>,</span>
+<a name="line-6451"></a>                                    <span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-6452"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>index</span> <span class='hs-comment'>-- ^ __begin__: begin[i] specifies the offset into the 'i'th dimension of</span>
+<a name="line-6453"></a>                            <span class='hs-comment'>-- 'input' to slice from.</span>
+<a name="line-6454"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>index</span> <span class='hs-comment'>-- ^ __size__: size[i] specifies the number of elements of the 'i'th dimension</span>
+<a name="line-6455"></a>                            <span class='hs-comment'>-- of 'input' to slice. If size[i] is -1, all remaining elements in dimension</span>
+<a name="line-6456"></a>                            <span class='hs-comment'>-- i are included in the slice (i.e. this is equivalent to setting</span>
+<a name="line-6457"></a>                            <span class='hs-comment'>-- size[i] = input.dim_size(i) - begin[i]).</span>
+<a name="line-6458"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-6459"></a><span class='hs-definition'>slice</span> <span class='hs-varid'>input</span> <span class='hs-varid'>begin</span> <span class='hs-varid'>size</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6460"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Slice"</span>
+<a name="line-6461"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Index"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>index</span><span class='hs-layout'>)</span>
+<a name="line-6462"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-6463"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>begin</span> <span class='hs-varid'>size</span>
+<a name="line-6464"></a><span class='hs-comment'>{-
+<a name="line-6465"></a>attr { name: "T" type: "type" }
+<a name="line-6466"></a>attr {
+<a name="line-6467"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-6468"></a>  name: "Index"
+<a name="line-6469"></a>  type: "type"
+<a name="line-6470"></a>}
+<a name="line-6471"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-6472"></a>input_arg {
+<a name="line-6473"></a>  description: "begin[i] specifies the offset into the \'i\'th dimension of\n\'input\' to slice from."
+<a name="line-6474"></a>  name: "begin"
+<a name="line-6475"></a>  type_attr: "Index"
+<a name="line-6476"></a>}
+<a name="line-6477"></a>input_arg {
+<a name="line-6478"></a>  description: "size[i] specifies the number of elements of the \'i\'th dimension\nof \'input\' to slice. If size[i] is -1, all remaining elements in dimension\ni are included in the slice (i.e. this is equivalent to setting\nsize[i] = input.dim_size(i) - begin[i])."
+<a name="line-6479"></a>  name: "size"
+<a name="line-6480"></a>  type_attr: "Index"
+<a name="line-6481"></a>}
+<a name="line-6482"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-6483"></a>-}</span>
+<a name="line-6484"></a>
+<a name="line-6485"></a><a name="stridedSliceGrad"></a><span class='hs-comment'>-- | Returns the gradient of `StridedSlice`.</span>
+<a name="line-6486"></a><span class='hs-comment'>--</span>
+<a name="line-6487"></a><span class='hs-comment'>-- Since `StridedSlice` cuts out pieces of its `input` which is size</span>
+<a name="line-6488"></a><span class='hs-comment'>-- `shape`, its gradient will have the same shape (which is passed here</span>
+<a name="line-6489"></a><span class='hs-comment'>-- as `shape`). The gradient will be zero in any element that the slice</span>
+<a name="line-6490"></a><span class='hs-comment'>-- does not select.</span>
+<a name="line-6491"></a><span class='hs-comment'>-- </span>
+<a name="line-6492"></a><span class='hs-comment'>-- Arguments are the same as StridedSliceGrad with the exception that</span>
+<a name="line-6493"></a><span class='hs-comment'>-- `dy` is the input gradient to be propagated and `shape` is the</span>
+<a name="line-6494"></a><span class='hs-comment'>-- shape of `StridedSlice`'s `input`.</span>
+<a name="line-6495"></a><span class='hs-definition'>stridedSliceGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>index</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>index</span><span class='hs-layout'>,</span>
+<a name="line-6496"></a>                                                     <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-6497"></a>                                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>index</span><span class='hs-layout'>,</span>
+<a name="line-6498"></a>                                                     <span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-6499"></a>                    <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>index</span> <span class='hs-comment'>-- ^ __shape__</span>
+<a name="line-6500"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>index</span> <span class='hs-comment'>-- ^ __begin__</span>
+<a name="line-6501"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>index</span> <span class='hs-comment'>-- ^ __end__</span>
+<a name="line-6502"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>index</span> <span class='hs-comment'>-- ^ __strides__</span>
+<a name="line-6503"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __dy__</span>
+<a name="line-6504"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-6505"></a><span class='hs-definition'>stridedSliceGrad</span> <span class='hs-varid'>shape</span> <span class='hs-varid'>begin</span> <span class='hs-varid'>end</span> <span class='hs-varid'>strides</span> <span class='hs-varid'>dy</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6506"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"StridedSliceGrad"</span>
+<a name="line-6507"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Index"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>index</span><span class='hs-layout'>)</span>
+<a name="line-6508"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-6509"></a>        <span class='hs-varid'>shape</span> <span class='hs-varid'>begin</span> <span class='hs-varid'>end</span> <span class='hs-varid'>strides</span> <span class='hs-varid'>dy</span>
+<a name="line-6510"></a><span class='hs-comment'>{-
+<a name="line-6511"></a>attr { name: "T" type: "type" }
+<a name="line-6512"></a>attr {
+<a name="line-6513"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-6514"></a>  name: "Index"
+<a name="line-6515"></a>  type: "type"
+<a name="line-6516"></a>}
+<a name="line-6517"></a>attr { default_value { i: 0 } name: "begin_mask" type: "int" }
+<a name="line-6518"></a>attr { default_value { i: 0 } name: "end_mask" type: "int" }
+<a name="line-6519"></a>attr { default_value { i: 0 } name: "ellipsis_mask" type: "int" }
+<a name="line-6520"></a>attr { default_value { i: 0 } name: "new_axis_mask" type: "int" }
+<a name="line-6521"></a>attr {
+<a name="line-6522"></a>  default_value { i: 0 } name: "shrink_axis_mask" type: "int"
+<a name="line-6523"></a>}
+<a name="line-6524"></a>input_arg { name: "shape" type_attr: "Index" }
+<a name="line-6525"></a>input_arg { name: "begin" type_attr: "Index" }
+<a name="line-6526"></a>input_arg { name: "end" type_attr: "Index" }
+<a name="line-6527"></a>input_arg { name: "strides" type_attr: "Index" }
+<a name="line-6528"></a>input_arg { name: "dy" type_attr: "T" }
+<a name="line-6529"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-6530"></a>-}</span>
+<a name="line-6531"></a>
+<a name="line-6532"></a><a name="sparseTensorDenseAdd"></a><span class='hs-comment'>-- | Adds up a `SparseTensor` and a dense `Tensor`, producing a dense `Tensor`.</span>
+<a name="line-6533"></a><span class='hs-comment'>--</span>
+<a name="line-6534"></a><span class='hs-comment'>-- This Op does not require `a_indices` be sorted in standard lexicographic order.</span>
+<a name="line-6535"></a><span class='hs-definition'>sparseTensorDenseAdd</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-6536"></a>                                                         <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-6537"></a>                                                                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-6538"></a>                                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-6539"></a>                                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-6540"></a>                                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-6541"></a>                                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-6542"></a>                                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-6543"></a>                                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-6544"></a>                                                                 <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-6545"></a>                                                                 <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-6546"></a>                                                         <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-6547"></a>                                                         <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-6548"></a>                                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-6549"></a>                        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __a_indices__: 2-D.  The `indices` of the `SparseTensor`, with shape `[nnz, ndims]`.</span>
+<a name="line-6550"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __a_values__: 1-D.  The `values` of the `SparseTensor`, with shape `[nnz]`.</span>
+<a name="line-6551"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __a_shape__: 1-D.  The `shape` of the `SparseTensor`, with shape `[ndims]`.</span>
+<a name="line-6552"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __b__: `ndims`-D Tensor.  With shape `a_shape`.</span>
+<a name="line-6553"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-6554"></a><span class='hs-definition'>sparseTensorDenseAdd</span> <span class='hs-varid'>a_indices</span> <span class='hs-varid'>a_values</span> <span class='hs-varid'>a_shape</span> <span class='hs-varid'>b</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6555"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseTensorDenseAdd"</span>
+<a name="line-6556"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-6557"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-6558"></a>        <span class='hs-varid'>a_indices</span> <span class='hs-varid'>a_values</span> <span class='hs-varid'>a_shape</span> <span class='hs-varid'>b</span>
+<a name="line-6559"></a><span class='hs-comment'>{-
+<a name="line-6560"></a>attr {
+<a name="line-6561"></a>  allowed_values {
+<a name="line-6562"></a>    list {
+<a name="line-6563"></a>      type: DT_FLOAT
+<a name="line-6564"></a>      type: DT_DOUBLE
+<a name="line-6565"></a>      type: DT_INT64
+<a name="line-6566"></a>      type: DT_INT32
+<a name="line-6567"></a>      type: DT_UINT8
+<a name="line-6568"></a>      type: DT_UINT16
+<a name="line-6569"></a>      type: DT_INT16
+<a name="line-6570"></a>      type: DT_INT8
+<a name="line-6571"></a>      type: DT_COMPLEX64
+<a name="line-6572"></a>      type: DT_COMPLEX128
+<a name="line-6573"></a>      type: DT_QINT8
+<a name="line-6574"></a>      type: DT_QUINT8
+<a name="line-6575"></a>      type: DT_QINT32
+<a name="line-6576"></a>      type: DT_HALF
+<a name="line-6577"></a>    }
+<a name="line-6578"></a>  }
+<a name="line-6579"></a>  name: "T"
+<a name="line-6580"></a>  type: "type"
+<a name="line-6581"></a>}
+<a name="line-6582"></a>attr {
+<a name="line-6583"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-6584"></a>  name: "Tindices"
+<a name="line-6585"></a>  type: "type"
+<a name="line-6586"></a>}
+<a name="line-6587"></a>input_arg {
+<a name="line-6588"></a>  description: "2-D.  The `indices` of the `SparseTensor`, with shape `[nnz, ndims]`."
+<a name="line-6589"></a>  name: "a_indices"
+<a name="line-6590"></a>  type_attr: "Tindices"
+<a name="line-6591"></a>}
+<a name="line-6592"></a>input_arg {
+<a name="line-6593"></a>  description: "1-D.  The `values` of the `SparseTensor`, with shape `[nnz]`."
+<a name="line-6594"></a>  name: "a_values"
+<a name="line-6595"></a>  type_attr: "T"
+<a name="line-6596"></a>}
+<a name="line-6597"></a>input_arg {
+<a name="line-6598"></a>  description: "1-D.  The `shape` of the `SparseTensor`, with shape `[ndims]`."
+<a name="line-6599"></a>  name: "a_shape"
+<a name="line-6600"></a>  type_attr: "Tindices"
+<a name="line-6601"></a>}
+<a name="line-6602"></a>input_arg {
+<a name="line-6603"></a>  description: "`ndims`-D Tensor.  With shape `a_shape`."
+<a name="line-6604"></a>  name: "b"
+<a name="line-6605"></a>  type_attr: "T"
+<a name="line-6606"></a>}
+<a name="line-6607"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-6608"></a>-}</span>
+<a name="line-6609"></a>
+<a name="line-6610"></a><a name="size"></a><span class='hs-comment'>-- | Returns the size of a tensor.</span>
+<a name="line-6611"></a><span class='hs-comment'>--</span>
+<a name="line-6612"></a><span class='hs-comment'>-- This operation returns an integer representing the number of elements in</span>
+<a name="line-6613"></a><span class='hs-comment'>-- `input`.</span>
+<a name="line-6614"></a><span class='hs-comment'>-- </span>
+<a name="line-6615"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-6616"></a><span class='hs-comment'>-- </span>
+<a name="line-6617"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-6618"></a><span class='hs-comment'>-- # 't' is [[[1, 1,, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]]</span>
+<a name="line-6619"></a><span class='hs-comment'>-- size(t) ==&gt; 12</span>
+<a name="line-6620"></a><span class='hs-comment'>-- ```</span>
+<a name="line-6621"></a><span class='hs-definition'>size</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varid'>out_type</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>,</span>
+<a name="line-6622"></a>                                <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-6623"></a>                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-6624"></a>        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-6625"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>out_type</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-6626"></a><span class='hs-definition'>size</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6627"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Size"</span>
+<a name="line-6628"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-6629"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"out_type"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-6630"></a>        <span class='hs-varid'>input</span>
+<a name="line-6631"></a><span class='hs-comment'>{-
+<a name="line-6632"></a>attr { name: "T" type: "type" }
+<a name="line-6633"></a>attr {
+<a name="line-6634"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-6635"></a>  default_value { type: DT_INT32 }
+<a name="line-6636"></a>  name: "out_type"
+<a name="line-6637"></a>  type: "type"
+<a name="line-6638"></a>}
+<a name="line-6639"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-6640"></a>output_arg { name: "output" type_attr: "out_type" }
+<a name="line-6641"></a>-}</span>
+<a name="line-6642"></a>
+<a name="line-6643"></a><a name="barrier"></a><span class='hs-comment'>-- | Defines a barrier that persists across different graph executions.</span>
+<a name="line-6644"></a><span class='hs-comment'>--</span>
+<a name="line-6645"></a><span class='hs-comment'>-- A barrier represents a key-value map, where each key is a string, and</span>
+<a name="line-6646"></a><span class='hs-comment'>-- each value is a tuple of tensors.</span>
+<a name="line-6647"></a><span class='hs-comment'>-- </span>
+<a name="line-6648"></a><span class='hs-comment'>-- At runtime, the barrier contains 'complete' and 'incomplete'</span>
+<a name="line-6649"></a><span class='hs-comment'>-- elements. A complete element has defined tensors for all components of</span>
+<a name="line-6650"></a><span class='hs-comment'>-- its value tuple, and may be accessed using BarrierTakeMany. An</span>
+<a name="line-6651"></a><span class='hs-comment'>-- incomplete element has some undefined components in its value tuple,</span>
+<a name="line-6652"></a><span class='hs-comment'>-- and may be updated using BarrierInsertMany.</span>
+<a name="line-6653"></a><span class='hs-definition'>barrier</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to the barrier.</span>
+<a name="line-6654"></a><span class='hs-definition'>barrier</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6655"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Barrier"</span><span class='hs-layout'>)</span>
+<a name="line-6656"></a>        
+<a name="line-6657"></a><span class='hs-comment'>{-
+<a name="line-6658"></a>attr {
+<a name="line-6659"></a>  description: "The type of each component in a value."
+<a name="line-6660"></a>  has_minimum: true
+<a name="line-6661"></a>  minimum: 1
+<a name="line-6662"></a>  name: "component_types"
+<a name="line-6663"></a>  type: "list(type)"
+<a name="line-6664"></a>}
+<a name="line-6665"></a>attr {
+<a name="line-6666"></a>  default_value { list { } }
+<a name="line-6667"></a>  description: "The shape of each component in a value. Each shape must be 1 in the\nfirst dimension. The length of this attr must be the same as the length of\ncomponent_types."
+<a name="line-6668"></a>  has_minimum: true
+<a name="line-6669"></a>  name: "shapes"
+<a name="line-6670"></a>  type: "list(shape)"
+<a name="line-6671"></a>}
+<a name="line-6672"></a>attr {
+<a name="line-6673"></a>  default_value { i: -1 }
+<a name="line-6674"></a>  description: "The capacity of the barrier.  The default capacity is MAX_INT32,\nwhich is the largest capacity of the underlying queue."
+<a name="line-6675"></a>  name: "capacity"
+<a name="line-6676"></a>  type: "int"
+<a name="line-6677"></a>}
+<a name="line-6678"></a>attr {
+<a name="line-6679"></a>  default_value { s: "" }
+<a name="line-6680"></a>  description: "If non-empty, this barrier is placed in the given container.\nOtherwise, a default container is used."
+<a name="line-6681"></a>  name: "container"
+<a name="line-6682"></a>  type: "string"
+<a name="line-6683"></a>}
+<a name="line-6684"></a>attr {
+<a name="line-6685"></a>  default_value { s: "" }
+<a name="line-6686"></a>  description: "If non-empty, this barrier will be shared under the given name\nacross multiple sessions."
+<a name="line-6687"></a>  name: "shared_name"
+<a name="line-6688"></a>  type: "string"
+<a name="line-6689"></a>}
+<a name="line-6690"></a>output_arg {
+<a name="line-6691"></a>  description: "The handle to the barrier."
+<a name="line-6692"></a>  is_ref: true
+<a name="line-6693"></a>  name: "handle"
+<a name="line-6694"></a>  type: DT_STRING
+<a name="line-6695"></a>}
+<a name="line-6696"></a>-}</span>
+<a name="line-6697"></a>
+<a name="line-6698"></a><span class='hs-comment'>-- | Computes the log of the absolute value of `Gamma(x)` element-wise.</span>
+<a name="line-6699"></a>
+<a name="line-6700"></a><a name="lgamma"></a><span class='hs-definition'>lgamma</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-6701"></a>                                               <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-6702"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-6703"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-6704"></a><span class='hs-definition'>lgamma</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6705"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Lgamma"</span>
+<a name="line-6706"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-6707"></a>        <span class='hs-varid'>x</span>
+<a name="line-6708"></a><span class='hs-comment'>{-
+<a name="line-6709"></a>attr {
+<a name="line-6710"></a>  allowed_values {
+<a name="line-6711"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-6712"></a>  }
+<a name="line-6713"></a>  name: "T"
+<a name="line-6714"></a>  type: "type"
+<a name="line-6715"></a>}
+<a name="line-6716"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-6717"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-6718"></a>-}</span>
+<a name="line-6719"></a>
+<a name="line-6720"></a><a name="decodeJpeg"></a><span class='hs-comment'>-- | Decode a JPEG-encoded image to a uint8 tensor.</span>
+<a name="line-6721"></a><span class='hs-comment'>--</span>
+<a name="line-6722"></a><span class='hs-comment'>-- The attr `channels` indicates the desired number of color channels for the</span>
+<a name="line-6723"></a><span class='hs-comment'>-- decoded image.</span>
+<a name="line-6724"></a><span class='hs-comment'>-- </span>
+<a name="line-6725"></a><span class='hs-comment'>-- Accepted values are:</span>
+<a name="line-6726"></a><span class='hs-comment'>-- </span>
+<a name="line-6727"></a><span class='hs-comment'>-- *   0: Use the number of channels in the JPEG-encoded image.</span>
+<a name="line-6728"></a><span class='hs-comment'>-- *   1: output a grayscale image.</span>
+<a name="line-6729"></a><span class='hs-comment'>-- *   3: output an RGB image.</span>
+<a name="line-6730"></a><span class='hs-comment'>-- </span>
+<a name="line-6731"></a><span class='hs-comment'>-- If needed, the JPEG-encoded image is transformed to match the requested number</span>
+<a name="line-6732"></a><span class='hs-comment'>-- of color channels.</span>
+<a name="line-6733"></a><span class='hs-comment'>-- </span>
+<a name="line-6734"></a><span class='hs-comment'>-- The attr `ratio` allows downscaling the image by an integer factor during</span>
+<a name="line-6735"></a><span class='hs-comment'>-- decoding.  Allowed values are: 1, 2, 4, and 8.  This is much faster than</span>
+<a name="line-6736"></a><span class='hs-comment'>-- downscaling the image later.</span>
+<a name="line-6737"></a><span class='hs-definition'>decodeJpeg</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __contents__: 0-D.  The JPEG-encoded image.</span>
+<a name="line-6738"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span> <span class='hs-comment'>-- ^ __image__: 3-D with shape `[height, width, channels]`..</span>
+<a name="line-6739"></a><span class='hs-definition'>decodeJpeg</span> <span class='hs-varid'>contents</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6740"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DecodeJpeg"</span><span class='hs-layout'>)</span>
+<a name="line-6741"></a>        <span class='hs-varid'>contents</span>
+<a name="line-6742"></a><span class='hs-comment'>{-
+<a name="line-6743"></a>attr {
+<a name="line-6744"></a>  default_value { i: 0 }
+<a name="line-6745"></a>  description: "Number of color channels for the decoded image."
+<a name="line-6746"></a>  name: "channels"
+<a name="line-6747"></a>  type: "int"
+<a name="line-6748"></a>}
+<a name="line-6749"></a>attr {
+<a name="line-6750"></a>  default_value { i: 1 }
+<a name="line-6751"></a>  description: "Downscaling ratio."
+<a name="line-6752"></a>  name: "ratio"
+<a name="line-6753"></a>  type: "int"
+<a name="line-6754"></a>}
+<a name="line-6755"></a>attr {
+<a name="line-6756"></a>  default_value { b: true }
+<a name="line-6757"></a>  description: "If true use a slower but nicer upscaling of the\nchroma planes (yuv420/422 only)."
+<a name="line-6758"></a>  name: "fancy_upscaling"
+<a name="line-6759"></a>  type: "bool"
+<a name="line-6760"></a>}
+<a name="line-6761"></a>attr {
+<a name="line-6762"></a>  default_value { b: false }
+<a name="line-6763"></a>  description: "If true try to recover an image from truncated input."
+<a name="line-6764"></a>  name: "try_recover_truncated"
+<a name="line-6765"></a>  type: "bool"
+<a name="line-6766"></a>}
+<a name="line-6767"></a>attr {
+<a name="line-6768"></a>  default_value { f: 1.0 }
+<a name="line-6769"></a>  description: "The minimum required fraction of lines before a truncated\ninput is accepted."
+<a name="line-6770"></a>  name: "acceptable_fraction"
+<a name="line-6771"></a>  type: "float"
+<a name="line-6772"></a>}
+<a name="line-6773"></a>input_arg {
+<a name="line-6774"></a>  description: "0-D.  The JPEG-encoded image."
+<a name="line-6775"></a>  name: "contents"
+<a name="line-6776"></a>  type: DT_STRING
+<a name="line-6777"></a>}
+<a name="line-6778"></a>output_arg {
+<a name="line-6779"></a>  description: "3-D with shape `[height, width, channels]`.."
+<a name="line-6780"></a>  name: "image"
+<a name="line-6781"></a>  type: DT_UINT8
+<a name="line-6782"></a>}
+<a name="line-6783"></a>-}</span>
+<a name="line-6784"></a>
+<a name="line-6785"></a><a name="shapeN"></a><span class='hs-comment'>-- | Returns shape of tensors.</span>
+<a name="line-6786"></a><span class='hs-comment'>--</span>
+<a name="line-6787"></a><span class='hs-comment'>-- This operation returns N 1-D integer tensors representing shape of `input[i]s`.</span>
+<a name="line-6788"></a><span class='hs-definition'>shapeN</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varid'>out_type</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>,</span>
+<a name="line-6789"></a>                                  <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-6790"></a>                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-6791"></a>          <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-6792"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>out_type</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-6793"></a><span class='hs-definition'>shapeN</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"N"</span><span class='hs-layout'>,</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"input"</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>input</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6794"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ShapeN"</span>
+<a name="line-6795"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-6796"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"out_type"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>)</span>
+<a name="line-6797"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"N"</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>input</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-6798"></a>        <span class='hs-varid'>input</span>
+<a name="line-6799"></a><span class='hs-comment'>{-
+<a name="line-6800"></a>attr { has_minimum: true minimum: 1 name: "N" type: "int" }
+<a name="line-6801"></a>attr { name: "T" type: "type" }
+<a name="line-6802"></a>attr {
+<a name="line-6803"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-6804"></a>  default_value { type: DT_INT32 }
+<a name="line-6805"></a>  name: "out_type"
+<a name="line-6806"></a>  type: "type"
+<a name="line-6807"></a>}
+<a name="line-6808"></a>input_arg { name: "input" number_attr: "N" type_attr: "T" }
+<a name="line-6809"></a>output_arg {
+<a name="line-6810"></a>  name: "output" number_attr: "N" type_attr: "out_type"
+<a name="line-6811"></a>}
+<a name="line-6812"></a>-}</span>
+<a name="line-6813"></a>
+<a name="line-6814"></a><a name="uniformCandidateSampler"></a><span class='hs-comment'>-- | Generates labels for candidate sampling with a uniform distribution.</span>
+<a name="line-6815"></a><span class='hs-comment'>--</span>
+<a name="line-6816"></a><span class='hs-comment'>-- See explanations of candidate sampling and the data formats at</span>
+<a name="line-6817"></a><span class='hs-comment'>-- go/candidate-sampling.</span>
+<a name="line-6818"></a><span class='hs-comment'>-- </span>
+<a name="line-6819"></a><span class='hs-comment'>-- For each batch, this op picks a single set of sampled candidate labels.</span>
+<a name="line-6820"></a><span class='hs-comment'>-- </span>
+<a name="line-6821"></a><span class='hs-comment'>-- The advantages of sampling candidates per-batch are simplicity and the</span>
+<a name="line-6822"></a><span class='hs-comment'>-- possibility of efficient dense matrix multiplication. The disadvantage is that</span>
+<a name="line-6823"></a><span class='hs-comment'>-- the sampled candidates must be chosen independently of the context and of the</span>
+<a name="line-6824"></a><span class='hs-comment'>-- true labels.</span>
+<a name="line-6825"></a><span class='hs-definition'>uniformCandidateSampler</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_sampled__: Number of candidates to randomly sample per batch.</span>
+<a name="line-6826"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_true__: Number of true labels per context.</span>
+<a name="line-6827"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __range_max__: The sampler will sample integers from the interval [0, range_max).</span>
+<a name="line-6828"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __unique__: If unique is true, we sample with rejection, so that all sampled</span>
+<a name="line-6829"></a>                                   <span class='hs-comment'>-- candidates in a batch are unique. This requires some approximation to</span>
+<a name="line-6830"></a>                                   <span class='hs-comment'>-- estimate the post-rejection sampling probabilities.</span>
+<a name="line-6831"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __true_classes__: A batch_size * num_true matrix, in which each row contains the</span>
+<a name="line-6832"></a>                                                       <span class='hs-comment'>-- IDs of the num_true target_classes in the corresponding original label.</span>
+<a name="line-6833"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>,</span>
+<a name="line-6834"></a>                               <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span>
+<a name="line-6835"></a>                           <span class='hs-comment'>-- ^ (__sampled_candidates__, __true_expected_count__, __sampled_expected_count__)</span>
+<a name="line-6836"></a>                           <span class='hs-comment'>--</span>
+<a name="line-6837"></a>                           <span class='hs-comment'>-- * __sampled_candidates__: A vector of length num_sampled, in which each element is</span>
+<a name="line-6838"></a>                           <span class='hs-comment'>-- the ID of a sampled candidate.</span>
+<a name="line-6839"></a>                           <span class='hs-comment'>--</span>
+<a name="line-6840"></a>                           <span class='hs-comment'>-- * __true_expected_count__: A batch_size * num_true matrix, representing</span>
+<a name="line-6841"></a>                           <span class='hs-comment'>-- the number of times each candidate is expected to occur in a batch</span>
+<a name="line-6842"></a>                           <span class='hs-comment'>-- of sampled candidates. If unique=true, then this is a probability.</span>
+<a name="line-6843"></a>                           <span class='hs-comment'>--</span>
+<a name="line-6844"></a>                           <span class='hs-comment'>-- * __sampled_expected_count__: A vector of length num_sampled, for each sampled</span>
+<a name="line-6845"></a>                           <span class='hs-comment'>-- candidate representing the number of times the candidate is expected</span>
+<a name="line-6846"></a>                           <span class='hs-comment'>-- to occur in a batch of sampled candidates.  If unique=true, then this is a</span>
+<a name="line-6847"></a>                           <span class='hs-comment'>-- probability.</span>
+<a name="line-6848"></a><span class='hs-definition'>uniformCandidateSampler</span> <span class='hs-varid'>num_sampled</span> <span class='hs-varid'>num_true</span> <span class='hs-varid'>range_max</span> <span class='hs-varid'>unique</span>
+<a name="line-6849"></a>                        <span class='hs-varid'>true_classes</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6850"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"UniformCandidateSampler"</span>
+<a name="line-6851"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_sampled"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_sampled</span>
+<a name="line-6852"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_true"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_true</span>
+<a name="line-6853"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"range_max"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>range_max</span>
+<a name="line-6854"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"unique"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>unique</span><span class='hs-layout'>)</span>
+<a name="line-6855"></a>        <span class='hs-varid'>true_classes</span>
+<a name="line-6856"></a><span class='hs-comment'>{-
+<a name="line-6857"></a>attr {
+<a name="line-6858"></a>  description: "Number of true labels per context."
+<a name="line-6859"></a>  has_minimum: true
+<a name="line-6860"></a>  minimum: 1
+<a name="line-6861"></a>  name: "num_true"
+<a name="line-6862"></a>  type: "int"
+<a name="line-6863"></a>}
+<a name="line-6864"></a>attr {
+<a name="line-6865"></a>  description: "Number of candidates to randomly sample per batch."
+<a name="line-6866"></a>  has_minimum: true
+<a name="line-6867"></a>  minimum: 1
+<a name="line-6868"></a>  name: "num_sampled"
+<a name="line-6869"></a>  type: "int"
+<a name="line-6870"></a>}
+<a name="line-6871"></a>attr {
+<a name="line-6872"></a>  description: "If unique is true, we sample with rejection, so that all sampled\ncandidates in a batch are unique. This requires some approximation to\nestimate the post-rejection sampling probabilities."
+<a name="line-6873"></a>  name: "unique"
+<a name="line-6874"></a>  type: "bool"
+<a name="line-6875"></a>}
+<a name="line-6876"></a>attr {
+<a name="line-6877"></a>  description: "The sampler will sample integers from the interval [0, range_max)."
+<a name="line-6878"></a>  has_minimum: true
+<a name="line-6879"></a>  minimum: 1
+<a name="line-6880"></a>  name: "range_max"
+<a name="line-6881"></a>  type: "int"
+<a name="line-6882"></a>}
+<a name="line-6883"></a>attr {
+<a name="line-6884"></a>  default_value { i: 0 }
+<a name="line-6885"></a>  description: "If either seed or seed2 are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-6886"></a>  name: "seed"
+<a name="line-6887"></a>  type: "int"
+<a name="line-6888"></a>}
+<a name="line-6889"></a>attr {
+<a name="line-6890"></a>  default_value { i: 0 }
+<a name="line-6891"></a>  description: "An second seed to avoid seed collision."
+<a name="line-6892"></a>  name: "seed2"
+<a name="line-6893"></a>  type: "int"
+<a name="line-6894"></a>}
+<a name="line-6895"></a>input_arg {
+<a name="line-6896"></a>  description: "A batch_size * num_true matrix, in which each row contains the\nIDs of the num_true target_classes in the corresponding original label."
+<a name="line-6897"></a>  name: "true_classes"
+<a name="line-6898"></a>  type: DT_INT64
+<a name="line-6899"></a>}
+<a name="line-6900"></a>output_arg {
+<a name="line-6901"></a>  description: "A vector of length num_sampled, in which each element is\nthe ID of a sampled candidate."
+<a name="line-6902"></a>  name: "sampled_candidates"
+<a name="line-6903"></a>  type: DT_INT64
+<a name="line-6904"></a>}
+<a name="line-6905"></a>output_arg {
+<a name="line-6906"></a>  description: "A batch_size * num_true matrix, representing\nthe number of times each candidate is expected to occur in a batch\nof sampled candidates. If unique=true, then this is a probability."
+<a name="line-6907"></a>  name: "true_expected_count"
+<a name="line-6908"></a>  type: DT_FLOAT
+<a name="line-6909"></a>}
+<a name="line-6910"></a>output_arg {
+<a name="line-6911"></a>  description: "A vector of length num_sampled, for each sampled\ncandidate representing the number of times the candidate is expected\nto occur in a batch of sampled candidates.  If unique=true, then this is a\nprobability."
+<a name="line-6912"></a>  name: "sampled_expected_count"
+<a name="line-6913"></a>  type: DT_FLOAT
+<a name="line-6914"></a>}
+<a name="line-6915"></a>-}</span>
+<a name="line-6916"></a>
+<a name="line-6917"></a><a name="unique"></a><span class='hs-comment'>-- | Finds unique elements in a 1-D tensor.</span>
+<a name="line-6918"></a><span class='hs-comment'>--</span>
+<a name="line-6919"></a><span class='hs-comment'>-- This operation returns a tensor `y` containing all of the unique elements of `x`</span>
+<a name="line-6920"></a><span class='hs-comment'>-- sorted in the same order that they occur in `x`. This operation also returns a</span>
+<a name="line-6921"></a><span class='hs-comment'>-- tensor `idx` the same size as `x` that contains the index of each value of `x`</span>
+<a name="line-6922"></a><span class='hs-comment'>-- in the unique output `y`. In other words:</span>
+<a name="line-6923"></a><span class='hs-comment'>-- </span>
+<a name="line-6924"></a><span class='hs-comment'>-- `y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`</span>
+<a name="line-6925"></a><span class='hs-comment'>-- </span>
+<a name="line-6926"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-6927"></a><span class='hs-comment'>-- </span>
+<a name="line-6928"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-6929"></a><span class='hs-comment'>-- # tensor 'x' is [1, 1, 2, 4, 4, 4, 7, 8, 8]</span>
+<a name="line-6930"></a><span class='hs-comment'>-- y, idx = unique(x)</span>
+<a name="line-6931"></a><span class='hs-comment'>-- y ==&gt; [1, 2, 4, 7, 8]</span>
+<a name="line-6932"></a><span class='hs-comment'>-- idx ==&gt; [0, 0, 1, 2, 2, 2, 3, 4, 4]</span>
+<a name="line-6933"></a><span class='hs-comment'>-- ```</span>
+<a name="line-6934"></a><span class='hs-definition'>unique</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varid'>out_idx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>out_idx</span><span class='hs-layout'>,</span>
+<a name="line-6935"></a>                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-6936"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>out_idx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-6937"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__: 1-D.</span>
+<a name="line-6938"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>out_idx</span><span class='hs-layout'>)</span>
+<a name="line-6939"></a>          <span class='hs-comment'>-- ^ (__y__, __idx__)</span>
+<a name="line-6940"></a>          <span class='hs-comment'>--</span>
+<a name="line-6941"></a>          <span class='hs-comment'>-- * __y__: 1-D.</span>
+<a name="line-6942"></a>          <span class='hs-comment'>--</span>
+<a name="line-6943"></a>          <span class='hs-comment'>-- * __idx__: 1-D.</span>
+<a name="line-6944"></a><span class='hs-definition'>unique</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6945"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Unique"</span>
+<a name="line-6946"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-6947"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"out_idx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>out_idx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-6948"></a>        <span class='hs-varid'>x</span>
+<a name="line-6949"></a><span class='hs-comment'>{-
+<a name="line-6950"></a>attr { name: "T" type: "type" }
+<a name="line-6951"></a>attr {
+<a name="line-6952"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-6953"></a>  default_value { type: DT_INT32 }
+<a name="line-6954"></a>  name: "out_idx"
+<a name="line-6955"></a>  type: "type"
+<a name="line-6956"></a>}
+<a name="line-6957"></a>input_arg { description: "1-D." name: "x" type_attr: "T" }
+<a name="line-6958"></a>output_arg { description: "1-D." name: "y" type_attr: "T" }
+<a name="line-6959"></a>output_arg { description: "1-D." name: "idx" type_attr: "out_idx" }
+<a name="line-6960"></a>-}</span>
+<a name="line-6961"></a>
+<a name="line-6962"></a><a name="drawBoundingBoxes"></a><span class='hs-comment'>-- | Draw bounding boxes on a batch of images.</span>
+<a name="line-6963"></a><span class='hs-comment'>--</span>
+<a name="line-6964"></a><span class='hs-comment'>-- Outputs a copy of `images` but draws on top of the pixels zero or more bounding</span>
+<a name="line-6965"></a><span class='hs-comment'>-- boxes specified by the locations in `boxes`. The coordinates of the each</span>
+<a name="line-6966"></a><span class='hs-comment'>-- bounding box in `boxes` are encoded as `[y_min, x_min, y_max, x_max]`. The</span>
+<a name="line-6967"></a><span class='hs-comment'>-- bounding box coordinates are floats in `[0.0, 1.0]` relative to the width and</span>
+<a name="line-6968"></a><span class='hs-comment'>-- height of the underlying image.</span>
+<a name="line-6969"></a><span class='hs-comment'>-- </span>
+<a name="line-6970"></a><span class='hs-comment'>-- For example, if an image is 100 x 200 pixels and the bounding box is</span>
+<a name="line-6971"></a><span class='hs-comment'>-- `[0.1, 0.2, 0.5, 0.9]`, the bottom-left and upper-right coordinates of the</span>
+<a name="line-6972"></a><span class='hs-comment'>-- bounding box will be `(10, 40)` to `(50, 180)`.</span>
+<a name="line-6973"></a><span class='hs-comment'>-- </span>
+<a name="line-6974"></a><span class='hs-comment'>-- Parts of the bounding box may fall outside the image.</span>
+<a name="line-6975"></a><span class='hs-definition'>drawBoundingBoxes</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-6976"></a>                                                             <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-6977"></a>                     <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __images__: 4-D with shape `[batch, height, width, depth]`. A batch of images.</span>
+<a name="line-6978"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __boxes__: 3-D with shape `[batch, num_bounding_boxes, 4]` containing bounding</span>
+<a name="line-6979"></a>                                        <span class='hs-comment'>-- boxes.</span>
+<a name="line-6980"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 4-D with the same shape as `images`. The batch of input images with</span>
+<a name="line-6981"></a>                     <span class='hs-comment'>-- bounding boxes drawn on the images.</span>
+<a name="line-6982"></a><span class='hs-definition'>drawBoundingBoxes</span> <span class='hs-varid'>images</span> <span class='hs-varid'>boxes</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-6983"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DrawBoundingBoxes"</span>
+<a name="line-6984"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-6985"></a>        <span class='hs-varid'>images</span> <span class='hs-varid'>boxes</span>
+<a name="line-6986"></a><span class='hs-comment'>{-
+<a name="line-6987"></a>attr {
+<a name="line-6988"></a>  allowed_values { list { type: DT_FLOAT type: DT_HALF } }
+<a name="line-6989"></a>  default_value { type: DT_FLOAT }
+<a name="line-6990"></a>  name: "T"
+<a name="line-6991"></a>  type: "type"
+<a name="line-6992"></a>}
+<a name="line-6993"></a>input_arg {
+<a name="line-6994"></a>  description: "4-D with shape `[batch, height, width, depth]`. A batch of images."
+<a name="line-6995"></a>  name: "images"
+<a name="line-6996"></a>  type_attr: "T"
+<a name="line-6997"></a>}
+<a name="line-6998"></a>input_arg {
+<a name="line-6999"></a>  description: "3-D with shape `[batch, num_bounding_boxes, 4]` containing bounding\nboxes."
+<a name="line-7000"></a>  name: "boxes"
+<a name="line-7001"></a>  type: DT_FLOAT
+<a name="line-7002"></a>}
+<a name="line-7003"></a>output_arg {
+<a name="line-7004"></a>  description: "4-D with the same shape as `images`. The batch of input images with\nbounding boxes drawn on the images."
+<a name="line-7005"></a>  name: "output"
+<a name="line-7006"></a>  type_attr: "T"
+<a name="line-7007"></a>}
+<a name="line-7008"></a>-}</span>
+<a name="line-7009"></a>
+<a name="line-7010"></a><a name="tensorArraySplit"></a><span class='hs-comment'>-- | Split the data from the input value into TensorArray elements.</span>
+<a name="line-7011"></a><span class='hs-comment'>--</span>
+<a name="line-7012"></a><span class='hs-comment'>-- Assuming that `lengths` takes on values</span>
+<a name="line-7013"></a><span class='hs-comment'>-- </span>
+<a name="line-7014"></a><span class='hs-comment'>--   ```(n0, n1, ..., n(T-1))```</span>
+<a name="line-7015"></a><span class='hs-comment'>-- </span>
+<a name="line-7016"></a><span class='hs-comment'>-- and that `value` has shape</span>
+<a name="line-7017"></a><span class='hs-comment'>-- </span>
+<a name="line-7018"></a><span class='hs-comment'>--   ```(n0 + n1 + ... + n(T-1) x d0 x d1 x ...)```,</span>
+<a name="line-7019"></a><span class='hs-comment'>-- </span>
+<a name="line-7020"></a><span class='hs-comment'>-- this splits values into a TensorArray with T tensors.</span>
+<a name="line-7021"></a><span class='hs-comment'>-- </span>
+<a name="line-7022"></a><span class='hs-comment'>-- TensorArray index t will be the subtensor of values with starting position</span>
+<a name="line-7023"></a><span class='hs-comment'>-- </span>
+<a name="line-7024"></a><span class='hs-comment'>--   ```(n0 + n1 + ... + n(t-1), 0, 0, ...)```</span>
+<a name="line-7025"></a><span class='hs-comment'>-- </span>
+<a name="line-7026"></a><span class='hs-comment'>-- and having size</span>
+<a name="line-7027"></a><span class='hs-comment'>-- </span>
+<a name="line-7028"></a><span class='hs-comment'>--   ```nt x d0 x d1 x ...```</span>
+<a name="line-7029"></a><span class='hs-definition'>tensorArraySplit</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-7030"></a>                    <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a TensorArray.</span>
+<a name="line-7031"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: The concatenated tensor to write to the TensorArray.</span>
+<a name="line-7032"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __lengths__: The vector of lengths, how to split the rows of value into the</span>
+<a name="line-7033"></a>                                                <span class='hs-comment'>-- TensorArray.</span>
+<a name="line-7034"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __flow_in__: A float scalar that enforces proper chaining of operations.</span>
+<a name="line-7035"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __flow_out__: A float scalar that enforces proper chaining of operations.</span>
+<a name="line-7036"></a><span class='hs-definition'>tensorArraySplit</span> <span class='hs-varid'>handle</span> <span class='hs-varid'>value</span> <span class='hs-varid'>lengths</span> <span class='hs-varid'>flow_in</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7037"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TensorArraySplit"</span>
+<a name="line-7038"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-7039"></a>        <span class='hs-varid'>handle</span> <span class='hs-varid'>value</span> <span class='hs-varid'>lengths</span> <span class='hs-varid'>flow_in</span>
+<a name="line-7040"></a><span class='hs-comment'>{-
+<a name="line-7041"></a>attr { name: "T" type: "type" }
+<a name="line-7042"></a>input_arg {
+<a name="line-7043"></a>  description: "The handle to a TensorArray."
+<a name="line-7044"></a>  is_ref: true
+<a name="line-7045"></a>  name: "handle"
+<a name="line-7046"></a>  type: DT_STRING
+<a name="line-7047"></a>}
+<a name="line-7048"></a>input_arg {
+<a name="line-7049"></a>  description: "The concatenated tensor to write to the TensorArray."
+<a name="line-7050"></a>  name: "value"
+<a name="line-7051"></a>  type_attr: "T"
+<a name="line-7052"></a>}
+<a name="line-7053"></a>input_arg {
+<a name="line-7054"></a>  description: "The vector of lengths, how to split the rows of value into the\nTensorArray."
+<a name="line-7055"></a>  name: "lengths"
+<a name="line-7056"></a>  type: DT_INT64
+<a name="line-7057"></a>}
+<a name="line-7058"></a>input_arg {
+<a name="line-7059"></a>  description: "A float scalar that enforces proper chaining of operations."
+<a name="line-7060"></a>  name: "flow_in"
+<a name="line-7061"></a>  type: DT_FLOAT
+<a name="line-7062"></a>}
+<a name="line-7063"></a>output_arg {
+<a name="line-7064"></a>  description: "A float scalar that enforces proper chaining of operations."
+<a name="line-7065"></a>  name: "flow_out"
+<a name="line-7066"></a>  type: DT_FLOAT
+<a name="line-7067"></a>}
+<a name="line-7068"></a>-}</span>
+<a name="line-7069"></a>
+<a name="line-7070"></a><span class='hs-comment'>-- | Splits a tensor into `num_split` tensors along one dimension.</span>
+<a name="line-7071"></a>
+<a name="line-7072"></a><a name="split"></a><span class='hs-definition'>split</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-7073"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_split__: The number of ways to split.  Must evenly divide</span>
+<a name="line-7074"></a>                        <span class='hs-comment'>-- `value.shape[split_dim]`.</span>
+<a name="line-7075"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __split_dim__: 0-D.  The dimension along which to split.  Must be in the range</span>
+<a name="line-7076"></a>                                     <span class='hs-comment'>-- `[0, rank(value))`.</span>
+<a name="line-7077"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: The tensor to split.</span>
+<a name="line-7078"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __output__: They are identically shaped tensors, whose shape matches that of `value`</span>
+<a name="line-7079"></a>         <span class='hs-comment'>-- except along `split_dim`, where their sizes are</span>
+<a name="line-7080"></a>         <span class='hs-comment'>-- `values.shape[split_dim] / num_split`.</span>
+<a name="line-7081"></a><span class='hs-definition'>split</span> <span class='hs-varid'>num_split</span> <span class='hs-varid'>split_dim</span> <span class='hs-varid'>value</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7082"></a>    <span class='hs-varid'>buildListOp</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>num_split</span><span class='hs-keyglyph'>]</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Split"</span>
+<a name="line-7083"></a>                             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-7084"></a>                             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_split"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_split</span><span class='hs-layout'>)</span>
+<a name="line-7085"></a>        <span class='hs-varid'>split_dim</span> <span class='hs-varid'>value</span>
+<a name="line-7086"></a><span class='hs-comment'>{-
+<a name="line-7087"></a>attr {
+<a name="line-7088"></a>  description: "The number of ways to split.  Must evenly divide\n`value.shape[split_dim]`."
+<a name="line-7089"></a>  has_minimum: true
+<a name="line-7090"></a>  minimum: 1
+<a name="line-7091"></a>  name: "num_split"
+<a name="line-7092"></a>  type: "int"
+<a name="line-7093"></a>}
+<a name="line-7094"></a>attr { name: "T" type: "type" }
+<a name="line-7095"></a>input_arg {
+<a name="line-7096"></a>  description: "0-D.  The dimension along which to split.  Must be in the range\n`[0, rank(value))`."
+<a name="line-7097"></a>  name: "split_dim"
+<a name="line-7098"></a>  type: DT_INT32
+<a name="line-7099"></a>}
+<a name="line-7100"></a>input_arg {
+<a name="line-7101"></a>  description: "The tensor to split." name: "value" type_attr: "T"
+<a name="line-7102"></a>}
+<a name="line-7103"></a>output_arg {
+<a name="line-7104"></a>  description: "They are identically shaped tensors, whose shape matches that of `value`\nexcept along `split_dim`, where their sizes are\n`values.shape[split_dim] / num_split`."
+<a name="line-7105"></a>  name: "output"
+<a name="line-7106"></a>  number_attr: "num_split"
+<a name="line-7107"></a>  type_attr: "T"
+<a name="line-7108"></a>}
+<a name="line-7109"></a>-}</span>
+<a name="line-7110"></a>
+<a name="line-7111"></a><a name="segmentMax"></a><span class='hs-comment'>-- | Computes the maximum along segments of a tensor.</span>
+<a name="line-7112"></a><span class='hs-comment'>--</span>
+<a name="line-7113"></a><span class='hs-comment'>-- Read [the section on Segmentation](../../api_docs/python/math_ops.md#segmentation)</span>
+<a name="line-7114"></a><span class='hs-comment'>-- for an explanation of segments.</span>
+<a name="line-7115"></a><span class='hs-comment'>-- </span>
+<a name="line-7116"></a><span class='hs-comment'>-- Computes a tensor such that</span>
+<a name="line-7117"></a><span class='hs-comment'>-- \\(output_i = \max_j(data_j)\\) where `max` is over `j` such</span>
+<a name="line-7118"></a><span class='hs-comment'>-- that `segment_ids[j] == i`.</span>
+<a name="line-7119"></a><span class='hs-comment'>-- </span>
+<a name="line-7120"></a><span class='hs-comment'>-- &lt;div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;"&gt;</span>
+<a name="line-7121"></a><span class='hs-comment'>-- &lt;img style="width:100%" src="../../images/SegmentMax.png" alt&gt;</span>
+<a name="line-7122"></a><span class='hs-comment'>-- &lt;/div&gt;</span>
+<a name="line-7123"></a><span class='hs-definition'>segmentMax</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-7124"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-7125"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-7126"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-7127"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-7128"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-7129"></a>                                                               <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-7130"></a>                                         <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-7131"></a>                                         <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-7132"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-7133"></a>              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__</span>
+<a name="line-7134"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __segment_ids__: A 1-D tensor whose rank is equal to the rank of `data`'s</span>
+<a name="line-7135"></a>                                    <span class='hs-comment'>-- first dimension.  Values should be sorted and can be repeated.</span>
+<a name="line-7136"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Has same shape as data, except for dimension 0 which</span>
+<a name="line-7137"></a>              <span class='hs-comment'>-- has size `k`, the number of segments.</span>
+<a name="line-7138"></a><span class='hs-definition'>segmentMax</span> <span class='hs-varid'>data'</span> <span class='hs-varid'>segment_ids</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7139"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SegmentMax"</span>
+<a name="line-7140"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-7141"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-7142"></a>        <span class='hs-varid'>data'</span> <span class='hs-varid'>segment_ids</span>
+<a name="line-7143"></a><span class='hs-comment'>{-
+<a name="line-7144"></a>attr {
+<a name="line-7145"></a>  allowed_values {
+<a name="line-7146"></a>    list {
+<a name="line-7147"></a>      type: DT_FLOAT
+<a name="line-7148"></a>      type: DT_DOUBLE
+<a name="line-7149"></a>      type: DT_INT32
+<a name="line-7150"></a>      type: DT_INT64
+<a name="line-7151"></a>      type: DT_UINT8
+<a name="line-7152"></a>      type: DT_INT16
+<a name="line-7153"></a>      type: DT_INT8
+<a name="line-7154"></a>      type: DT_UINT16
+<a name="line-7155"></a>      type: DT_HALF
+<a name="line-7156"></a>    }
+<a name="line-7157"></a>  }
+<a name="line-7158"></a>  name: "T"
+<a name="line-7159"></a>  type: "type"
+<a name="line-7160"></a>}
+<a name="line-7161"></a>attr {
+<a name="line-7162"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-7163"></a>  name: "Tindices"
+<a name="line-7164"></a>  type: "type"
+<a name="line-7165"></a>}
+<a name="line-7166"></a>input_arg { name: "data" type_attr: "T" }
+<a name="line-7167"></a>input_arg {
+<a name="line-7168"></a>  description: "A 1-D tensor whose rank is equal to the rank of `data`\'s\nfirst dimension.  Values should be sorted and can be repeated."
+<a name="line-7169"></a>  name: "segment_ids"
+<a name="line-7170"></a>  type_attr: "Tindices"
+<a name="line-7171"></a>}
+<a name="line-7172"></a>output_arg {
+<a name="line-7173"></a>  description: "Has same shape as data, except for dimension 0 which\nhas size `k`, the number of segments."
+<a name="line-7174"></a>  name: "output"
+<a name="line-7175"></a>  type_attr: "T"
+<a name="line-7176"></a>}
+<a name="line-7177"></a>-}</span>
+<a name="line-7178"></a>
+<a name="line-7179"></a><a name="abort"></a><span class='hs-comment'>-- | Raise a exception to abort the process when called.</span>
+<a name="line-7180"></a><span class='hs-comment'>--</span>
+<a name="line-7181"></a><span class='hs-comment'>-- Returns nothing but an exception.</span>
+<a name="line-7182"></a><span class='hs-definition'>abort</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-7183"></a><span class='hs-definition'>abort</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7184"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Abort"</span><span class='hs-layout'>)</span>
+<a name="line-7185"></a>        
+<a name="line-7186"></a><span class='hs-comment'>{-
+<a name="line-7187"></a>attr {
+<a name="line-7188"></a>  default_value { s: "" }
+<a name="line-7189"></a>  description: "A string which is the message associated with the exception."
+<a name="line-7190"></a>  name: "error_msg"
+<a name="line-7191"></a>  type: "string"
+<a name="line-7192"></a>}
+<a name="line-7193"></a>-}</span>
+<a name="line-7194"></a>
+<a name="line-7195"></a><a name="sparseReorder"></a><span class='hs-comment'>-- | Reorders a SparseTensor into the canonical, row-major ordering.</span>
+<a name="line-7196"></a><span class='hs-comment'>--</span>
+<a name="line-7197"></a><span class='hs-comment'>-- Note that by convention, all sparse ops preserve the canonical ordering along</span>
+<a name="line-7198"></a><span class='hs-comment'>-- increasing dimension number. The only time ordering can be violated is during</span>
+<a name="line-7199"></a><span class='hs-comment'>-- manual manipulation of the indices and values vectors to add entries.</span>
+<a name="line-7200"></a><span class='hs-comment'>-- </span>
+<a name="line-7201"></a><span class='hs-comment'>-- Reordering does not affect the shape of the SparseTensor.</span>
+<a name="line-7202"></a><span class='hs-comment'>-- </span>
+<a name="line-7203"></a><span class='hs-comment'>-- If the tensor has rank `R` and `N` non-empty values, `input_indices` has</span>
+<a name="line-7204"></a><span class='hs-comment'>-- shape `[N, R]`, input_values has length `N`, and input_shape has length `R`.</span>
+<a name="line-7205"></a><span class='hs-definition'>sparseReorder</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-7206"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __input_indices__: 2-D.  `N x R` matrix with the indices of non-empty values in a</span>
+<a name="line-7207"></a>                                          <span class='hs-comment'>-- SparseTensor, possibly not in canonical ordering.</span>
+<a name="line-7208"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input_values__: 1-D.  `N` non-empty values corresponding to `input_indices`.</span>
+<a name="line-7209"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __input_shape__: 1-D.  Shape of the input SparseTensor.</span>
+<a name="line-7210"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-7211"></a>                 <span class='hs-comment'>-- ^ (__output_indices__, __output_values__)</span>
+<a name="line-7212"></a>                 <span class='hs-comment'>--</span>
+<a name="line-7213"></a>                 <span class='hs-comment'>-- * __output_indices__: 2-D.  `N x R` matrix with the same indices as input_indices, but</span>
+<a name="line-7214"></a>                 <span class='hs-comment'>-- in canonical row-major ordering.</span>
+<a name="line-7215"></a>                 <span class='hs-comment'>--</span>
+<a name="line-7216"></a>                 <span class='hs-comment'>-- * __output_values__: 1-D.  `N` non-empty values corresponding to `output_indices`.</span>
+<a name="line-7217"></a><span class='hs-definition'>sparseReorder</span> <span class='hs-varid'>input_indices</span> <span class='hs-varid'>input_values</span> <span class='hs-varid'>input_shape</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7218"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseReorder"</span>
+<a name="line-7219"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-7220"></a>        <span class='hs-varid'>input_indices</span> <span class='hs-varid'>input_values</span> <span class='hs-varid'>input_shape</span>
+<a name="line-7221"></a><span class='hs-comment'>{-
+<a name="line-7222"></a>attr { name: "T" type: "type" }
+<a name="line-7223"></a>input_arg {
+<a name="line-7224"></a>  description: "2-D.  `N x R` matrix with the indices of non-empty values in a\nSparseTensor, possibly not in canonical ordering."
+<a name="line-7225"></a>  name: "input_indices"
+<a name="line-7226"></a>  type: DT_INT64
+<a name="line-7227"></a>}
+<a name="line-7228"></a>input_arg {
+<a name="line-7229"></a>  description: "1-D.  `N` non-empty values corresponding to `input_indices`."
+<a name="line-7230"></a>  name: "input_values"
+<a name="line-7231"></a>  type_attr: "T"
+<a name="line-7232"></a>}
+<a name="line-7233"></a>input_arg {
+<a name="line-7234"></a>  description: "1-D.  Shape of the input SparseTensor."
+<a name="line-7235"></a>  name: "input_shape"
+<a name="line-7236"></a>  type: DT_INT64
+<a name="line-7237"></a>}
+<a name="line-7238"></a>output_arg {
+<a name="line-7239"></a>  description: "2-D.  `N x R` matrix with the same indices as input_indices, but\nin canonical row-major ordering."
+<a name="line-7240"></a>  name: "output_indices"
+<a name="line-7241"></a>  type: DT_INT64
+<a name="line-7242"></a>}
+<a name="line-7243"></a>output_arg {
+<a name="line-7244"></a>  description: "1-D.  `N` non-empty values corresponding to `output_indices`."
+<a name="line-7245"></a>  name: "output_values"
+<a name="line-7246"></a>  type_attr: "T"
+<a name="line-7247"></a>}
+<a name="line-7248"></a>-}</span>
+<a name="line-7249"></a>
+<a name="line-7250"></a><a name="rsqrtGrad"></a><span class='hs-comment'>-- | Computes the gradient for the rsqrt of `x` wrt its input.</span>
+<a name="line-7251"></a><span class='hs-comment'>--</span>
+<a name="line-7252"></a><span class='hs-comment'>-- Specifically, `grad = dy * -0.5 * y^3`, where `y = rsqrt(x)`, and `dy`</span>
+<a name="line-7253"></a><span class='hs-comment'>-- is the corresponding input gradient.</span>
+<a name="line-7254"></a><span class='hs-definition'>rsqrtGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-7255"></a>                               <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-7256"></a>                                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-7257"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-7258"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-7259"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-7260"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-7261"></a><span class='hs-definition'>rsqrtGrad</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7262"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RsqrtGrad"</span>
+<a name="line-7263"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-7264"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-7265"></a><span class='hs-comment'>{-
+<a name="line-7266"></a>attr {
+<a name="line-7267"></a>  allowed_values {
+<a name="line-7268"></a>    list {
+<a name="line-7269"></a>      type: DT_HALF
+<a name="line-7270"></a>      type: DT_FLOAT
+<a name="line-7271"></a>      type: DT_DOUBLE
+<a name="line-7272"></a>      type: DT_COMPLEX64
+<a name="line-7273"></a>      type: DT_COMPLEX128
+<a name="line-7274"></a>    }
+<a name="line-7275"></a>  }
+<a name="line-7276"></a>  name: "T"
+<a name="line-7277"></a>  type: "type"
+<a name="line-7278"></a>}
+<a name="line-7279"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-7280"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-7281"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-7282"></a>-}</span>
+<a name="line-7283"></a>
+<a name="line-7284"></a><a name="reverseSequence"></a><span class='hs-comment'>-- | Reverses variable length slices.</span>
+<a name="line-7285"></a><span class='hs-comment'>--</span>
+<a name="line-7286"></a><span class='hs-comment'>-- This op first slices `input` along the dimension `batch_dim`, and for each</span>
+<a name="line-7287"></a><span class='hs-comment'>-- slice `i`, reverses the first `seq_lengths[i]` elements along</span>
+<a name="line-7288"></a><span class='hs-comment'>-- the dimension `seq_dim`.</span>
+<a name="line-7289"></a><span class='hs-comment'>-- </span>
+<a name="line-7290"></a><span class='hs-comment'>-- The elements of `seq_lengths` must obey `seq_lengths[i] &lt; input.dims[seq_dim]`,</span>
+<a name="line-7291"></a><span class='hs-comment'>-- and `seq_lengths` must be a vector of length `input.dims[batch_dim]`.</span>
+<a name="line-7292"></a><span class='hs-comment'>-- </span>
+<a name="line-7293"></a><span class='hs-comment'>-- The output slice `i` along dimension `batch_dim` is then given by input</span>
+<a name="line-7294"></a><span class='hs-comment'>-- slice `i`, with the first `seq_lengths[i]` slices along dimension</span>
+<a name="line-7295"></a><span class='hs-comment'>-- `seq_dim` reversed.</span>
+<a name="line-7296"></a><span class='hs-comment'>-- </span>
+<a name="line-7297"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-7298"></a><span class='hs-comment'>-- </span>
+<a name="line-7299"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-7300"></a><span class='hs-comment'>-- # Given this:</span>
+<a name="line-7301"></a><span class='hs-comment'>-- batch_dim = 0</span>
+<a name="line-7302"></a><span class='hs-comment'>-- seq_dim = 1</span>
+<a name="line-7303"></a><span class='hs-comment'>-- input.dims = (4, 8, ...)</span>
+<a name="line-7304"></a><span class='hs-comment'>-- seq_lengths = [7, 2, 3, 5]</span>
+<a name="line-7305"></a><span class='hs-comment'>-- </span>
+<a name="line-7306"></a><span class='hs-comment'>-- # then slices of input are reversed on seq_dim, but only up to seq_lengths:</span>
+<a name="line-7307"></a><span class='hs-comment'>-- output[0, 0:7, :, ...] = input[0, 7:0:-1, :, ...]</span>
+<a name="line-7308"></a><span class='hs-comment'>-- output[1, 0:2, :, ...] = input[1, 2:0:-1, :, ...]</span>
+<a name="line-7309"></a><span class='hs-comment'>-- output[2, 0:3, :, ...] = input[2, 3:0:-1, :, ...]</span>
+<a name="line-7310"></a><span class='hs-comment'>-- output[3, 0:5, :, ...] = input[3, 5:0:-1, :, ...]</span>
+<a name="line-7311"></a><span class='hs-comment'>-- </span>
+<a name="line-7312"></a><span class='hs-comment'>-- # while entries past seq_lens are copied through:</span>
+<a name="line-7313"></a><span class='hs-comment'>-- output[0, 7:, :, ...] = input[0, 7:, :, ...]</span>
+<a name="line-7314"></a><span class='hs-comment'>-- output[1, 2:, :, ...] = input[1, 2:, :, ...]</span>
+<a name="line-7315"></a><span class='hs-comment'>-- output[2, 3:, :, ...] = input[2, 3:, :, ...]</span>
+<a name="line-7316"></a><span class='hs-comment'>-- output[3, 2:, :, ...] = input[3, 2:, :, ...]</span>
+<a name="line-7317"></a><span class='hs-comment'>-- ```</span>
+<a name="line-7318"></a><span class='hs-comment'>-- </span>
+<a name="line-7319"></a><span class='hs-comment'>-- In contrast, if:</span>
+<a name="line-7320"></a><span class='hs-comment'>-- </span>
+<a name="line-7321"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-7322"></a><span class='hs-comment'>-- # Given this:</span>
+<a name="line-7323"></a><span class='hs-comment'>-- batch_dim = 2</span>
+<a name="line-7324"></a><span class='hs-comment'>-- seq_dim = 0</span>
+<a name="line-7325"></a><span class='hs-comment'>-- input.dims = (8, ?, 4, ...)</span>
+<a name="line-7326"></a><span class='hs-comment'>-- seq_lengths = [7, 2, 3, 5]</span>
+<a name="line-7327"></a><span class='hs-comment'>-- </span>
+<a name="line-7328"></a><span class='hs-comment'>-- # then slices of input are reversed on seq_dim, but only up to seq_lengths:</span>
+<a name="line-7329"></a><span class='hs-comment'>-- output[0:7, :, 0, :, ...] = input[7:0:-1, :, 0, :, ...]</span>
+<a name="line-7330"></a><span class='hs-comment'>-- output[0:2, :, 1, :, ...] = input[2:0:-1, :, 1, :, ...]</span>
+<a name="line-7331"></a><span class='hs-comment'>-- output[0:3, :, 2, :, ...] = input[3:0:-1, :, 2, :, ...]</span>
+<a name="line-7332"></a><span class='hs-comment'>-- output[0:5, :, 3, :, ...] = input[5:0:-1, :, 3, :, ...]</span>
+<a name="line-7333"></a><span class='hs-comment'>-- </span>
+<a name="line-7334"></a><span class='hs-comment'>-- # while entries past seq_lens are copied through:</span>
+<a name="line-7335"></a><span class='hs-comment'>-- output[7:, :, 0, :, ...] = input[7:, :, 0, :, ...]</span>
+<a name="line-7336"></a><span class='hs-comment'>-- output[2:, :, 1, :, ...] = input[2:, :, 1, :, ...]</span>
+<a name="line-7337"></a><span class='hs-comment'>-- output[3:, :, 2, :, ...] = input[3:, :, 2, :, ...]</span>
+<a name="line-7338"></a><span class='hs-comment'>-- output[2:, :, 3, :, ...] = input[2:, :, 3, :, ...]</span>
+<a name="line-7339"></a><span class='hs-comment'>-- ```</span>
+<a name="line-7340"></a><span class='hs-definition'>reverseSequence</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tlen</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tlen</span><span class='hs-layout'>,</span>
+<a name="line-7341"></a>                                          <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-7342"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tlen</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-7343"></a>                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __seq_dim__: The dimension which is partially reversed.</span>
+<a name="line-7344"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: The input to reverse.</span>
+<a name="line-7345"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tlen</span> <span class='hs-comment'>-- ^ __seq_lengths__: 1-D with length `input.dims(batch_dim)` and</span>
+<a name="line-7346"></a>                                     <span class='hs-comment'>-- `max(seq_lengths) &lt; input.dims(seq_dim)`</span>
+<a name="line-7347"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The partially reversed input. It has the same shape as `input`.</span>
+<a name="line-7348"></a><span class='hs-definition'>reverseSequence</span> <span class='hs-varid'>seq_dim</span> <span class='hs-varid'>input</span> <span class='hs-varid'>seq_lengths</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7349"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ReverseSequence"</span>
+<a name="line-7350"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-7351"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tlen"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tlen</span><span class='hs-layout'>)</span>
+<a name="line-7352"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"seq_dim"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>seq_dim</span><span class='hs-layout'>)</span>
+<a name="line-7353"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>seq_lengths</span>
+<a name="line-7354"></a><span class='hs-comment'>{-
+<a name="line-7355"></a>attr {
+<a name="line-7356"></a>  description: "The dimension which is partially reversed."
+<a name="line-7357"></a>  name: "seq_dim"
+<a name="line-7358"></a>  type: "int"
+<a name="line-7359"></a>}
+<a name="line-7360"></a>attr {
+<a name="line-7361"></a>  default_value { i: 0 }
+<a name="line-7362"></a>  description: "The dimension along which reversal is performed."
+<a name="line-7363"></a>  name: "batch_dim"
+<a name="line-7364"></a>  type: "int"
+<a name="line-7365"></a>}
+<a name="line-7366"></a>attr { name: "T" type: "type" }
+<a name="line-7367"></a>attr {
+<a name="line-7368"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-7369"></a>  default_value { type: DT_INT64 }
+<a name="line-7370"></a>  name: "Tlen"
+<a name="line-7371"></a>  type: "type"
+<a name="line-7372"></a>}
+<a name="line-7373"></a>input_arg {
+<a name="line-7374"></a>  description: "The input to reverse." name: "input" type_attr: "T"
+<a name="line-7375"></a>}
+<a name="line-7376"></a>input_arg {
+<a name="line-7377"></a>  description: "1-D with length `input.dims(batch_dim)` and\n`max(seq_lengths) &lt; input.dims(seq_dim)`"
+<a name="line-7378"></a>  name: "seq_lengths"
+<a name="line-7379"></a>  type_attr: "Tlen"
+<a name="line-7380"></a>}
+<a name="line-7381"></a>output_arg {
+<a name="line-7382"></a>  description: "The partially reversed input. It has the same shape as `input`."
+<a name="line-7383"></a>  name: "output"
+<a name="line-7384"></a>  type_attr: "T"
+<a name="line-7385"></a>}
+<a name="line-7386"></a>-}</span>
+<a name="line-7387"></a>
+<a name="line-7388"></a><a name="readerNumRecordsProduced"></a><span class='hs-comment'>-- | Returns the number of records this Reader has produced.</span>
+<a name="line-7389"></a><span class='hs-comment'>--</span>
+<a name="line-7390"></a><span class='hs-comment'>-- This is the same as the number of ReaderRead executions that have</span>
+<a name="line-7391"></a><span class='hs-comment'>-- succeeded.</span>
+<a name="line-7392"></a><span class='hs-definition'>readerNumRecordsProduced</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __reader_handle__: Handle to a Reader.</span>
+<a name="line-7393"></a>                            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __records_produced__</span>
+<a name="line-7394"></a><span class='hs-definition'>readerNumRecordsProduced</span> <span class='hs-varid'>reader_handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7395"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ReaderNumRecordsProduced"</span><span class='hs-layout'>)</span>
+<a name="line-7396"></a>        <span class='hs-varid'>reader_handle</span>
+<a name="line-7397"></a><span class='hs-comment'>{-
+<a name="line-7398"></a>input_arg {
+<a name="line-7399"></a>  description: "Handle to a Reader."
+<a name="line-7400"></a>  is_ref: true
+<a name="line-7401"></a>  name: "reader_handle"
+<a name="line-7402"></a>  type: DT_STRING
+<a name="line-7403"></a>}
+<a name="line-7404"></a>output_arg { name: "records_produced" type: DT_INT64 }
+<a name="line-7405"></a>-}</span>
+<a name="line-7406"></a>
+<a name="line-7407"></a><a name="deserializeManySparse"></a><span class='hs-comment'>-- | Deserialize and concatenate `SparseTensors` from a serialized minibatch.</span>
+<a name="line-7408"></a><span class='hs-comment'>--</span>
+<a name="line-7409"></a><span class='hs-comment'>-- The input `serialized_sparse` must be a string matrix of shape `[N x 3]` where</span>
+<a name="line-7410"></a><span class='hs-comment'>-- `N` is the minibatch size and the rows correspond to packed outputs of</span>
+<a name="line-7411"></a><span class='hs-comment'>-- `SerializeSparse`.  The ranks of the original `SparseTensor` objects</span>
+<a name="line-7412"></a><span class='hs-comment'>-- must all match.  When the final `SparseTensor` is created, it has rank one</span>
+<a name="line-7413"></a><span class='hs-comment'>-- higher than the ranks of the incoming `SparseTensor` objects</span>
+<a name="line-7414"></a><span class='hs-comment'>-- (they have been concatenated along a new row dimension).</span>
+<a name="line-7415"></a><span class='hs-comment'>-- </span>
+<a name="line-7416"></a><span class='hs-comment'>-- The output `SparseTensor` object's shape values for all dimensions but the</span>
+<a name="line-7417"></a><span class='hs-comment'>-- first are the max across the input `SparseTensor` objects' shape values</span>
+<a name="line-7418"></a><span class='hs-comment'>-- for the corresponding dimensions.  Its first shape value is `N`, the minibatch</span>
+<a name="line-7419"></a><span class='hs-comment'>-- size.</span>
+<a name="line-7420"></a><span class='hs-comment'>-- </span>
+<a name="line-7421"></a><span class='hs-comment'>-- The input `SparseTensor` objects' indices are assumed ordered in</span>
+<a name="line-7422"></a><span class='hs-comment'>-- standard lexicographic order.  If this is not the case, after this</span>
+<a name="line-7423"></a><span class='hs-comment'>-- step run `SparseReorder` to restore index ordering.</span>
+<a name="line-7424"></a><span class='hs-comment'>-- </span>
+<a name="line-7425"></a><span class='hs-comment'>-- For example, if the serialized input is a `[2 x 3]` matrix representing two</span>
+<a name="line-7426"></a><span class='hs-comment'>-- original `SparseTensor` objects:</span>
+<a name="line-7427"></a><span class='hs-comment'>-- </span>
+<a name="line-7428"></a><span class='hs-comment'>--     index = [ 0]</span>
+<a name="line-7429"></a><span class='hs-comment'>--             [10]</span>
+<a name="line-7430"></a><span class='hs-comment'>--             [20]</span>
+<a name="line-7431"></a><span class='hs-comment'>--     values = [1, 2, 3]</span>
+<a name="line-7432"></a><span class='hs-comment'>--     shape = [50]</span>
+<a name="line-7433"></a><span class='hs-comment'>-- </span>
+<a name="line-7434"></a><span class='hs-comment'>-- and</span>
+<a name="line-7435"></a><span class='hs-comment'>-- </span>
+<a name="line-7436"></a><span class='hs-comment'>--     index = [ 2]</span>
+<a name="line-7437"></a><span class='hs-comment'>--             [10]</span>
+<a name="line-7438"></a><span class='hs-comment'>--     values = [4, 5]</span>
+<a name="line-7439"></a><span class='hs-comment'>--     shape = [30]</span>
+<a name="line-7440"></a><span class='hs-comment'>-- </span>
+<a name="line-7441"></a><span class='hs-comment'>-- then the final deserialized `SparseTensor` will be:</span>
+<a name="line-7442"></a><span class='hs-comment'>-- </span>
+<a name="line-7443"></a><span class='hs-comment'>--     index = [0  0]</span>
+<a name="line-7444"></a><span class='hs-comment'>--             [0 10]</span>
+<a name="line-7445"></a><span class='hs-comment'>--             [0 20]</span>
+<a name="line-7446"></a><span class='hs-comment'>--             [1  2]</span>
+<a name="line-7447"></a><span class='hs-comment'>--             [1 10]</span>
+<a name="line-7448"></a><span class='hs-comment'>--     values = [1, 2, 3, 4, 5]</span>
+<a name="line-7449"></a><span class='hs-comment'>--     shape = [2 50]</span>
+<a name="line-7450"></a><span class='hs-definition'>deserializeManySparse</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-7451"></a>                         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __serialized_sparse__: 2-D, The `N` serialized `SparseTensor` objects.</span>
+<a name="line-7452"></a>                                                              <span class='hs-comment'>-- Must have 3 columns.</span>
+<a name="line-7453"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>,</span>
+<a name="line-7454"></a>                             <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-7455"></a>                         <span class='hs-comment'>-- ^ (__sparse_indices__, __sparse_values__, __sparse_shape__)</span>
+<a name="line-7456"></a>                         <span class='hs-comment'>--</span>
+<a name="line-7457"></a>                         <span class='hs-comment'>-- * __sparse_indices__</span>
+<a name="line-7458"></a>                         <span class='hs-comment'>--</span>
+<a name="line-7459"></a>                         <span class='hs-comment'>-- * __sparse_values__</span>
+<a name="line-7460"></a>                         <span class='hs-comment'>--</span>
+<a name="line-7461"></a>                         <span class='hs-comment'>-- * __sparse_shape__</span>
+<a name="line-7462"></a><span class='hs-definition'>deserializeManySparse</span> <span class='hs-varid'>serialized_sparse</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7463"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DeserializeManySparse"</span>
+<a name="line-7464"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-7465"></a>        <span class='hs-varid'>serialized_sparse</span>
+<a name="line-7466"></a><span class='hs-comment'>{-
+<a name="line-7467"></a>attr {
+<a name="line-7468"></a>  description: "The `dtype` of the serialized `SparseTensor` objects."
+<a name="line-7469"></a>  name: "dtype"
+<a name="line-7470"></a>  type: "type"
+<a name="line-7471"></a>}
+<a name="line-7472"></a>input_arg {
+<a name="line-7473"></a>  description: "2-D, The `N` serialized `SparseTensor` objects.\nMust have 3 columns."
+<a name="line-7474"></a>  name: "serialized_sparse"
+<a name="line-7475"></a>  type: DT_STRING
+<a name="line-7476"></a>}
+<a name="line-7477"></a>output_arg { name: "sparse_indices" type: DT_INT64 }
+<a name="line-7478"></a>output_arg { name: "sparse_values" type_attr: "dtype" }
+<a name="line-7479"></a>output_arg { name: "sparse_shape" type: DT_INT64 }
+<a name="line-7480"></a>-}</span>
+<a name="line-7481"></a>
+<a name="line-7482"></a><a name="immutableConst"></a><span class='hs-comment'>-- | Returns immutable tensor from memory region.</span>
+<a name="line-7483"></a><span class='hs-comment'>--</span>
+<a name="line-7484"></a><span class='hs-comment'>-- The current implementation memmaps the tensor from a file.</span>
+<a name="line-7485"></a><span class='hs-definition'>immutableConst</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-7486"></a>                  <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __tensor__</span>
+<a name="line-7487"></a><span class='hs-definition'>immutableConst</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7488"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ImmutableConst"</span>
+<a name="line-7489"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-7490"></a>        
+<a name="line-7491"></a><span class='hs-comment'>{-
+<a name="line-7492"></a>attr {
+<a name="line-7493"></a>  description: "Type of the returned tensor."
+<a name="line-7494"></a>  name: "dtype"
+<a name="line-7495"></a>  type: "type"
+<a name="line-7496"></a>}
+<a name="line-7497"></a>attr {
+<a name="line-7498"></a>  description: "Shape of the returned tensor."
+<a name="line-7499"></a>  name: "shape"
+<a name="line-7500"></a>  type: "shape"
+<a name="line-7501"></a>}
+<a name="line-7502"></a>attr {
+<a name="line-7503"></a>  description: "Name of readonly memory region used by the tensor, see\nNewReadOnlyMemoryRegionFromFile in tensorflow::Env."
+<a name="line-7504"></a>  name: "memory_region_name"
+<a name="line-7505"></a>  type: "string"
+<a name="line-7506"></a>}
+<a name="line-7507"></a>output_arg { name: "tensor" type_attr: "dtype" }
+<a name="line-7508"></a>-}</span>
+<a name="line-7509"></a>
+<a name="line-7510"></a><a name="minimum"></a><span class='hs-comment'>-- | Returns the min of x and y (i.e. x &lt; y ? x : y) element-wise.</span>
+<a name="line-7511"></a><span class='hs-comment'>--</span>
+<a name="line-7512"></a><span class='hs-comment'>-- *NOTE*: `Minimum` supports broadcasting. More about broadcasting</span>
+<a name="line-7513"></a><span class='hs-comment'>-- [here](<a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)">http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)</a></span>
+<a name="line-7514"></a><span class='hs-definition'>minimum</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-7515"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-7516"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-7517"></a>                                                   <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-7518"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-7519"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-7520"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-7521"></a><span class='hs-definition'>minimum</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7522"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Minimum"</span>
+<a name="line-7523"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-7524"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-7525"></a><span class='hs-comment'>{-
+<a name="line-7526"></a>attr {
+<a name="line-7527"></a>  allowed_values {
+<a name="line-7528"></a>    list {
+<a name="line-7529"></a>      type: DT_HALF
+<a name="line-7530"></a>      type: DT_FLOAT
+<a name="line-7531"></a>      type: DT_DOUBLE
+<a name="line-7532"></a>      type: DT_INT32
+<a name="line-7533"></a>      type: DT_INT64
+<a name="line-7534"></a>    }
+<a name="line-7535"></a>  }
+<a name="line-7536"></a>  name: "T"
+<a name="line-7537"></a>  type: "type"
+<a name="line-7538"></a>}
+<a name="line-7539"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-7540"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-7541"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-7542"></a>-}</span>
+<a name="line-7543"></a>
+<a name="line-7544"></a><a name="initializeTableFromTextFile"></a><span class='hs-comment'>-- | Initializes a table from a text file.</span>
+<a name="line-7545"></a><span class='hs-comment'>--</span>
+<a name="line-7546"></a><span class='hs-comment'>-- It inserts one key-value pair into the table for each line of the file.</span>
+<a name="line-7547"></a><span class='hs-comment'>-- The key and value is extracted from the whole line content, elements from the</span>
+<a name="line-7548"></a><span class='hs-comment'>-- split line based on `delimiter` or the line number (starting from zero).</span>
+<a name="line-7549"></a><span class='hs-comment'>-- Where to extract the key and value from a line is specified by `key_index` and</span>
+<a name="line-7550"></a><span class='hs-comment'>-- `value_index`.</span>
+<a name="line-7551"></a><span class='hs-comment'>-- </span>
+<a name="line-7552"></a><span class='hs-comment'>-- - A value of -1 means use the line number(starting from zero), expects `int64`.</span>
+<a name="line-7553"></a><span class='hs-comment'>-- - A value of -2 means use the whole line content, expects `string`.</span>
+<a name="line-7554"></a><span class='hs-comment'>-- - A value &gt;= 0 means use the index (starting at zero) of the split line based</span>
+<a name="line-7555"></a><span class='hs-comment'>--   on `delimiter`.</span>
+<a name="line-7556"></a><span class='hs-definition'>initializeTableFromTextFile</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __key_index__: Column index in a line to get the table `key` values from.</span>
+<a name="line-7557"></a>                               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __value_index__: Column index that represents information of a line to get the table</span>
+<a name="line-7558"></a>                                                 <span class='hs-comment'>-- `value` values from.</span>
+<a name="line-7559"></a>                               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __table_handle__: Handle to a table which will be initialized.</span>
+<a name="line-7560"></a>                               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __filename__: Filename of a vocabulary text file.</span>
+<a name="line-7561"></a>                               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-7562"></a><span class='hs-definition'>initializeTableFromTextFile</span> <span class='hs-varid'>key_index</span> <span class='hs-varid'>value_index</span> <span class='hs-varid'>table_handle</span>
+<a name="line-7563"></a>                            <span class='hs-varid'>filename</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7564"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"InitializeTableFromTextFile"</span>
+<a name="line-7565"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"key_index"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>key_index</span>
+<a name="line-7566"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"value_index"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>value_index</span><span class='hs-layout'>)</span>
+<a name="line-7567"></a>        <span class='hs-varid'>table_handle</span> <span class='hs-varid'>filename</span>
+<a name="line-7568"></a><span class='hs-comment'>{-
+<a name="line-7569"></a>attr {
+<a name="line-7570"></a>  description: "Column index in a line to get the table `key` values from."
+<a name="line-7571"></a>  has_minimum: true
+<a name="line-7572"></a>  minimum: -2
+<a name="line-7573"></a>  name: "key_index"
+<a name="line-7574"></a>  type: "int"
+<a name="line-7575"></a>}
+<a name="line-7576"></a>attr {
+<a name="line-7577"></a>  description: "Column index that represents information of a line to get the table\n`value` values from."
+<a name="line-7578"></a>  has_minimum: true
+<a name="line-7579"></a>  minimum: -2
+<a name="line-7580"></a>  name: "value_index"
+<a name="line-7581"></a>  type: "int"
+<a name="line-7582"></a>}
+<a name="line-7583"></a>attr {
+<a name="line-7584"></a>  default_value { i: -1 }
+<a name="line-7585"></a>  description: "Number of elements of the file, use -1 if unknown."
+<a name="line-7586"></a>  has_minimum: true
+<a name="line-7587"></a>  minimum: -1
+<a name="line-7588"></a>  name: "vocab_size"
+<a name="line-7589"></a>  type: "int"
+<a name="line-7590"></a>}
+<a name="line-7591"></a>attr {
+<a name="line-7592"></a>  default_value { s: "\t" }
+<a name="line-7593"></a>  description: "Delimiter to separate fields in a line."
+<a name="line-7594"></a>  name: "delimiter"
+<a name="line-7595"></a>  type: "string"
+<a name="line-7596"></a>}
+<a name="line-7597"></a>input_arg {
+<a name="line-7598"></a>  description: "Handle to a table which will be initialized."
+<a name="line-7599"></a>  is_ref: true
+<a name="line-7600"></a>  name: "table_handle"
+<a name="line-7601"></a>  type: DT_STRING
+<a name="line-7602"></a>}
+<a name="line-7603"></a>input_arg {
+<a name="line-7604"></a>  description: "Filename of a vocabulary text file."
+<a name="line-7605"></a>  name: "filename"
+<a name="line-7606"></a>  type: DT_STRING
+<a name="line-7607"></a>}
+<a name="line-7608"></a>-}</span>
+<a name="line-7609"></a>
+<a name="line-7610"></a><a name="diagPart"></a><span class='hs-comment'>-- | Returns the diagonal part of the tensor.</span>
+<a name="line-7611"></a><span class='hs-comment'>--</span>
+<a name="line-7612"></a><span class='hs-comment'>-- This operation returns a tensor with the `diagonal` part</span>
+<a name="line-7613"></a><span class='hs-comment'>-- of the `input`. The `diagonal` part is computed as follows:</span>
+<a name="line-7614"></a><span class='hs-comment'>-- </span>
+<a name="line-7615"></a><span class='hs-comment'>-- Assume `input` has dimensions `[D1,..., Dk, D1,..., Dk]`, then the output is a</span>
+<a name="line-7616"></a><span class='hs-comment'>-- tensor of rank `k` with dimensions `[D1,..., Dk]` where:</span>
+<a name="line-7617"></a><span class='hs-comment'>-- </span>
+<a name="line-7618"></a><span class='hs-comment'>-- `diagonal[i1,..., ik] = input[i1, ..., ik, i1,..., ik]`.</span>
+<a name="line-7619"></a><span class='hs-comment'>-- </span>
+<a name="line-7620"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-7621"></a><span class='hs-comment'>-- </span>
+<a name="line-7622"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-7623"></a><span class='hs-comment'>-- # 'input' is [[1, 0, 0, 0]</span>
+<a name="line-7624"></a><span class='hs-comment'>--               [0, 2, 0, 0]</span>
+<a name="line-7625"></a><span class='hs-comment'>--               [0, 0, 3, 0]</span>
+<a name="line-7626"></a><span class='hs-comment'>--               [0, 0, 0, 4]]</span>
+<a name="line-7627"></a><span class='hs-comment'>-- </span>
+<a name="line-7628"></a><span class='hs-comment'>-- tf.diag_part(input) ==&gt; [1, 2, 3, 4]</span>
+<a name="line-7629"></a><span class='hs-comment'>-- ```</span>
+<a name="line-7630"></a><span class='hs-definition'>diagPart</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-7631"></a>                                                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-7632"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-7633"></a>                                                 <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-7634"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Rank k tensor where k is 2, 4, or 6.</span>
+<a name="line-7635"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __diagonal__: The extracted diagonal.</span>
+<a name="line-7636"></a><span class='hs-definition'>diagPart</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7637"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DiagPart"</span>
+<a name="line-7638"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-7639"></a>        <span class='hs-varid'>input</span>
+<a name="line-7640"></a><span class='hs-comment'>{-
+<a name="line-7641"></a>attr {
+<a name="line-7642"></a>  allowed_values {
+<a name="line-7643"></a>    list {
+<a name="line-7644"></a>      type: DT_FLOAT
+<a name="line-7645"></a>      type: DT_DOUBLE
+<a name="line-7646"></a>      type: DT_INT32
+<a name="line-7647"></a>      type: DT_INT64
+<a name="line-7648"></a>      type: DT_COMPLEX64
+<a name="line-7649"></a>      type: DT_COMPLEX128
+<a name="line-7650"></a>    }
+<a name="line-7651"></a>  }
+<a name="line-7652"></a>  name: "T"
+<a name="line-7653"></a>  type: "type"
+<a name="line-7654"></a>}
+<a name="line-7655"></a>input_arg {
+<a name="line-7656"></a>  description: "Rank k tensor where k is 2, 4, or 6."
+<a name="line-7657"></a>  name: "input"
+<a name="line-7658"></a>  type_attr: "T"
+<a name="line-7659"></a>}
+<a name="line-7660"></a>output_arg {
+<a name="line-7661"></a>  description: "The extracted diagonal."
+<a name="line-7662"></a>  name: "diagonal"
+<a name="line-7663"></a>  type_attr: "T"
+<a name="line-7664"></a>}
+<a name="line-7665"></a>-}</span>
+<a name="line-7666"></a>
+<a name="line-7667"></a><a name="log"></a><span class='hs-comment'>-- | Computes natural logarithm of x element-wise.</span>
+<a name="line-7668"></a><span class='hs-comment'>--</span>
+<a name="line-7669"></a><span class='hs-comment'>-- I.e., \\(y = \log_e x\\).</span>
+<a name="line-7670"></a><span class='hs-definition'>log</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-7671"></a>                                            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-7672"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-7673"></a>                                            <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-7674"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-7675"></a><span class='hs-definition'>log</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7676"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Log"</span>
+<a name="line-7677"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-7678"></a>        <span class='hs-varid'>x</span>
+<a name="line-7679"></a><span class='hs-comment'>{-
+<a name="line-7680"></a>attr {
+<a name="line-7681"></a>  allowed_values {
+<a name="line-7682"></a>    list {
+<a name="line-7683"></a>      type: DT_HALF
+<a name="line-7684"></a>      type: DT_FLOAT
+<a name="line-7685"></a>      type: DT_DOUBLE
+<a name="line-7686"></a>      type: DT_COMPLEX64
+<a name="line-7687"></a>      type: DT_COMPLEX128
+<a name="line-7688"></a>    }
+<a name="line-7689"></a>  }
+<a name="line-7690"></a>  name: "T"
+<a name="line-7691"></a>  type: "type"
+<a name="line-7692"></a>}
+<a name="line-7693"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-7694"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-7695"></a>-}</span>
+<a name="line-7696"></a>
+<a name="line-7697"></a><a name="tensorArrayScatter"></a><span class='hs-comment'>-- | Scatter the data from the input value into specific TensorArray elements.</span>
+<a name="line-7698"></a><span class='hs-comment'>--</span>
+<a name="line-7699"></a><span class='hs-comment'>-- `indices` must be a vector, its length must match the first dim of `value`.</span>
+<a name="line-7700"></a><span class='hs-definition'>tensorArrayScatter</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-7701"></a>                      <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a TensorArray.</span>
+<a name="line-7702"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __indices__: The locations at which to write the tensor elements.</span>
+<a name="line-7703"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: The concatenated tensor to write to the TensorArray.</span>
+<a name="line-7704"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __flow_in__: A float scalar that enforces proper chaining of operations.</span>
+<a name="line-7705"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __flow_out__: A float scalar that enforces proper chaining of operations.</span>
+<a name="line-7706"></a><span class='hs-definition'>tensorArrayScatter</span> <span class='hs-varid'>handle</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>value</span> <span class='hs-varid'>flow_in</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7707"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TensorArrayScatter"</span>
+<a name="line-7708"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-7709"></a>        <span class='hs-varid'>handle</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>value</span> <span class='hs-varid'>flow_in</span>
+<a name="line-7710"></a><span class='hs-comment'>{-
+<a name="line-7711"></a>attr { name: "T" type: "type" }
+<a name="line-7712"></a>input_arg {
+<a name="line-7713"></a>  description: "The handle to a TensorArray."
+<a name="line-7714"></a>  is_ref: true
+<a name="line-7715"></a>  name: "handle"
+<a name="line-7716"></a>  type: DT_STRING
+<a name="line-7717"></a>}
+<a name="line-7718"></a>input_arg {
+<a name="line-7719"></a>  description: "The locations at which to write the tensor elements."
+<a name="line-7720"></a>  name: "indices"
+<a name="line-7721"></a>  type: DT_INT32
+<a name="line-7722"></a>}
+<a name="line-7723"></a>input_arg {
+<a name="line-7724"></a>  description: "The concatenated tensor to write to the TensorArray."
+<a name="line-7725"></a>  name: "value"
+<a name="line-7726"></a>  type_attr: "T"
+<a name="line-7727"></a>}
+<a name="line-7728"></a>input_arg {
+<a name="line-7729"></a>  description: "A float scalar that enforces proper chaining of operations."
+<a name="line-7730"></a>  name: "flow_in"
+<a name="line-7731"></a>  type: DT_FLOAT
+<a name="line-7732"></a>}
+<a name="line-7733"></a>output_arg {
+<a name="line-7734"></a>  description: "A float scalar that enforces proper chaining of operations."
+<a name="line-7735"></a>  name: "flow_out"
+<a name="line-7736"></a>  type: DT_FLOAT
+<a name="line-7737"></a>}
+<a name="line-7738"></a>-}</span>
+<a name="line-7739"></a>
+<a name="line-7740"></a><a name="rank"></a><span class='hs-comment'>-- | Returns the rank of a tensor.</span>
+<a name="line-7741"></a><span class='hs-comment'>--</span>
+<a name="line-7742"></a><span class='hs-comment'>-- This operation returns an integer representing the rank of `input`.</span>
+<a name="line-7743"></a><span class='hs-comment'>-- </span>
+<a name="line-7744"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-7745"></a><span class='hs-comment'>-- </span>
+<a name="line-7746"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-7747"></a><span class='hs-comment'>-- # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]</span>
+<a name="line-7748"></a><span class='hs-comment'>-- # shape of tensor 't' is [2, 2, 3]</span>
+<a name="line-7749"></a><span class='hs-comment'>-- rank(t) ==&gt; 3</span>
+<a name="line-7750"></a><span class='hs-comment'>-- ```</span>
+<a name="line-7751"></a><span class='hs-comment'>-- </span>
+<a name="line-7752"></a><span class='hs-comment'>-- **Note**: The rank of a tensor is not the same as the rank of a matrix. The rank</span>
+<a name="line-7753"></a><span class='hs-comment'>-- of a tensor is the number of indices required to uniquely select each element</span>
+<a name="line-7754"></a><span class='hs-comment'>-- of the tensor. Rank is also known as "order", "degree", or "ndims."</span>
+<a name="line-7755"></a><span class='hs-definition'>rank</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-7756"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-7757"></a><span class='hs-definition'>rank</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7758"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Rank"</span>
+<a name="line-7759"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-7760"></a>        <span class='hs-varid'>input</span>
+<a name="line-7761"></a><span class='hs-comment'>{-
+<a name="line-7762"></a>attr { name: "T" type: "type" }
+<a name="line-7763"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-7764"></a>output_arg { name: "output" type: DT_INT32 }
+<a name="line-7765"></a>-}</span>
+<a name="line-7766"></a>
+<a name="line-7767"></a><span class='hs-comment'>-- | Return a tensor with the same shape and contents as the input tensor or value.</span>
+<a name="line-7768"></a>
+<a name="line-7769"></a><a name="identity"></a><span class='hs-definition'>identity</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-7770"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-7771"></a><span class='hs-definition'>identity</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7772"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Identity"</span>
+<a name="line-7773"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-7774"></a>        <span class='hs-varid'>input</span>
+<a name="line-7775"></a><span class='hs-comment'>{-
+<a name="line-7776"></a>attr { name: "T" type: "type" }
+<a name="line-7777"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-7778"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-7779"></a>-}</span>
+<a name="line-7780"></a>
+<a name="line-7781"></a><a name="adjustContrastv2"></a><span class='hs-comment'>-- | Adjust the contrast of one or more images.</span>
+<a name="line-7782"></a><span class='hs-comment'>--</span>
+<a name="line-7783"></a><span class='hs-comment'>-- `images` is a tensor of at least 3 dimensions.  The last 3 dimensions are</span>
+<a name="line-7784"></a><span class='hs-comment'>-- interpreted as `[height, width, channels]`.  The other dimensions only</span>
+<a name="line-7785"></a><span class='hs-comment'>-- represent a collection of images, such as `[batch, height, width, channels].`</span>
+<a name="line-7786"></a><span class='hs-comment'>-- </span>
+<a name="line-7787"></a><span class='hs-comment'>-- Contrast is adjusted independently for each channel of each image.</span>
+<a name="line-7788"></a><span class='hs-comment'>-- </span>
+<a name="line-7789"></a><span class='hs-comment'>-- For each channel, the Op first computes the mean of the image pixels in the</span>
+<a name="line-7790"></a><span class='hs-comment'>-- channel and then adjusts each component of each pixel to</span>
+<a name="line-7791"></a><span class='hs-comment'>-- `(x - mean) * contrast_factor + mean`.</span>
+<a name="line-7792"></a><span class='hs-definition'>adjustContrastv2</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __images__: Images to adjust.  At least 3-D.</span>
+<a name="line-7793"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __contrast_factor__: A float multiplier for adjusting contrast.</span>
+<a name="line-7794"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __output__: The contrast-adjusted image or images.</span>
+<a name="line-7795"></a><span class='hs-definition'>adjustContrastv2</span> <span class='hs-varid'>images</span> <span class='hs-varid'>contrast_factor</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7796"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"AdjustContrastv2"</span><span class='hs-layout'>)</span>
+<a name="line-7797"></a>        <span class='hs-varid'>images</span> <span class='hs-varid'>contrast_factor</span>
+<a name="line-7798"></a><span class='hs-comment'>{-
+<a name="line-7799"></a>input_arg {
+<a name="line-7800"></a>  description: "Images to adjust.  At least 3-D."
+<a name="line-7801"></a>  name: "images"
+<a name="line-7802"></a>  type: DT_FLOAT
+<a name="line-7803"></a>}
+<a name="line-7804"></a>input_arg {
+<a name="line-7805"></a>  description: "A float multiplier for adjusting contrast."
+<a name="line-7806"></a>  name: "contrast_factor"
+<a name="line-7807"></a>  type: DT_FLOAT
+<a name="line-7808"></a>}
+<a name="line-7809"></a>output_arg {
+<a name="line-7810"></a>  description: "The contrast-adjusted image or images."
+<a name="line-7811"></a>  name: "output"
+<a name="line-7812"></a>  type: DT_FLOAT
+<a name="line-7813"></a>}
+<a name="line-7814"></a>-}</span>
+<a name="line-7815"></a>
+<a name="line-7816"></a><a name="sparseApplyProximalAdagrad"></a><span class='hs-comment'>-- | Sparse update entries in '*var' and '*accum' according to FOBOS algorithm.</span>
+<a name="line-7817"></a><span class='hs-comment'>--</span>
+<a name="line-7818"></a><span class='hs-comment'>-- That is for rows we have grad for, we update var and accum as follows:</span>
+<a name="line-7819"></a><span class='hs-comment'>-- accum += grad * grad</span>
+<a name="line-7820"></a><span class='hs-comment'>-- prox_v = var</span>
+<a name="line-7821"></a><span class='hs-comment'>-- prox_v -= lr * grad * (1 / sqrt(accum))</span>
+<a name="line-7822"></a><span class='hs-comment'>-- var = sign(prox_v)/(1+lr*l2) * max{|prox_v|-lr*l1,0}</span>
+<a name="line-7823"></a><span class='hs-definition'>sparseApplyProximalAdagrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>t</span>
+<a name="line-7824"></a>                              <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-7825"></a>                                          <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-7826"></a>                                                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-7827"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-7828"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-7829"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-7830"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-7831"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-7832"></a>                                                  <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-7833"></a>                                          <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-7834"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-7835"></a>                              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-7836"></a>                              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __accum__: Should be from a Variable().</span>
+<a name="line-7837"></a>                              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr__: Learning rate. Must be a scalar.</span>
+<a name="line-7838"></a>                              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l1__: L1 regularization. Must be a scalar.</span>
+<a name="line-7839"></a>                              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l2__: L2 regularization. Must be a scalar.</span>
+<a name="line-7840"></a>                              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: The gradient.</span>
+<a name="line-7841"></a>                              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __indices__: A vector of indices into the first dimension of var and accum.</span>
+<a name="line-7842"></a>                              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-7843"></a><span class='hs-definition'>sparseApplyProximalAdagrad</span> <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>l1</span> <span class='hs-varid'>l2</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>indices</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7844"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseApplyProximalAdagrad"</span>
+<a name="line-7845"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-7846"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-7847"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>l1</span> <span class='hs-varid'>l2</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>indices</span>
+<a name="line-7848"></a><span class='hs-comment'>{-
+<a name="line-7849"></a>attr {
+<a name="line-7850"></a>  allowed_values {
+<a name="line-7851"></a>    list {
+<a name="line-7852"></a>      type: DT_FLOAT
+<a name="line-7853"></a>      type: DT_DOUBLE
+<a name="line-7854"></a>      type: DT_INT64
+<a name="line-7855"></a>      type: DT_INT32
+<a name="line-7856"></a>      type: DT_UINT8
+<a name="line-7857"></a>      type: DT_UINT16
+<a name="line-7858"></a>      type: DT_INT16
+<a name="line-7859"></a>      type: DT_INT8
+<a name="line-7860"></a>      type: DT_COMPLEX64
+<a name="line-7861"></a>      type: DT_COMPLEX128
+<a name="line-7862"></a>      type: DT_QINT8
+<a name="line-7863"></a>      type: DT_QUINT8
+<a name="line-7864"></a>      type: DT_QINT32
+<a name="line-7865"></a>      type: DT_HALF
+<a name="line-7866"></a>    }
+<a name="line-7867"></a>  }
+<a name="line-7868"></a>  name: "T"
+<a name="line-7869"></a>  type: "type"
+<a name="line-7870"></a>}
+<a name="line-7871"></a>attr {
+<a name="line-7872"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-7873"></a>  name: "Tindices"
+<a name="line-7874"></a>  type: "type"
+<a name="line-7875"></a>}
+<a name="line-7876"></a>attr {
+<a name="line-7877"></a>  default_value { b: false }
+<a name="line-7878"></a>  description: "If True, updating of the var and accum tensors will be protected by\na lock; otherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-7879"></a>  name: "use_locking"
+<a name="line-7880"></a>  type: "bool"
+<a name="line-7881"></a>}
+<a name="line-7882"></a>input_arg {
+<a name="line-7883"></a>  description: "Should be from a Variable()."
+<a name="line-7884"></a>  is_ref: true
+<a name="line-7885"></a>  name: "var"
+<a name="line-7886"></a>  type_attr: "T"
+<a name="line-7887"></a>}
+<a name="line-7888"></a>input_arg {
+<a name="line-7889"></a>  description: "Should be from a Variable()."
+<a name="line-7890"></a>  is_ref: true
+<a name="line-7891"></a>  name: "accum"
+<a name="line-7892"></a>  type_attr: "T"
+<a name="line-7893"></a>}
+<a name="line-7894"></a>input_arg {
+<a name="line-7895"></a>  description: "Learning rate. Must be a scalar."
+<a name="line-7896"></a>  name: "lr"
+<a name="line-7897"></a>  type_attr: "T"
+<a name="line-7898"></a>}
+<a name="line-7899"></a>input_arg {
+<a name="line-7900"></a>  description: "L1 regularization. Must be a scalar."
+<a name="line-7901"></a>  name: "l1"
+<a name="line-7902"></a>  type_attr: "T"
+<a name="line-7903"></a>}
+<a name="line-7904"></a>input_arg {
+<a name="line-7905"></a>  description: "L2 regularization. Must be a scalar."
+<a name="line-7906"></a>  name: "l2"
+<a name="line-7907"></a>  type_attr: "T"
+<a name="line-7908"></a>}
+<a name="line-7909"></a>input_arg {
+<a name="line-7910"></a>  description: "The gradient." name: "grad" type_attr: "T"
+<a name="line-7911"></a>}
+<a name="line-7912"></a>input_arg {
+<a name="line-7913"></a>  description: "A vector of indices into the first dimension of var and accum."
+<a name="line-7914"></a>  name: "indices"
+<a name="line-7915"></a>  type_attr: "Tindices"
+<a name="line-7916"></a>}
+<a name="line-7917"></a>output_arg {
+<a name="line-7918"></a>  description: "Same as \"var\"."
+<a name="line-7919"></a>  is_ref: true
+<a name="line-7920"></a>  name: "out"
+<a name="line-7921"></a>  type_attr: "T"
+<a name="line-7922"></a>}
+<a name="line-7923"></a>-}</span>
+<a name="line-7924"></a>
+<a name="line-7925"></a><a name="gather"></a><span class='hs-comment'>-- | Gather slices from `params` according to `indices`.</span>
+<a name="line-7926"></a><span class='hs-comment'>--</span>
+<a name="line-7927"></a><span class='hs-comment'>-- `indices` must be an integer tensor of any dimension (usually 0-D or 1-D).</span>
+<a name="line-7928"></a><span class='hs-comment'>-- Produces an output tensor with shape `indices.shape + params.shape[1:]` where:</span>
+<a name="line-7929"></a><span class='hs-comment'>-- </span>
+<a name="line-7930"></a><span class='hs-comment'>--     # Scalar indices</span>
+<a name="line-7931"></a><span class='hs-comment'>--     output[:, ..., :] = params[indices, :, ... :]</span>
+<a name="line-7932"></a><span class='hs-comment'>-- </span>
+<a name="line-7933"></a><span class='hs-comment'>--     # Vector indices</span>
+<a name="line-7934"></a><span class='hs-comment'>--     output[i, :, ..., :] = params[indices[i], :, ... :]</span>
+<a name="line-7935"></a><span class='hs-comment'>-- </span>
+<a name="line-7936"></a><span class='hs-comment'>--     # Higher rank indices</span>
+<a name="line-7937"></a><span class='hs-comment'>--     output[i, ..., j, :, ... :] = params[indices[i, ..., j], :, ..., :]</span>
+<a name="line-7938"></a><span class='hs-comment'>-- </span>
+<a name="line-7939"></a><span class='hs-comment'>-- If `indices` is a permutation and `len(indices) == params.shape[0]` then</span>
+<a name="line-7940"></a><span class='hs-comment'>-- this operation will permute `params` accordingly.</span>
+<a name="line-7941"></a><span class='hs-comment'>-- </span>
+<a name="line-7942"></a><span class='hs-comment'>-- &lt;div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;"&gt;</span>
+<a name="line-7943"></a><span class='hs-comment'>-- &lt;img style="width:100%" src="../../images/Gather.png" alt&gt;</span>
+<a name="line-7944"></a><span class='hs-comment'>-- &lt;/div&gt;</span>
+<a name="line-7945"></a><span class='hs-definition'>gather</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tindices</span> <span class='hs-varid'>tparams</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-7946"></a>                                           <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-7947"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-7948"></a>                                           <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tparams</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-7949"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>tparams</span> <span class='hs-comment'>-- ^ __params__</span>
+<a name="line-7950"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __indices__</span>
+<a name="line-7951"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>tparams</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-7952"></a><span class='hs-definition'>gather</span> <span class='hs-varid'>params</span> <span class='hs-varid'>indices</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7953"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Gather"</span>
+<a name="line-7954"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span>
+<a name="line-7955"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tparams"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tparams</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-7956"></a>        <span class='hs-varid'>params</span> <span class='hs-varid'>indices</span>
+<a name="line-7957"></a><span class='hs-comment'>{-
+<a name="line-7958"></a>attr {
+<a name="line-7959"></a>  default_value { b: true } name: "validate_indices" type: "bool"
+<a name="line-7960"></a>}
+<a name="line-7961"></a>attr { name: "Tparams" type: "type" }
+<a name="line-7962"></a>attr {
+<a name="line-7963"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-7964"></a>  name: "Tindices"
+<a name="line-7965"></a>  type: "type"
+<a name="line-7966"></a>}
+<a name="line-7967"></a>input_arg { name: "params" type_attr: "Tparams" }
+<a name="line-7968"></a>input_arg { name: "indices" type_attr: "Tindices" }
+<a name="line-7969"></a>output_arg { name: "output" type_attr: "Tparams" }
+<a name="line-7970"></a>-}</span>
+<a name="line-7971"></a>
+<a name="line-7972"></a><a name="isVariableInitialized"></a><span class='hs-comment'>-- | Checks whether a tensor has been initialized.</span>
+<a name="line-7973"></a><span class='hs-comment'>--</span>
+<a name="line-7974"></a><span class='hs-comment'>-- Outputs boolean scalar indicating whether the tensor has been initialized.</span>
+<a name="line-7975"></a><span class='hs-definition'>isVariableInitialized</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-7976"></a>                         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __ref__: Should be from a `Variable` node. May be uninitialized.</span>
+<a name="line-7977"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __is_initialized__</span>
+<a name="line-7978"></a><span class='hs-definition'>isVariableInitialized</span> <span class='hs-varid'>ref</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-7979"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"IsVariableInitialized"</span>
+<a name="line-7980"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-7981"></a>        <span class='hs-varid'>ref</span>
+<a name="line-7982"></a><span class='hs-comment'>{-
+<a name="line-7983"></a>attr {
+<a name="line-7984"></a>  description: "The type of elements in the variable tensor."
+<a name="line-7985"></a>  name: "dtype"
+<a name="line-7986"></a>  type: "type"
+<a name="line-7987"></a>}
+<a name="line-7988"></a>input_arg {
+<a name="line-7989"></a>  description: "Should be from a `Variable` node. May be uninitialized."
+<a name="line-7990"></a>  is_ref: true
+<a name="line-7991"></a>  name: "ref"
+<a name="line-7992"></a>  type_attr: "dtype"
+<a name="line-7993"></a>}
+<a name="line-7994"></a>output_arg { name: "is_initialized" type: DT_BOOL }
+<a name="line-7995"></a>-}</span>
+<a name="line-7996"></a>
+<a name="line-7997"></a><span class='hs-comment'>-- | Concatenates tensors along one dimension.</span>
+<a name="line-7998"></a>
+<a name="line-7999"></a><a name="concat"></a><span class='hs-definition'>concat</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-8000"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __concat_dim__: 0-D.  The dimension along which to concatenate.  Must be in the</span>
+<a name="line-8001"></a>                                   <span class='hs-comment'>-- range [0, rank(values)).</span>
+<a name="line-8002"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __values__: The `N` Tensors to concatenate. Their ranks and types must match,</span>
+<a name="line-8003"></a>                           <span class='hs-comment'>-- and their sizes must match in all dimensions except `concat_dim`.</span>
+<a name="line-8004"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: A `Tensor` with the concatenation of values stacked along the</span>
+<a name="line-8005"></a>          <span class='hs-comment'>-- `concat_dim` dimension.  This tensor's shape matches that of `values` except</span>
+<a name="line-8006"></a>          <span class='hs-comment'>-- in `concat_dim` where it has the sum of the sizes.</span>
+<a name="line-8007"></a><span class='hs-definition'>concat</span> <span class='hs-varid'>concat_dim</span> <span class='hs-varid'>values</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"N"</span><span class='hs-layout'>,</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"values"</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>values</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8008"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Concat"</span>
+<a name="line-8009"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-8010"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"N"</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>values</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-8011"></a>        <span class='hs-varid'>concat_dim</span> <span class='hs-varid'>values</span>
+<a name="line-8012"></a><span class='hs-comment'>{-
+<a name="line-8013"></a>attr { has_minimum: true minimum: 2 name: "N" type: "int" }
+<a name="line-8014"></a>attr { name: "T" type: "type" }
+<a name="line-8015"></a>input_arg {
+<a name="line-8016"></a>  description: "0-D.  The dimension along which to concatenate.  Must be in the\nrange [0, rank(values))."
+<a name="line-8017"></a>  name: "concat_dim"
+<a name="line-8018"></a>  type: DT_INT32
+<a name="line-8019"></a>}
+<a name="line-8020"></a>input_arg {
+<a name="line-8021"></a>  description: "The `N` Tensors to concatenate. Their ranks and types must match,\nand their sizes must match in all dimensions except `concat_dim`."
+<a name="line-8022"></a>  name: "values"
+<a name="line-8023"></a>  number_attr: "N"
+<a name="line-8024"></a>  type_attr: "T"
+<a name="line-8025"></a>}
+<a name="line-8026"></a>output_arg {
+<a name="line-8027"></a>  description: "A `Tensor` with the concatenation of values stacked along the\n`concat_dim` dimension.  This tensor\'s shape matches that of `values` except\nin `concat_dim` where it has the sum of the sizes."
+<a name="line-8028"></a>  name: "output"
+<a name="line-8029"></a>  type_attr: "T"
+<a name="line-8030"></a>}
+<a name="line-8031"></a>-}</span>
+<a name="line-8032"></a>
+<a name="line-8033"></a><a name="randomUniformInt"></a><span class='hs-comment'>-- | Outputs random integers from a uniform distribution.</span>
+<a name="line-8034"></a><span class='hs-comment'>--</span>
+<a name="line-8035"></a><span class='hs-comment'>-- The generated values are uniform integers in the range `[minval, maxval)`.</span>
+<a name="line-8036"></a><span class='hs-comment'>-- The lower bound `minval` is included in the range, while the upper bound</span>
+<a name="line-8037"></a><span class='hs-comment'>-- `maxval` is excluded.</span>
+<a name="line-8038"></a><span class='hs-comment'>-- </span>
+<a name="line-8039"></a><span class='hs-comment'>-- The random integers are slightly biased unless `maxval - minval` is an exact</span>
+<a name="line-8040"></a><span class='hs-comment'>-- power of two.  The bias is small for values of `maxval - minval` significantly</span>
+<a name="line-8041"></a><span class='hs-comment'>-- smaller than the range of the output (either `2^32` or `2^64`).</span>
+<a name="line-8042"></a><span class='hs-definition'>randomUniformInt</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tout</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-8043"></a>                                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-8044"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-8045"></a>                                              <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tout</span><span class='hs-layout'>,</span>
+<a name="line-8046"></a>                                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-8047"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tout</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-8048"></a>                    <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __shape__: The shape of the output tensor.</span>
+<a name="line-8049"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tout</span> <span class='hs-comment'>-- ^ __minval__: 0-D.  Inclusive lower bound on the generated integers.</span>
+<a name="line-8050"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>tout</span> <span class='hs-comment'>-- ^ __maxval__: 0-D.  Exclusive upper bound on the generated integers.</span>
+<a name="line-8051"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>tout</span> <span class='hs-comment'>-- ^ __output__: A tensor of the specified shape filled with uniform random integers.</span>
+<a name="line-8052"></a><span class='hs-definition'>randomUniformInt</span> <span class='hs-varid'>shape</span> <span class='hs-varid'>minval</span> <span class='hs-varid'>maxval</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8053"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RandomUniformInt"</span>
+<a name="line-8054"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-8055"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tout"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tout</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-8056"></a>        <span class='hs-varid'>shape</span> <span class='hs-varid'>minval</span> <span class='hs-varid'>maxval</span>
+<a name="line-8057"></a><span class='hs-comment'>{-
+<a name="line-8058"></a>attr {
+<a name="line-8059"></a>  default_value { i: 0 }
+<a name="line-8060"></a>  description: "If either `seed` or `seed2` are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-8061"></a>  name: "seed"
+<a name="line-8062"></a>  type: "int"
+<a name="line-8063"></a>}
+<a name="line-8064"></a>attr {
+<a name="line-8065"></a>  default_value { i: 0 }
+<a name="line-8066"></a>  description: "A second seed to avoid seed collision."
+<a name="line-8067"></a>  name: "seed2"
+<a name="line-8068"></a>  type: "int"
+<a name="line-8069"></a>}
+<a name="line-8070"></a>attr {
+<a name="line-8071"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-8072"></a>  name: "Tout"
+<a name="line-8073"></a>  type: "type"
+<a name="line-8074"></a>}
+<a name="line-8075"></a>attr {
+<a name="line-8076"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-8077"></a>  name: "T"
+<a name="line-8078"></a>  type: "type"
+<a name="line-8079"></a>}
+<a name="line-8080"></a>input_arg {
+<a name="line-8081"></a>  description: "The shape of the output tensor."
+<a name="line-8082"></a>  name: "shape"
+<a name="line-8083"></a>  type_attr: "T"
+<a name="line-8084"></a>}
+<a name="line-8085"></a>input_arg {
+<a name="line-8086"></a>  description: "0-D.  Inclusive lower bound on the generated integers."
+<a name="line-8087"></a>  name: "minval"
+<a name="line-8088"></a>  type_attr: "Tout"
+<a name="line-8089"></a>}
+<a name="line-8090"></a>input_arg {
+<a name="line-8091"></a>  description: "0-D.  Exclusive upper bound on the generated integers."
+<a name="line-8092"></a>  name: "maxval"
+<a name="line-8093"></a>  type_attr: "Tout"
+<a name="line-8094"></a>}
+<a name="line-8095"></a>output_arg {
+<a name="line-8096"></a>  description: "A tensor of the specified shape filled with uniform random integers."
+<a name="line-8097"></a>  name: "output"
+<a name="line-8098"></a>  type_attr: "Tout"
+<a name="line-8099"></a>}
+<a name="line-8100"></a>-}</span>
+<a name="line-8101"></a>
+<a name="line-8102"></a><a name="stopGradient"></a><span class='hs-comment'>-- | Stops gradient computation.</span>
+<a name="line-8103"></a><span class='hs-comment'>--</span>
+<a name="line-8104"></a><span class='hs-comment'>-- When executed in a graph, this op outputs its input tensor as-is.</span>
+<a name="line-8105"></a><span class='hs-comment'>-- </span>
+<a name="line-8106"></a><span class='hs-comment'>-- When building ops to compute gradients, this op prevents the contribution of</span>
+<a name="line-8107"></a><span class='hs-comment'>-- its inputs to be taken into account.  Normally, the gradient generator adds ops</span>
+<a name="line-8108"></a><span class='hs-comment'>-- to a graph to compute the derivatives of a specified 'loss' by recursively</span>
+<a name="line-8109"></a><span class='hs-comment'>-- finding out inputs that contributed to its computation.  If you insert this op</span>
+<a name="line-8110"></a><span class='hs-comment'>-- in the graph it inputs are masked from the gradient generator.  They are not</span>
+<a name="line-8111"></a><span class='hs-comment'>-- taken into account for computing gradients.</span>
+<a name="line-8112"></a><span class='hs-comment'>-- </span>
+<a name="line-8113"></a><span class='hs-comment'>-- This is useful any time you want to compute a value with TensorFlow but need</span>
+<a name="line-8114"></a><span class='hs-comment'>-- to pretend that the value was a constant. Some examples include:</span>
+<a name="line-8115"></a><span class='hs-comment'>-- </span>
+<a name="line-8116"></a><span class='hs-comment'>-- *  The *EM* algorithm where the *M-step* should not involve backpropagation</span>
+<a name="line-8117"></a><span class='hs-comment'>--    through the output of the *E-step*.</span>
+<a name="line-8118"></a><span class='hs-comment'>-- *  Contrastive divergence training of Boltzmann machines where, when</span>
+<a name="line-8119"></a><span class='hs-comment'>--    differentiating the energy function, the training must not backpropagate</span>
+<a name="line-8120"></a><span class='hs-comment'>--    through the graph that generated the samples from the model.</span>
+<a name="line-8121"></a><span class='hs-comment'>-- *  Adversarial training, where no backprop should happen through the adversarial</span>
+<a name="line-8122"></a><span class='hs-comment'>--    example generation process.</span>
+<a name="line-8123"></a><span class='hs-definition'>stopGradient</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-8124"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-8125"></a><span class='hs-definition'>stopGradient</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8126"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"StopGradient"</span>
+<a name="line-8127"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-8128"></a>        <span class='hs-varid'>input</span>
+<a name="line-8129"></a><span class='hs-comment'>{-
+<a name="line-8130"></a>attr { name: "T" type: "type" }
+<a name="line-8131"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-8132"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-8133"></a>-}</span>
+<a name="line-8134"></a>
+<a name="line-8135"></a><a name="avgPool"></a><span class='hs-comment'>-- | Performs average pooling on the input.</span>
+<a name="line-8136"></a><span class='hs-comment'>--</span>
+<a name="line-8137"></a><span class='hs-comment'>-- Each entry in `output` is the mean of the corresponding size `ksize`</span>
+<a name="line-8138"></a><span class='hs-comment'>-- window in `value`.</span>
+<a name="line-8139"></a><span class='hs-definition'>avgPool</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-8140"></a>                                                <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-8141"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: 4-D with shape `[batch, height, width, channels]`.</span>
+<a name="line-8142"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The average pooled output tensor.</span>
+<a name="line-8143"></a><span class='hs-definition'>avgPool</span> <span class='hs-varid'>value</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8144"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"AvgPool"</span>
+<a name="line-8145"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-8146"></a>        <span class='hs-varid'>value</span>
+<a name="line-8147"></a><span class='hs-comment'>{-
+<a name="line-8148"></a>attr {
+<a name="line-8149"></a>  description: "The size of the sliding window for each dimension of `value`."
+<a name="line-8150"></a>  has_minimum: true
+<a name="line-8151"></a>  minimum: 4
+<a name="line-8152"></a>  name: "ksize"
+<a name="line-8153"></a>  type: "list(int)"
+<a name="line-8154"></a>}
+<a name="line-8155"></a>attr {
+<a name="line-8156"></a>  description: "The stride of the sliding window for each dimension of `value`."
+<a name="line-8157"></a>  has_minimum: true
+<a name="line-8158"></a>  minimum: 4
+<a name="line-8159"></a>  name: "strides"
+<a name="line-8160"></a>  type: "list(int)"
+<a name="line-8161"></a>}
+<a name="line-8162"></a>attr {
+<a name="line-8163"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-8164"></a>  description: "The type of padding algorithm to use."
+<a name="line-8165"></a>  name: "padding"
+<a name="line-8166"></a>  type: "string"
+<a name="line-8167"></a>}
+<a name="line-8168"></a>attr {
+<a name="line-8169"></a>  allowed_values { list { s: "NHWC" s: "NCHW" } }
+<a name="line-8170"></a>  default_value { s: "NHWC" }
+<a name="line-8171"></a>  description: "Specify the data format of the input and output data. With the\ndefault format \"NHWC\", the data is stored in the order of:\n    [batch, in_height, in_width, in_channels].\nAlternatively, the format could be \"NCHW\", the data storage order of:\n    [batch, in_channels, in_height, in_width]."
+<a name="line-8172"></a>  name: "data_format"
+<a name="line-8173"></a>  type: "string"
+<a name="line-8174"></a>}
+<a name="line-8175"></a>attr {
+<a name="line-8176"></a>  allowed_values {
+<a name="line-8177"></a>    list { type: DT_FLOAT type: DT_HALF type: DT_DOUBLE }
+<a name="line-8178"></a>  }
+<a name="line-8179"></a>  name: "T"
+<a name="line-8180"></a>  type: "type"
+<a name="line-8181"></a>}
+<a name="line-8182"></a>input_arg {
+<a name="line-8183"></a>  description: "4-D with shape `[batch, height, width, channels]`."
+<a name="line-8184"></a>  name: "value"
+<a name="line-8185"></a>  type_attr: "T"
+<a name="line-8186"></a>}
+<a name="line-8187"></a>output_arg {
+<a name="line-8188"></a>  description: "The average pooled output tensor."
+<a name="line-8189"></a>  name: "output"
+<a name="line-8190"></a>  type_attr: "T"
+<a name="line-8191"></a>}
+<a name="line-8192"></a>-}</span>
+<a name="line-8193"></a>
+<a name="line-8194"></a><a name="wholeFileReader"></a><span class='hs-comment'>-- | A Reader that outputs the entire contents of a file as a value.</span>
+<a name="line-8195"></a><span class='hs-comment'>--</span>
+<a name="line-8196"></a><span class='hs-comment'>-- To use, enqueue filenames in a Queue.  The output of ReaderRead will</span>
+<a name="line-8197"></a><span class='hs-comment'>-- be a filename (key) and the contents of that file (value).</span>
+<a name="line-8198"></a><span class='hs-definition'>wholeFileReader</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __reader_handle__: The handle to reference the Reader.</span>
+<a name="line-8199"></a><span class='hs-definition'>wholeFileReader</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8200"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"WholeFileReader"</span><span class='hs-layout'>)</span>
+<a name="line-8201"></a>        
+<a name="line-8202"></a><span class='hs-comment'>{-
+<a name="line-8203"></a>attr {
+<a name="line-8204"></a>  default_value { s: "" }
+<a name="line-8205"></a>  description: "If non-empty, this reader is placed in the given container.\nOtherwise, a default container is used."
+<a name="line-8206"></a>  name: "container"
+<a name="line-8207"></a>  type: "string"
+<a name="line-8208"></a>}
+<a name="line-8209"></a>attr {
+<a name="line-8210"></a>  default_value { s: "" }
+<a name="line-8211"></a>  description: "If non-empty, this reader is named in the given bucket\nwith this shared_name. Otherwise, the node name is used instead."
+<a name="line-8212"></a>  name: "shared_name"
+<a name="line-8213"></a>  type: "string"
+<a name="line-8214"></a>}
+<a name="line-8215"></a>output_arg {
+<a name="line-8216"></a>  description: "The handle to reference the Reader."
+<a name="line-8217"></a>  is_ref: true
+<a name="line-8218"></a>  name: "reader_handle"
+<a name="line-8219"></a>  type: DT_STRING
+<a name="line-8220"></a>}
+<a name="line-8221"></a>-}</span>
+<a name="line-8222"></a>
+<a name="line-8223"></a><a name="switch"></a><span class='hs-comment'>-- | Forwards `data` to the output port determined by `pred`.</span>
+<a name="line-8224"></a><span class='hs-comment'>--</span>
+<a name="line-8225"></a><span class='hs-comment'>-- If `pred` is true, the `data` input is forwarded to `output_true`. Otherwise,</span>
+<a name="line-8226"></a><span class='hs-comment'>-- the data goes to `output_false`.</span>
+<a name="line-8227"></a><span class='hs-comment'>-- </span>
+<a name="line-8228"></a><span class='hs-comment'>-- See also `RefSwitch` and `Merge`.</span>
+<a name="line-8229"></a><span class='hs-definition'>switch</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-8230"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__: The tensor to be forwarded to the appropriate output.</span>
+<a name="line-8231"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __pred__: A scalar that specifies which output port will receive data.</span>
+<a name="line-8232"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-8233"></a>          <span class='hs-comment'>-- ^ (__output_false__, __output_true__)</span>
+<a name="line-8234"></a>          <span class='hs-comment'>--</span>
+<a name="line-8235"></a>          <span class='hs-comment'>-- * __output_false__: If `pred` is false, data will be forwarded to this output.</span>
+<a name="line-8236"></a>          <span class='hs-comment'>--</span>
+<a name="line-8237"></a>          <span class='hs-comment'>-- * __output_true__: If `pred` is true, data will be forwarded to this output.</span>
+<a name="line-8238"></a><span class='hs-definition'>switch</span> <span class='hs-varid'>data'</span> <span class='hs-varid'>pred</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8239"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Switch"</span>
+<a name="line-8240"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-8241"></a>        <span class='hs-varid'>data'</span> <span class='hs-varid'>pred</span>
+<a name="line-8242"></a><span class='hs-comment'>{-
+<a name="line-8243"></a>attr { name: "T" type: "type" }
+<a name="line-8244"></a>input_arg {
+<a name="line-8245"></a>  description: "The tensor to be forwarded to the appropriate output."
+<a name="line-8246"></a>  name: "data"
+<a name="line-8247"></a>  type_attr: "T"
+<a name="line-8248"></a>}
+<a name="line-8249"></a>input_arg {
+<a name="line-8250"></a>  description: "A scalar that specifies which output port will receive data."
+<a name="line-8251"></a>  name: "pred"
+<a name="line-8252"></a>  type: DT_BOOL
+<a name="line-8253"></a>}
+<a name="line-8254"></a>output_arg {
+<a name="line-8255"></a>  description: "If `pred` is false, data will be forwarded to this output."
+<a name="line-8256"></a>  name: "output_false"
+<a name="line-8257"></a>  type_attr: "T"
+<a name="line-8258"></a>}
+<a name="line-8259"></a>output_arg {
+<a name="line-8260"></a>  description: "If `pred` is true, data will be forwarded to this output."
+<a name="line-8261"></a>  name: "output_true"
+<a name="line-8262"></a>  type_attr: "T"
+<a name="line-8263"></a>}
+<a name="line-8264"></a>-}</span>
+<a name="line-8265"></a>
+<a name="line-8266"></a><a name="randomStandardNormal"></a><span class='hs-comment'>-- | Outputs random values from a normal distribution.</span>
+<a name="line-8267"></a><span class='hs-comment'>--</span>
+<a name="line-8268"></a><span class='hs-comment'>-- The generated values will have mean 0 and standard deviation 1.</span>
+<a name="line-8269"></a><span class='hs-definition'>randomStandardNormal</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-8270"></a>                                             <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-8271"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-8272"></a>                                             <span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>,</span>
+<a name="line-8273"></a>                                             <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-8274"></a>                                                     <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-8275"></a>                        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __shape__: The shape of the output tensor.</span>
+<a name="line-8276"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __output__: A tensor of the specified shape filled with random normal values.</span>
+<a name="line-8277"></a><span class='hs-definition'>randomStandardNormal</span> <span class='hs-varid'>shape</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8278"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RandomStandardNormal"</span>
+<a name="line-8279"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-8280"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-8281"></a>        <span class='hs-varid'>shape</span>
+<a name="line-8282"></a><span class='hs-comment'>{-
+<a name="line-8283"></a>attr {
+<a name="line-8284"></a>  default_value { i: 0 }
+<a name="line-8285"></a>  description: "If either `seed` or `seed2` are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-8286"></a>  name: "seed"
+<a name="line-8287"></a>  type: "int"
+<a name="line-8288"></a>}
+<a name="line-8289"></a>attr {
+<a name="line-8290"></a>  default_value { i: 0 }
+<a name="line-8291"></a>  description: "A second seed to avoid seed collision."
+<a name="line-8292"></a>  name: "seed2"
+<a name="line-8293"></a>  type: "int"
+<a name="line-8294"></a>}
+<a name="line-8295"></a>attr {
+<a name="line-8296"></a>  allowed_values {
+<a name="line-8297"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-8298"></a>  }
+<a name="line-8299"></a>  description: "The type of the output."
+<a name="line-8300"></a>  name: "dtype"
+<a name="line-8301"></a>  type: "type"
+<a name="line-8302"></a>}
+<a name="line-8303"></a>attr {
+<a name="line-8304"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-8305"></a>  name: "T"
+<a name="line-8306"></a>  type: "type"
+<a name="line-8307"></a>}
+<a name="line-8308"></a>input_arg {
+<a name="line-8309"></a>  description: "The shape of the output tensor."
+<a name="line-8310"></a>  name: "shape"
+<a name="line-8311"></a>  type_attr: "T"
+<a name="line-8312"></a>}
+<a name="line-8313"></a>output_arg {
+<a name="line-8314"></a>  description: "A tensor of the specified shape filled with random normal values."
+<a name="line-8315"></a>  name: "output"
+<a name="line-8316"></a>  type_attr: "dtype"
+<a name="line-8317"></a>}
+<a name="line-8318"></a>-}</span>
+<a name="line-8319"></a>
+<a name="line-8320"></a><a name="sigmoid"></a><span class='hs-comment'>-- | Computes sigmoid of `x` element-wise.</span>
+<a name="line-8321"></a><span class='hs-comment'>--</span>
+<a name="line-8322"></a><span class='hs-comment'>-- Specifically, `y = 1 / (1 + exp(-x))`.</span>
+<a name="line-8323"></a><span class='hs-definition'>sigmoid</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-8324"></a>                                                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-8325"></a>                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-8326"></a>                                                <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-8327"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-8328"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-8329"></a><span class='hs-definition'>sigmoid</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8330"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Sigmoid"</span>
+<a name="line-8331"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-8332"></a>        <span class='hs-varid'>x</span>
+<a name="line-8333"></a><span class='hs-comment'>{-
+<a name="line-8334"></a>attr {
+<a name="line-8335"></a>  allowed_values {
+<a name="line-8336"></a>    list {
+<a name="line-8337"></a>      type: DT_HALF
+<a name="line-8338"></a>      type: DT_FLOAT
+<a name="line-8339"></a>      type: DT_DOUBLE
+<a name="line-8340"></a>      type: DT_COMPLEX64
+<a name="line-8341"></a>      type: DT_COMPLEX128
+<a name="line-8342"></a>    }
+<a name="line-8343"></a>  }
+<a name="line-8344"></a>  name: "T"
+<a name="line-8345"></a>  type: "type"
+<a name="line-8346"></a>}
+<a name="line-8347"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-8348"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-8349"></a>-}</span>
+<a name="line-8350"></a>
+<a name="line-8351"></a><a name="sampleDistortedBoundingBox"></a><span class='hs-comment'>-- | Generate a single randomly distorted bounding box for an image.</span>
+<a name="line-8352"></a><span class='hs-comment'>--</span>
+<a name="line-8353"></a><span class='hs-comment'>-- Bounding box annotations are often supplied in addition to ground-truth labels</span>
+<a name="line-8354"></a><span class='hs-comment'>-- in image recognition or object localization tasks. A common technique for</span>
+<a name="line-8355"></a><span class='hs-comment'>-- training such a system is to randomly distort an image while preserving</span>
+<a name="line-8356"></a><span class='hs-comment'>-- its content, i.e. *data augmentation*. This Op outputs a randomly distorted</span>
+<a name="line-8357"></a><span class='hs-comment'>-- localization of an object, i.e. bounding box, given an `image_size`,</span>
+<a name="line-8358"></a><span class='hs-comment'>-- `bounding_boxes` and a series of constraints.</span>
+<a name="line-8359"></a><span class='hs-comment'>-- </span>
+<a name="line-8360"></a><span class='hs-comment'>-- The output of this Op is a single bounding box that may be used to crop the</span>
+<a name="line-8361"></a><span class='hs-comment'>-- original image. The output is returned as 3 tensors: `begin`, `size` and</span>
+<a name="line-8362"></a><span class='hs-comment'>-- `bboxes`. The first 2 tensors can be fed directly into `tf.slice` to crop the</span>
+<a name="line-8363"></a><span class='hs-comment'>-- image. The latter may be supplied to `tf.image.draw_bounding_box` to visualize</span>
+<a name="line-8364"></a><span class='hs-comment'>-- what the bounding box looks like.</span>
+<a name="line-8365"></a><span class='hs-comment'>-- </span>
+<a name="line-8366"></a><span class='hs-comment'>-- Bounding boxes are supplied and returned as `[y_min, x_min, y_max, x_max]`. The</span>
+<a name="line-8367"></a><span class='hs-comment'>-- bounding box coordinates are floats in `[0.0, 1.0]` relative to the width and</span>
+<a name="line-8368"></a><span class='hs-comment'>-- height of the underlying image.</span>
+<a name="line-8369"></a><span class='hs-comment'>-- </span>
+<a name="line-8370"></a><span class='hs-comment'>-- For example,</span>
+<a name="line-8371"></a><span class='hs-comment'>-- </span>
+<a name="line-8372"></a><span class='hs-comment'>--     # Generate a single distorted bounding box.</span>
+<a name="line-8373"></a><span class='hs-comment'>--     begin, size, bbox_for_draw = tf.image.sample_distorted_bounding_box(</span>
+<a name="line-8374"></a><span class='hs-comment'>--         tf.shape(image),</span>
+<a name="line-8375"></a><span class='hs-comment'>--         bounding_boxes=bounding_boxes)</span>
+<a name="line-8376"></a><span class='hs-comment'>-- </span>
+<a name="line-8377"></a><span class='hs-comment'>--     # Draw the bounding box in an image summary.</span>
+<a name="line-8378"></a><span class='hs-comment'>--     image_with_box = tf.image.draw_bounding_boxes(tf.expand_dims(image, 0),</span>
+<a name="line-8379"></a><span class='hs-comment'>--                                                   bbox_for_draw)</span>
+<a name="line-8380"></a><span class='hs-comment'>--     tf.image_summary('images_with_box', image_with_box)</span>
+<a name="line-8381"></a><span class='hs-comment'>-- </span>
+<a name="line-8382"></a><span class='hs-comment'>--     # Employ the bounding box to distort the image.</span>
+<a name="line-8383"></a><span class='hs-comment'>--     distorted_image = tf.slice(image, begin, size)</span>
+<a name="line-8384"></a><span class='hs-comment'>-- </span>
+<a name="line-8385"></a><span class='hs-comment'>-- Note that if no bounding box information is available, setting</span>
+<a name="line-8386"></a><span class='hs-comment'>-- `use_image_if_no_bounding_boxes = true` will assume there is a single implicit</span>
+<a name="line-8387"></a><span class='hs-comment'>-- bounding box covering the whole image. If `use_image_if_no_bounding_boxes` is</span>
+<a name="line-8388"></a><span class='hs-comment'>-- false and no bounding boxes are supplied, an error is raised.</span>
+<a name="line-8389"></a><span class='hs-definition'>sampleDistortedBoundingBox</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-8390"></a>                                                <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-8391"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-8392"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-8393"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-8394"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-8395"></a>                              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __image_size__: 1-D, containing `[height, width, channels]`.</span>
+<a name="line-8396"></a>                              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __bounding_boxes__: 3-D with shape `[batch, N, 4]` describing the N bounding boxes</span>
+<a name="line-8397"></a>                                                 <span class='hs-comment'>-- associated with the image.</span>
+<a name="line-8398"></a>                              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-8399"></a>                                  <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span>
+<a name="line-8400"></a>                              <span class='hs-comment'>-- ^ (__begin__, __size__, __bboxes__)</span>
+<a name="line-8401"></a>                              <span class='hs-comment'>--</span>
+<a name="line-8402"></a>                              <span class='hs-comment'>-- * __begin__: 1-D, containing `[offset_height, offset_width, 0]`. Provide as input to</span>
+<a name="line-8403"></a>                              <span class='hs-comment'>-- `tf.slice`.</span>
+<a name="line-8404"></a>                              <span class='hs-comment'>--</span>
+<a name="line-8405"></a>                              <span class='hs-comment'>-- * __size__: 1-D, containing `[target_height, target_width, -1]`. Provide as input to</span>
+<a name="line-8406"></a>                              <span class='hs-comment'>-- `tf.slice`.</span>
+<a name="line-8407"></a>                              <span class='hs-comment'>--</span>
+<a name="line-8408"></a>                              <span class='hs-comment'>-- * __bboxes__: 3-D with shape `[1, 1, 4]` containing the distorted bounding box.</span>
+<a name="line-8409"></a>                              <span class='hs-comment'>-- Provide as input to `tf.image.draw_bounding_boxes`.</span>
+<a name="line-8410"></a><span class='hs-definition'>sampleDistortedBoundingBox</span> <span class='hs-varid'>image_size</span> <span class='hs-varid'>bounding_boxes</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8411"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SampleDistortedBoundingBox"</span>
+<a name="line-8412"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-8413"></a>        <span class='hs-varid'>image_size</span> <span class='hs-varid'>bounding_boxes</span>
+<a name="line-8414"></a><span class='hs-comment'>{-
+<a name="line-8415"></a>attr {
+<a name="line-8416"></a>  allowed_values {
+<a name="line-8417"></a>    list {
+<a name="line-8418"></a>      type: DT_UINT8
+<a name="line-8419"></a>      type: DT_INT8
+<a name="line-8420"></a>      type: DT_INT16
+<a name="line-8421"></a>      type: DT_INT32
+<a name="line-8422"></a>      type: DT_INT64
+<a name="line-8423"></a>    }
+<a name="line-8424"></a>  }
+<a name="line-8425"></a>  name: "T"
+<a name="line-8426"></a>  type: "type"
+<a name="line-8427"></a>}
+<a name="line-8428"></a>attr {
+<a name="line-8429"></a>  default_value { i: 0 }
+<a name="line-8430"></a>  description: "If either `seed` or `seed2` are set to non-zero, the random number\ngenerator is seeded by the given `seed`.  Otherwise, it is seeded by a random\nseed."
+<a name="line-8431"></a>  name: "seed"
+<a name="line-8432"></a>  type: "int"
+<a name="line-8433"></a>}
+<a name="line-8434"></a>attr {
+<a name="line-8435"></a>  default_value { i: 0 }
+<a name="line-8436"></a>  description: "A second seed to avoid seed collision."
+<a name="line-8437"></a>  name: "seed2"
+<a name="line-8438"></a>  type: "int"
+<a name="line-8439"></a>}
+<a name="line-8440"></a>attr {
+<a name="line-8441"></a>  default_value { f: 0.1 }
+<a name="line-8442"></a>  description: "The cropped area of the image must contain at least this\nfraction of any bounding box supplied."
+<a name="line-8443"></a>  name: "min_object_covered"
+<a name="line-8444"></a>  type: "float"
+<a name="line-8445"></a>}
+<a name="line-8446"></a>attr {
+<a name="line-8447"></a>  default_value { list { f: 0.75 f: 1.33 } }
+<a name="line-8448"></a>  description: "The cropped area of the image must have an aspect ratio =\nwidth / height within this range."
+<a name="line-8449"></a>  name: "aspect_ratio_range"
+<a name="line-8450"></a>  type: "list(float)"
+<a name="line-8451"></a>}
+<a name="line-8452"></a>attr {
+<a name="line-8453"></a>  default_value { list { f: 5.0e-2 f: 1.0 } }
+<a name="line-8454"></a>  description: "The cropped area of the image must contain a fraction of the\nsupplied image within in this range."
+<a name="line-8455"></a>  name: "area_range"
+<a name="line-8456"></a>  type: "list(float)"
+<a name="line-8457"></a>}
+<a name="line-8458"></a>attr {
+<a name="line-8459"></a>  default_value { i: 100 }
+<a name="line-8460"></a>  description: "Number of attempts at generating a cropped region of the image\nof the specified constraints. After `max_attempts` failures, return the entire\nimage."
+<a name="line-8461"></a>  name: "max_attempts"
+<a name="line-8462"></a>  type: "int"
+<a name="line-8463"></a>}
+<a name="line-8464"></a>attr {
+<a name="line-8465"></a>  default_value { b: false }
+<a name="line-8466"></a>  description: "Controls behavior if no bounding boxes supplied.\nIf true, assume an implicit bounding box covering the whole input. If false,\nraise an error."
+<a name="line-8467"></a>  name: "use_image_if_no_bounding_boxes"
+<a name="line-8468"></a>  type: "bool"
+<a name="line-8469"></a>}
+<a name="line-8470"></a>input_arg {
+<a name="line-8471"></a>  description: "1-D, containing `[height, width, channels]`."
+<a name="line-8472"></a>  name: "image_size"
+<a name="line-8473"></a>  type_attr: "T"
+<a name="line-8474"></a>}
+<a name="line-8475"></a>input_arg {
+<a name="line-8476"></a>  description: "3-D with shape `[batch, N, 4]` describing the N bounding boxes\nassociated with the image."
+<a name="line-8477"></a>  name: "bounding_boxes"
+<a name="line-8478"></a>  type: DT_FLOAT
+<a name="line-8479"></a>}
+<a name="line-8480"></a>output_arg {
+<a name="line-8481"></a>  description: "1-D, containing `[offset_height, offset_width, 0]`. Provide as input to\n`tf.slice`."
+<a name="line-8482"></a>  name: "begin"
+<a name="line-8483"></a>  type_attr: "T"
+<a name="line-8484"></a>}
+<a name="line-8485"></a>output_arg {
+<a name="line-8486"></a>  description: "1-D, containing `[target_height, target_width, -1]`. Provide as input to\n`tf.slice`."
+<a name="line-8487"></a>  name: "size"
+<a name="line-8488"></a>  type_attr: "T"
+<a name="line-8489"></a>}
+<a name="line-8490"></a>output_arg {
+<a name="line-8491"></a>  description: "3-D with shape `[1, 1, 4]` containing the distorted bounding box.\nProvide as input to `tf.image.draw_bounding_boxes`."
+<a name="line-8492"></a>  name: "bboxes"
+<a name="line-8493"></a>  type: DT_FLOAT
+<a name="line-8494"></a>}
+<a name="line-8495"></a>-}</span>
+<a name="line-8496"></a>
+<a name="line-8497"></a><a name="greater"></a><span class='hs-comment'>-- | Returns the truth value of (x &gt; y) element-wise.</span>
+<a name="line-8498"></a><span class='hs-comment'>--</span>
+<a name="line-8499"></a><span class='hs-comment'>-- *NOTE*: `Greater` supports broadcasting. More about broadcasting</span>
+<a name="line-8500"></a><span class='hs-comment'>-- [here](<a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)">http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)</a></span>
+<a name="line-8501"></a><span class='hs-definition'>greater</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-8502"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-8503"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-8504"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-8505"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-8506"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-8507"></a>                                                   <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-8508"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-8509"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-8510"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-8511"></a><span class='hs-definition'>greater</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8512"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Greater"</span>
+<a name="line-8513"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-8514"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-8515"></a><span class='hs-comment'>{-
+<a name="line-8516"></a>attr {
+<a name="line-8517"></a>  allowed_values {
+<a name="line-8518"></a>    list {
+<a name="line-8519"></a>      type: DT_FLOAT
+<a name="line-8520"></a>      type: DT_DOUBLE
+<a name="line-8521"></a>      type: DT_INT32
+<a name="line-8522"></a>      type: DT_INT64
+<a name="line-8523"></a>      type: DT_UINT8
+<a name="line-8524"></a>      type: DT_INT16
+<a name="line-8525"></a>      type: DT_INT8
+<a name="line-8526"></a>      type: DT_UINT16
+<a name="line-8527"></a>      type: DT_HALF
+<a name="line-8528"></a>    }
+<a name="line-8529"></a>  }
+<a name="line-8530"></a>  name: "T"
+<a name="line-8531"></a>  type: "type"
+<a name="line-8532"></a>}
+<a name="line-8533"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-8534"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-8535"></a>output_arg { name: "z" type: DT_BOOL }
+<a name="line-8536"></a>-}</span>
+<a name="line-8537"></a>
+<a name="line-8538"></a><span class='hs-comment'>-- | Makes its input available to the next iteration.</span>
+<a name="line-8539"></a>
+<a name="line-8540"></a><a name="refNextIteration"></a><span class='hs-definition'>refNextIteration</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-8541"></a>                    <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__: The tensor to be made available to the next iteration.</span>
+<a name="line-8542"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The same tensor as `data`.</span>
+<a name="line-8543"></a><span class='hs-definition'>refNextIteration</span> <span class='hs-varid'>data'</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8544"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RefNextIteration"</span>
+<a name="line-8545"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-8546"></a>        <span class='hs-varid'>data'</span>
+<a name="line-8547"></a><span class='hs-comment'>{-
+<a name="line-8548"></a>attr { name: "T" type: "type" }
+<a name="line-8549"></a>input_arg {
+<a name="line-8550"></a>  description: "The tensor to be made available to the next iteration."
+<a name="line-8551"></a>  is_ref: true
+<a name="line-8552"></a>  name: "data"
+<a name="line-8553"></a>  type_attr: "T"
+<a name="line-8554"></a>}
+<a name="line-8555"></a>output_arg {
+<a name="line-8556"></a>  description: "The same tensor as `data`."
+<a name="line-8557"></a>  is_ref: true
+<a name="line-8558"></a>  name: "output"
+<a name="line-8559"></a>  type_attr: "T"
+<a name="line-8560"></a>}
+<a name="line-8561"></a>-}</span>
+<a name="line-8562"></a>
+<a name="line-8563"></a><a name="spaceToDepth"></a><span class='hs-comment'>-- | SpaceToDepth for tensors of type T.</span>
+<a name="line-8564"></a><span class='hs-comment'>--</span>
+<a name="line-8565"></a><span class='hs-comment'>-- Rearranges blocks of spatial data, into depth. More specifically,</span>
+<a name="line-8566"></a><span class='hs-comment'>-- this op outputs a copy of the input tensor where values from the `height`</span>
+<a name="line-8567"></a><span class='hs-comment'>-- and `width` dimensions are moved to the `depth` dimension.</span>
+<a name="line-8568"></a><span class='hs-comment'>-- The attr `block_size` indicates the input block size and how the data is moved.</span>
+<a name="line-8569"></a><span class='hs-comment'>-- </span>
+<a name="line-8570"></a><span class='hs-comment'>--   * Non-overlapping blocks of size `block_size x block size` are rearranged</span>
+<a name="line-8571"></a><span class='hs-comment'>--     into depth at each location.</span>
+<a name="line-8572"></a><span class='hs-comment'>--   * The depth of the output tensor is `input_depth * block_size * block_size`.</span>
+<a name="line-8573"></a><span class='hs-comment'>--   * The input tensor's height and width must be divisible by block_size.</span>
+<a name="line-8574"></a><span class='hs-comment'>-- </span>
+<a name="line-8575"></a><span class='hs-comment'>-- That is, assuming the input is in the shape:</span>
+<a name="line-8576"></a><span class='hs-comment'>-- `[batch, height, width, depth]`,</span>
+<a name="line-8577"></a><span class='hs-comment'>-- the shape of the output will be:</span>
+<a name="line-8578"></a><span class='hs-comment'>-- `[batch, height/block_size, width/block_size, depth*block_size*block_size]`</span>
+<a name="line-8579"></a><span class='hs-comment'>-- </span>
+<a name="line-8580"></a><span class='hs-comment'>-- This operation requires that the input tensor be of rank 4, and that</span>
+<a name="line-8581"></a><span class='hs-comment'>-- `block_size` be &gt;=1 and a divisor of both the input `height` and `width`.</span>
+<a name="line-8582"></a><span class='hs-comment'>-- </span>
+<a name="line-8583"></a><span class='hs-comment'>-- This operation is useful for resizing the activations between convolutions</span>
+<a name="line-8584"></a><span class='hs-comment'>-- (but keeping all data), e.g. instead of pooling. It is also useful for training</span>
+<a name="line-8585"></a><span class='hs-comment'>-- purely convolutional models.</span>
+<a name="line-8586"></a><span class='hs-comment'>-- </span>
+<a name="line-8587"></a><span class='hs-comment'>-- For example, given this input of shape `[1, 2, 2, 1]`, and block_size of 2:</span>
+<a name="line-8588"></a><span class='hs-comment'>-- </span>
+<a name="line-8589"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-8590"></a><span class='hs-comment'>-- x = [[[[1], [2]],</span>
+<a name="line-8591"></a><span class='hs-comment'>--       [[3], [4]]]]</span>
+<a name="line-8592"></a><span class='hs-comment'>-- ```</span>
+<a name="line-8593"></a><span class='hs-comment'>-- </span>
+<a name="line-8594"></a><span class='hs-comment'>-- This operation will output a tensor of shape `[1, 1, 1, 4]`:</span>
+<a name="line-8595"></a><span class='hs-comment'>-- </span>
+<a name="line-8596"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-8597"></a><span class='hs-comment'>-- [[[[1, 2, 3, 4]]]]</span>
+<a name="line-8598"></a><span class='hs-comment'>-- ```</span>
+<a name="line-8599"></a><span class='hs-comment'>-- </span>
+<a name="line-8600"></a><span class='hs-comment'>-- Here, the input has a batch of 1 and each batch element has shape `[2, 2, 1]`,</span>
+<a name="line-8601"></a><span class='hs-comment'>-- the corresponding output will have a single element (i.e. width and height are</span>
+<a name="line-8602"></a><span class='hs-comment'>-- both 1) and will have a depth of 4 channels (1 * block_size * block_size).</span>
+<a name="line-8603"></a><span class='hs-comment'>-- The output element shape is `[1, 1, 4]`.</span>
+<a name="line-8604"></a><span class='hs-comment'>-- </span>
+<a name="line-8605"></a><span class='hs-comment'>-- For an input tensor with larger depth, here of shape `[1, 2, 2, 3]`, e.g.</span>
+<a name="line-8606"></a><span class='hs-comment'>-- </span>
+<a name="line-8607"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-8608"></a><span class='hs-comment'>-- x = [[[[1, 2, 3], [4, 5, 6]],</span>
+<a name="line-8609"></a><span class='hs-comment'>--       [[7, 8, 9], [10, 11, 12]]]]</span>
+<a name="line-8610"></a><span class='hs-comment'>-- ```</span>
+<a name="line-8611"></a><span class='hs-comment'>-- </span>
+<a name="line-8612"></a><span class='hs-comment'>-- This operation, for block_size of 2, will return the following tensor of shape</span>
+<a name="line-8613"></a><span class='hs-comment'>-- `[1, 1, 1, 12]`</span>
+<a name="line-8614"></a><span class='hs-comment'>-- </span>
+<a name="line-8615"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-8616"></a><span class='hs-comment'>-- [[[[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]]]]</span>
+<a name="line-8617"></a><span class='hs-comment'>-- ```</span>
+<a name="line-8618"></a><span class='hs-comment'>-- </span>
+<a name="line-8619"></a><span class='hs-comment'>-- Similarly, for the following input of shape `[1 4 4 1]`, and a block size of 2:</span>
+<a name="line-8620"></a><span class='hs-comment'>-- </span>
+<a name="line-8621"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-8622"></a><span class='hs-comment'>-- x = [[[[1],   [2],  [5],  [6]],</span>
+<a name="line-8623"></a><span class='hs-comment'>--       [[3],   [4],  [7],  [8]],</span>
+<a name="line-8624"></a><span class='hs-comment'>--       [[9],  [10], [13],  [14]],</span>
+<a name="line-8625"></a><span class='hs-comment'>--       [[11], [12], [15],  [16]]]]</span>
+<a name="line-8626"></a><span class='hs-comment'>-- ```</span>
+<a name="line-8627"></a><span class='hs-comment'>-- </span>
+<a name="line-8628"></a><span class='hs-comment'>-- the operator will return the following tensor of shape `[1 2 2 4]`:</span>
+<a name="line-8629"></a><span class='hs-comment'>-- </span>
+<a name="line-8630"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-8631"></a><span class='hs-comment'>-- x = [[[[1, 2, 3, 4],</span>
+<a name="line-8632"></a><span class='hs-comment'>--        [5, 6, 7, 8]],</span>
+<a name="line-8633"></a><span class='hs-comment'>--       [[9, 10, 11, 12],</span>
+<a name="line-8634"></a><span class='hs-comment'>--        [13, 14, 15, 16]]]]</span>
+<a name="line-8635"></a><span class='hs-comment'>-- ```</span>
+<a name="line-8636"></a><span class='hs-definition'>spaceToDepth</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-8637"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __block_size__: The size of the spatial block.</span>
+<a name="line-8638"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-8639"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-8640"></a><span class='hs-definition'>spaceToDepth</span> <span class='hs-varid'>block_size</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8641"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SpaceToDepth"</span>
+<a name="line-8642"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-8643"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"block_size"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>block_size</span><span class='hs-layout'>)</span>
+<a name="line-8644"></a>        <span class='hs-varid'>input</span>
+<a name="line-8645"></a><span class='hs-comment'>{-
+<a name="line-8646"></a>attr { name: "T" type: "type" }
+<a name="line-8647"></a>attr {
+<a name="line-8648"></a>  description: "The size of the spatial block."
+<a name="line-8649"></a>  has_minimum: true
+<a name="line-8650"></a>  minimum: 2
+<a name="line-8651"></a>  name: "block_size"
+<a name="line-8652"></a>  type: "int"
+<a name="line-8653"></a>}
+<a name="line-8654"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-8655"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-8656"></a>-}</span>
+<a name="line-8657"></a>
+<a name="line-8658"></a><a name="controlTrigger"></a><span class='hs-comment'>-- | Does nothing. Serves as a control trigger for scheduling.</span>
+<a name="line-8659"></a><span class='hs-comment'>--</span>
+<a name="line-8660"></a><span class='hs-comment'>-- Only useful as a placeholder for control edges.</span>
+<a name="line-8661"></a><span class='hs-definition'>controlTrigger</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-8662"></a><span class='hs-definition'>controlTrigger</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8663"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ControlTrigger"</span><span class='hs-layout'>)</span>
+<a name="line-8664"></a>        
+<a name="line-8665"></a><span class='hs-comment'>{-
+<a name="line-8666"></a>
+<a name="line-8667"></a>-}</span>
+<a name="line-8668"></a>
+<a name="line-8669"></a><a name="scatterDiv"></a><span class='hs-comment'>-- | Divides a variable reference by sparse updates.</span>
+<a name="line-8670"></a><span class='hs-comment'>--</span>
+<a name="line-8671"></a><span class='hs-comment'>-- This operation computes</span>
+<a name="line-8672"></a><span class='hs-comment'>-- </span>
+<a name="line-8673"></a><span class='hs-comment'>--     # Scalar indices</span>
+<a name="line-8674"></a><span class='hs-comment'>--     ref[indices, ...] /= updates[...]</span>
+<a name="line-8675"></a><span class='hs-comment'>-- </span>
+<a name="line-8676"></a><span class='hs-comment'>--     # Vector indices (for each i)</span>
+<a name="line-8677"></a><span class='hs-comment'>--     ref[indices[i], ...] /= updates[i, ...]</span>
+<a name="line-8678"></a><span class='hs-comment'>-- </span>
+<a name="line-8679"></a><span class='hs-comment'>--     # High rank indices (for each i, ..., j)</span>
+<a name="line-8680"></a><span class='hs-comment'>--     ref[indices[i, ..., j], ...] /= updates[i, ..., j, ...]</span>
+<a name="line-8681"></a><span class='hs-comment'>-- </span>
+<a name="line-8682"></a><span class='hs-comment'>-- This operation outputs `ref` after the update is done.</span>
+<a name="line-8683"></a><span class='hs-comment'>-- This makes it easier to chain operations that need to use the reset value.</span>
+<a name="line-8684"></a><span class='hs-comment'>-- </span>
+<a name="line-8685"></a><span class='hs-comment'>-- Duplicate entries are handled correctly: if multiple `indices` reference</span>
+<a name="line-8686"></a><span class='hs-comment'>-- the same location, their contributions divide.</span>
+<a name="line-8687"></a><span class='hs-comment'>-- </span>
+<a name="line-8688"></a><span class='hs-comment'>-- Requires `updates.shape = indices.shape + ref.shape[1:]`.</span>
+<a name="line-8689"></a><span class='hs-definition'>scatterDiv</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-8690"></a>                                            <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-8691"></a>                                                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-8692"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-8693"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-8694"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-8695"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-8696"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-8697"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-8698"></a>                                                    <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-8699"></a>                                            <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-8700"></a>                                            <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-8701"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-8702"></a>              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __ref__: Should be from a `Variable` node.</span>
+<a name="line-8703"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __indices__: A tensor of indices into the first dimension of `ref`.</span>
+<a name="line-8704"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __updates__: A tensor of values that `ref` is divided by.</span>
+<a name="line-8705"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output_ref__: = Same as `ref`.  Returned as a convenience for operations that want</span>
+<a name="line-8706"></a>              <span class='hs-comment'>-- to use the updated values after the update is done.</span>
+<a name="line-8707"></a><span class='hs-definition'>scatterDiv</span> <span class='hs-varid'>ref</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>updates</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8708"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ScatterDiv"</span>
+<a name="line-8709"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-8710"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-8711"></a>        <span class='hs-varid'>ref</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>updates</span>
+<a name="line-8712"></a><span class='hs-comment'>{-
+<a name="line-8713"></a>attr {
+<a name="line-8714"></a>  allowed_values {
+<a name="line-8715"></a>    list {
+<a name="line-8716"></a>      type: DT_FLOAT
+<a name="line-8717"></a>      type: DT_DOUBLE
+<a name="line-8718"></a>      type: DT_INT64
+<a name="line-8719"></a>      type: DT_INT32
+<a name="line-8720"></a>      type: DT_UINT8
+<a name="line-8721"></a>      type: DT_UINT16
+<a name="line-8722"></a>      type: DT_INT16
+<a name="line-8723"></a>      type: DT_INT8
+<a name="line-8724"></a>      type: DT_COMPLEX64
+<a name="line-8725"></a>      type: DT_COMPLEX128
+<a name="line-8726"></a>      type: DT_QINT8
+<a name="line-8727"></a>      type: DT_QUINT8
+<a name="line-8728"></a>      type: DT_QINT32
+<a name="line-8729"></a>      type: DT_HALF
+<a name="line-8730"></a>    }
+<a name="line-8731"></a>  }
+<a name="line-8732"></a>  name: "T"
+<a name="line-8733"></a>  type: "type"
+<a name="line-8734"></a>}
+<a name="line-8735"></a>attr {
+<a name="line-8736"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-8737"></a>  name: "Tindices"
+<a name="line-8738"></a>  type: "type"
+<a name="line-8739"></a>}
+<a name="line-8740"></a>attr {
+<a name="line-8741"></a>  default_value { b: false }
+<a name="line-8742"></a>  description: "If True, the operation will be protected by a lock;\notherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-8743"></a>  name: "use_locking"
+<a name="line-8744"></a>  type: "bool"
+<a name="line-8745"></a>}
+<a name="line-8746"></a>input_arg {
+<a name="line-8747"></a>  description: "Should be from a `Variable` node."
+<a name="line-8748"></a>  is_ref: true
+<a name="line-8749"></a>  name: "ref"
+<a name="line-8750"></a>  type_attr: "T"
+<a name="line-8751"></a>}
+<a name="line-8752"></a>input_arg {
+<a name="line-8753"></a>  description: "A tensor of indices into the first dimension of `ref`."
+<a name="line-8754"></a>  name: "indices"
+<a name="line-8755"></a>  type_attr: "Tindices"
+<a name="line-8756"></a>}
+<a name="line-8757"></a>input_arg {
+<a name="line-8758"></a>  description: "A tensor of values that `ref` is divided by."
+<a name="line-8759"></a>  name: "updates"
+<a name="line-8760"></a>  type_attr: "T"
+<a name="line-8761"></a>}
+<a name="line-8762"></a>output_arg {
+<a name="line-8763"></a>  description: "= Same as `ref`.  Returned as a convenience for operations that want\nto use the updated values after the update is done."
+<a name="line-8764"></a>  is_ref: true
+<a name="line-8765"></a>  name: "output_ref"
+<a name="line-8766"></a>  type_attr: "T"
+<a name="line-8767"></a>}
+<a name="line-8768"></a>-}</span>
+<a name="line-8769"></a>
+<a name="line-8770"></a><a name="copy"></a><span class='hs-comment'>-- | Copy Op.</span>
+<a name="line-8771"></a><span class='hs-comment'>--</span>
+<a name="line-8772"></a><span class='hs-comment'>-- Performs CPU-to-CPU or GPU-to-GPU deep-copying of tensor, depending on the</span>
+<a name="line-8773"></a><span class='hs-comment'>-- device on which the tensor is allocated.</span>
+<a name="line-8774"></a><span class='hs-comment'>-- </span>
+<a name="line-8775"></a><span class='hs-comment'>-- Unlike the CopyHost Op, this op does not have HostMemory constraint on its</span>
+<a name="line-8776"></a><span class='hs-comment'>-- input or output.</span>
+<a name="line-8777"></a><span class='hs-definition'>copy</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-8778"></a>        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Input tensor.</span>
+<a name="line-8779"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Output tensor, deep-copied from input.</span>
+<a name="line-8780"></a><span class='hs-definition'>copy</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8781"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Copy"</span>
+<a name="line-8782"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-8783"></a>        <span class='hs-varid'>input</span>
+<a name="line-8784"></a><span class='hs-comment'>{-
+<a name="line-8785"></a>attr { name: "T" type: "type" }
+<a name="line-8786"></a>attr {
+<a name="line-8787"></a>  default_value { s: "" }
+<a name="line-8788"></a>  description: "The name of the input tensor."
+<a name="line-8789"></a>  name: "tensor_name"
+<a name="line-8790"></a>  type: "string"
+<a name="line-8791"></a>}
+<a name="line-8792"></a>input_arg {
+<a name="line-8793"></a>  description: "Input tensor." name: "input" type_attr: "T"
+<a name="line-8794"></a>}
+<a name="line-8795"></a>output_arg {
+<a name="line-8796"></a>  description: "Output tensor, deep-copied from input."
+<a name="line-8797"></a>  name: "output"
+<a name="line-8798"></a>  type_attr: "T"
+<a name="line-8799"></a>}
+<a name="line-8800"></a>-}</span>
+<a name="line-8801"></a>
+<a name="line-8802"></a><span class='hs-comment'>-- | Computes the gradient of the crop_and_resize op wrt the input boxes tensor.</span>
+<a name="line-8803"></a>
+<a name="line-8804"></a><a name="cropAndResizeGradBoxes"></a><span class='hs-definition'>cropAndResizeGradBoxes</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-8805"></a>                                                  <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-8806"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-8807"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-8808"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-8809"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-8810"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-8811"></a>                                                          <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-8812"></a>                          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __grads__: A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`.</span>
+<a name="line-8813"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __image__: A 4-D tensor of shape `[batch, image_height, image_width, depth]`.</span>
+<a name="line-8814"></a>                                         <span class='hs-comment'>-- Both `image_height` and `image_width` need to be positive.</span>
+<a name="line-8815"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __boxes__: A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor</span>
+<a name="line-8816"></a>                                             <span class='hs-comment'>-- specifies the coordinates of a box in the `box_ind[i]` image and is specified</span>
+<a name="line-8817"></a>                                             <span class='hs-comment'>-- in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of</span>
+<a name="line-8818"></a>                                             <span class='hs-comment'>-- `y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the</span>
+<a name="line-8819"></a>                                             <span class='hs-comment'>-- `[0, 1]` interval of normalized image height is mapped to</span>
+<a name="line-8820"></a>                                             <span class='hs-comment'>-- `[0, image_height - 1] in image height coordinates. We do allow y1 &gt; y2, in</span>
+<a name="line-8821"></a>                                             <span class='hs-comment'>-- which case the sampled crop is an up-down flipped version of the original</span>
+<a name="line-8822"></a>                                             <span class='hs-comment'>-- image. The width dimension is treated similarly. Normalized coordinates</span>
+<a name="line-8823"></a>                                             <span class='hs-comment'>-- outside the `[0, 1]` range are allowed, in which case we use</span>
+<a name="line-8824"></a>                                             <span class='hs-comment'>-- `extrapolation_value` to extrapolate the input image values.</span>
+<a name="line-8825"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __box_ind__: A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.</span>
+<a name="line-8826"></a>                                                      <span class='hs-comment'>-- The value of `box_ind[i]` specifies the image that the `i`-th box refers to.</span>
+<a name="line-8827"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __output__: A 2-D tensor of shape `[num_boxes, 4]`.</span>
+<a name="line-8828"></a><span class='hs-definition'>cropAndResizeGradBoxes</span> <span class='hs-varid'>grads</span> <span class='hs-varid'>image</span> <span class='hs-varid'>boxes</span> <span class='hs-varid'>box_ind</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8829"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"CropAndResizeGradBoxes"</span>
+<a name="line-8830"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-8831"></a>        <span class='hs-varid'>grads</span> <span class='hs-varid'>image</span> <span class='hs-varid'>boxes</span> <span class='hs-varid'>box_ind</span>
+<a name="line-8832"></a><span class='hs-comment'>{-
+<a name="line-8833"></a>attr {
+<a name="line-8834"></a>  allowed_values {
+<a name="line-8835"></a>    list {
+<a name="line-8836"></a>      type: DT_UINT8
+<a name="line-8837"></a>      type: DT_INT8
+<a name="line-8838"></a>      type: DT_INT16
+<a name="line-8839"></a>      type: DT_INT32
+<a name="line-8840"></a>      type: DT_INT64
+<a name="line-8841"></a>      type: DT_HALF
+<a name="line-8842"></a>      type: DT_FLOAT
+<a name="line-8843"></a>      type: DT_DOUBLE
+<a name="line-8844"></a>    }
+<a name="line-8845"></a>  }
+<a name="line-8846"></a>  name: "T"
+<a name="line-8847"></a>  type: "type"
+<a name="line-8848"></a>}
+<a name="line-8849"></a>attr {
+<a name="line-8850"></a>  allowed_values { list { s: "bilinear" } }
+<a name="line-8851"></a>  default_value { s: "bilinear" }
+<a name="line-8852"></a>  description: "A string specifying the interpolation method. Only \'bilinear\' is\nsupported for now."
+<a name="line-8853"></a>  name: "method"
+<a name="line-8854"></a>  type: "string"
+<a name="line-8855"></a>}
+<a name="line-8856"></a>input_arg {
+<a name="line-8857"></a>  description: "A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`."
+<a name="line-8858"></a>  name: "grads"
+<a name="line-8859"></a>  type: DT_FLOAT
+<a name="line-8860"></a>}
+<a name="line-8861"></a>input_arg {
+<a name="line-8862"></a>  description: "A 4-D tensor of shape `[batch, image_height, image_width, depth]`.\nBoth `image_height` and `image_width` need to be positive."
+<a name="line-8863"></a>  name: "image"
+<a name="line-8864"></a>  type_attr: "T"
+<a name="line-8865"></a>}
+<a name="line-8866"></a>input_arg {
+<a name="line-8867"></a>  description: "A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor\nspecifies the coordinates of a box in the `box_ind[i]` image and is specified\nin normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of\n`y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the\n`[0, 1]` interval of normalized image height is mapped to\n`[0, image_height - 1] in image height coordinates. We do allow y1 &gt; y2, in\nwhich case the sampled crop is an up-down flipped version of the original\nimage. The width dimension is treated similarly. Normalized coordinates\noutside the `[0, 1]` range are allowed, in which case we use\n`extrapolation_value` to extrapolate the input image values."
+<a name="line-8868"></a>  name: "boxes"
+<a name="line-8869"></a>  type: DT_FLOAT
+<a name="line-8870"></a>}
+<a name="line-8871"></a>input_arg {
+<a name="line-8872"></a>  description: "A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.\nThe value of `box_ind[i]` specifies the image that the `i`-th box refers to."
+<a name="line-8873"></a>  name: "box_ind"
+<a name="line-8874"></a>  type: DT_INT32
+<a name="line-8875"></a>}
+<a name="line-8876"></a>output_arg {
+<a name="line-8877"></a>  description: "A 2-D tensor of shape `[num_boxes, 4]`."
+<a name="line-8878"></a>  name: "output"
+<a name="line-8879"></a>  type: DT_FLOAT
+<a name="line-8880"></a>}
+<a name="line-8881"></a>-}</span>
+<a name="line-8882"></a>
+<a name="line-8883"></a><a name="sparseSegmentMean"></a><span class='hs-comment'>-- | Computes the mean along sparse segments of a tensor.</span>
+<a name="line-8884"></a><span class='hs-comment'>--</span>
+<a name="line-8885"></a><span class='hs-comment'>-- Read [the section on</span>
+<a name="line-8886"></a><span class='hs-comment'>-- Segmentation](../../api_docs/python/math_ops.md#segmentation) for an explanation</span>
+<a name="line-8887"></a><span class='hs-comment'>-- of segments.</span>
+<a name="line-8888"></a><span class='hs-comment'>-- </span>
+<a name="line-8889"></a><span class='hs-comment'>-- Like `SegmentMean`, but `segment_ids` can have rank less than `data`'s first</span>
+<a name="line-8890"></a><span class='hs-comment'>-- dimension, selecting a subset of dimension 0, specified by `indices`.</span>
+<a name="line-8891"></a><span class='hs-definition'>sparseSegmentMean</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-8892"></a>                                                                     <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-8893"></a>                                               <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span>
+<a name="line-8894"></a>                                               <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-8895"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-8896"></a>                     <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__</span>
+<a name="line-8897"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __indices__: A 1-D tensor. Has same rank as `segment_ids`.</span>
+<a name="line-8898"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __segment_ids__: A 1-D tensor. Values should be sorted and can be repeated.</span>
+<a name="line-8899"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Has same shape as data, except for dimension 0 which</span>
+<a name="line-8900"></a>                     <span class='hs-comment'>-- has size `k`, the number of segments.</span>
+<a name="line-8901"></a><span class='hs-definition'>sparseSegmentMean</span> <span class='hs-varid'>data'</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>segment_ids</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8902"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseSegmentMean"</span>
+<a name="line-8903"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-8904"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-8905"></a>        <span class='hs-varid'>data'</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>segment_ids</span>
+<a name="line-8906"></a><span class='hs-comment'>{-
+<a name="line-8907"></a>attr {
+<a name="line-8908"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-8909"></a>  name: "T"
+<a name="line-8910"></a>  type: "type"
+<a name="line-8911"></a>}
+<a name="line-8912"></a>attr {
+<a name="line-8913"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-8914"></a>  default_value { type: DT_INT32 }
+<a name="line-8915"></a>  name: "Tidx"
+<a name="line-8916"></a>  type: "type"
+<a name="line-8917"></a>}
+<a name="line-8918"></a>input_arg { name: "data" type_attr: "T" }
+<a name="line-8919"></a>input_arg {
+<a name="line-8920"></a>  description: "A 1-D tensor. Has same rank as `segment_ids`."
+<a name="line-8921"></a>  name: "indices"
+<a name="line-8922"></a>  type_attr: "Tidx"
+<a name="line-8923"></a>}
+<a name="line-8924"></a>input_arg {
+<a name="line-8925"></a>  description: "A 1-D tensor. Values should be sorted and can be repeated."
+<a name="line-8926"></a>  name: "segment_ids"
+<a name="line-8927"></a>  type: DT_INT32
+<a name="line-8928"></a>}
+<a name="line-8929"></a>output_arg {
+<a name="line-8930"></a>  description: "Has same shape as data, except for dimension 0 which\nhas size `k`, the number of segments."
+<a name="line-8931"></a>  name: "output"
+<a name="line-8932"></a>  type_attr: "T"
+<a name="line-8933"></a>}
+<a name="line-8934"></a>-}</span>
+<a name="line-8935"></a>
+<a name="line-8936"></a><a name="assign"></a><span class='hs-comment'>-- | Update 'ref' by assigning 'value' to it.</span>
+<a name="line-8937"></a><span class='hs-comment'>--</span>
+<a name="line-8938"></a><span class='hs-comment'>-- This operation outputs "ref" after the assignment is done.</span>
+<a name="line-8939"></a><span class='hs-comment'>-- This makes it easier to chain operations that need to use the reset value.</span>
+<a name="line-8940"></a><span class='hs-definition'>assign</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-8941"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __ref__: Should be from a `Variable` node. May be uninitialized.</span>
+<a name="line-8942"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: The value to be assigned to the variable.</span>
+<a name="line-8943"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output_ref__: = Same as "ref".  Returned as a convenience for operations that want</span>
+<a name="line-8944"></a>          <span class='hs-comment'>-- to use the new value after the variable has been reset.</span>
+<a name="line-8945"></a><span class='hs-definition'>assign</span> <span class='hs-varid'>ref</span> <span class='hs-varid'>value</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-8946"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Assign"</span>
+<a name="line-8947"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-8948"></a>        <span class='hs-varid'>ref</span> <span class='hs-varid'>value</span>
+<a name="line-8949"></a><span class='hs-comment'>{-
+<a name="line-8950"></a>attr { name: "T" type: "type" }
+<a name="line-8951"></a>attr {
+<a name="line-8952"></a>  default_value { b: true }
+<a name="line-8953"></a>  description: "If true, the operation will validate that the shape\nof \'value\' matches the shape of the Tensor being assigned to.  If false,\n\'ref\' will take on the shape of \'value\'."
+<a name="line-8954"></a>  name: "validate_shape"
+<a name="line-8955"></a>  type: "bool"
+<a name="line-8956"></a>}
+<a name="line-8957"></a>attr {
+<a name="line-8958"></a>  default_value { b: true }
+<a name="line-8959"></a>  description: "If True, the assignment will be protected by a lock;\notherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-8960"></a>  name: "use_locking"
+<a name="line-8961"></a>  type: "bool"
+<a name="line-8962"></a>}
+<a name="line-8963"></a>input_arg {
+<a name="line-8964"></a>  description: "Should be from a `Variable` node. May be uninitialized."
+<a name="line-8965"></a>  is_ref: true
+<a name="line-8966"></a>  name: "ref"
+<a name="line-8967"></a>  type_attr: "T"
+<a name="line-8968"></a>}
+<a name="line-8969"></a>input_arg {
+<a name="line-8970"></a>  description: "The value to be assigned to the variable."
+<a name="line-8971"></a>  name: "value"
+<a name="line-8972"></a>  type_attr: "T"
+<a name="line-8973"></a>}
+<a name="line-8974"></a>output_arg {
+<a name="line-8975"></a>  description: "= Same as \"ref\".  Returned as a convenience for operations that want\nto use the new value after the variable has been reset."
+<a name="line-8976"></a>  is_ref: true
+<a name="line-8977"></a>  name: "output_ref"
+<a name="line-8978"></a>  type_attr: "T"
+<a name="line-8979"></a>}
+<a name="line-8980"></a>-}</span>
+<a name="line-8981"></a>
+<a name="line-8982"></a><a name="restore"></a><span class='hs-comment'>-- | Restores a tensor from checkpoint files.</span>
+<a name="line-8983"></a><span class='hs-comment'>--</span>
+<a name="line-8984"></a><span class='hs-comment'>-- Reads a tensor stored in one or several files. If there are several files (for</span>
+<a name="line-8985"></a><span class='hs-comment'>-- instance because a tensor was saved as slices), `file_pattern` may contain</span>
+<a name="line-8986"></a><span class='hs-comment'>-- wildcard symbols (`*` and `?`) in the filename portion only, not in the</span>
+<a name="line-8987"></a><span class='hs-comment'>-- directory portion.</span>
+<a name="line-8988"></a><span class='hs-comment'>-- </span>
+<a name="line-8989"></a><span class='hs-comment'>-- If a `file_pattern` matches several files, `preferred_shard` can be used to hint</span>
+<a name="line-8990"></a><span class='hs-comment'>-- in which file the requested tensor is likely to be found. This op will first</span>
+<a name="line-8991"></a><span class='hs-comment'>-- open the file at index `preferred_shard` in the list of matching files and try</span>
+<a name="line-8992"></a><span class='hs-comment'>-- to restore tensors from that file.  Only if some tensors or tensor slices are</span>
+<a name="line-8993"></a><span class='hs-comment'>-- not found in that first file, then the Op opens all the files. Setting</span>
+<a name="line-8994"></a><span class='hs-comment'>-- `preferred_shard` to match the value passed as the `shard` input</span>
+<a name="line-8995"></a><span class='hs-comment'>-- of a matching `Save` Op may speed up Restore.  This attribute only affects</span>
+<a name="line-8996"></a><span class='hs-comment'>-- performance, not correctness.  The default value -1 means files are processed in</span>
+<a name="line-8997"></a><span class='hs-comment'>-- order.</span>
+<a name="line-8998"></a><span class='hs-comment'>-- </span>
+<a name="line-8999"></a><span class='hs-comment'>-- See also `RestoreSlice`.</span>
+<a name="line-9000"></a><span class='hs-definition'>restore</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>dt</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dt</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9001"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __file_pattern__: Must have a single element. The pattern of the files from</span>
+<a name="line-9002"></a>                                                <span class='hs-comment'>-- which we read the tensor.</span>
+<a name="line-9003"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __tensor_name__: Must have a single element. The name of the tensor to be</span>
+<a name="line-9004"></a>                                                   <span class='hs-comment'>-- restored.</span>
+<a name="line-9005"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dt</span> <span class='hs-comment'>-- ^ __tensor__: The restored tensor.</span>
+<a name="line-9006"></a><span class='hs-definition'>restore</span> <span class='hs-varid'>file_pattern</span> <span class='hs-varid'>tensor_name</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9007"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Restore"</span>
+<a name="line-9008"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dt"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dt</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9009"></a>        <span class='hs-varid'>file_pattern</span> <span class='hs-varid'>tensor_name</span>
+<a name="line-9010"></a><span class='hs-comment'>{-
+<a name="line-9011"></a>attr {
+<a name="line-9012"></a>  description: "The type of the tensor to be restored."
+<a name="line-9013"></a>  name: "dt"
+<a name="line-9014"></a>  type: "type"
+<a name="line-9015"></a>}
+<a name="line-9016"></a>attr {
+<a name="line-9017"></a>  default_value { i: -1 }
+<a name="line-9018"></a>  description: "Index of file to open first if multiple files match\n`file_pattern`."
+<a name="line-9019"></a>  name: "preferred_shard"
+<a name="line-9020"></a>  type: "int"
+<a name="line-9021"></a>}
+<a name="line-9022"></a>input_arg {
+<a name="line-9023"></a>  description: "Must have a single element. The pattern of the files from\nwhich we read the tensor."
+<a name="line-9024"></a>  name: "file_pattern"
+<a name="line-9025"></a>  type: DT_STRING
+<a name="line-9026"></a>}
+<a name="line-9027"></a>input_arg {
+<a name="line-9028"></a>  description: "Must have a single element. The name of the tensor to be\nrestored."
+<a name="line-9029"></a>  name: "tensor_name"
+<a name="line-9030"></a>  type: DT_STRING
+<a name="line-9031"></a>}
+<a name="line-9032"></a>output_arg {
+<a name="line-9033"></a>  description: "The restored tensor." name: "tensor" type_attr: "dt"
+<a name="line-9034"></a>}
+<a name="line-9035"></a>-}</span>
+<a name="line-9036"></a>
+<a name="line-9037"></a><span class='hs-comment'>-- | Computes gradients of the maxpooling function.</span>
+<a name="line-9038"></a>
+<a name="line-9039"></a><a name="maxPoolGradWithArgmax"></a><span class='hs-definition'>maxPoolGradWithArgmax</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varid'>targmax</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-9040"></a>                                                      <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-9041"></a>                                                              <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-9042"></a>                                                      <span class='hs-conid'>TensorType</span> <span class='hs-varid'>targmax</span><span class='hs-layout'>,</span>
+<a name="line-9043"></a>                                                      <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-9044"></a>                                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>targmax</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9045"></a>                         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: The original input.</span>
+<a name="line-9046"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: 4-D with shape `[batch, height, width, channels]`.  Gradients w.r.t. the</span>
+<a name="line-9047"></a>                                        <span class='hs-comment'>-- output of `max_pool`.</span>
+<a name="line-9048"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>targmax</span> <span class='hs-comment'>-- ^ __argmax__: The indices of the maximum values chosen for each output of `max_pool`.</span>
+<a name="line-9049"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Gradients w.r.t. the input of `max_pool`.</span>
+<a name="line-9050"></a><span class='hs-definition'>maxPoolGradWithArgmax</span> <span class='hs-varid'>input</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>argmax</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9051"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MaxPoolGradWithArgmax"</span>
+<a name="line-9052"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-9053"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Targmax"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>targmax</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9054"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>argmax</span>
+<a name="line-9055"></a><span class='hs-comment'>{-
+<a name="line-9056"></a>attr {
+<a name="line-9057"></a>  description: "The size of the window for each dimension of the input tensor."
+<a name="line-9058"></a>  has_minimum: true
+<a name="line-9059"></a>  minimum: 4
+<a name="line-9060"></a>  name: "ksize"
+<a name="line-9061"></a>  type: "list(int)"
+<a name="line-9062"></a>}
+<a name="line-9063"></a>attr {
+<a name="line-9064"></a>  description: "The stride of the sliding window for each dimension of the\ninput tensor."
+<a name="line-9065"></a>  has_minimum: true
+<a name="line-9066"></a>  minimum: 4
+<a name="line-9067"></a>  name: "strides"
+<a name="line-9068"></a>  type: "list(int)"
+<a name="line-9069"></a>}
+<a name="line-9070"></a>attr {
+<a name="line-9071"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-9072"></a>  description: "The type of padding algorithm to use."
+<a name="line-9073"></a>  name: "padding"
+<a name="line-9074"></a>  type: "string"
+<a name="line-9075"></a>}
+<a name="line-9076"></a>attr {
+<a name="line-9077"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-9078"></a>  name: "Targmax"
+<a name="line-9079"></a>  type: "type"
+<a name="line-9080"></a>}
+<a name="line-9081"></a>attr {
+<a name="line-9082"></a>  allowed_values { list { type: DT_FLOAT type: DT_HALF } }
+<a name="line-9083"></a>  default_value { type: DT_FLOAT }
+<a name="line-9084"></a>  name: "T"
+<a name="line-9085"></a>  type: "type"
+<a name="line-9086"></a>}
+<a name="line-9087"></a>input_arg {
+<a name="line-9088"></a>  description: "The original input." name: "input" type_attr: "T"
+<a name="line-9089"></a>}
+<a name="line-9090"></a>input_arg {
+<a name="line-9091"></a>  description: "4-D with shape `[batch, height, width, channels]`.  Gradients w.r.t. the\noutput of `max_pool`."
+<a name="line-9092"></a>  name: "grad"
+<a name="line-9093"></a>  type_attr: "T"
+<a name="line-9094"></a>}
+<a name="line-9095"></a>input_arg {
+<a name="line-9096"></a>  description: "The indices of the maximum values chosen for each output of `max_pool`."
+<a name="line-9097"></a>  name: "argmax"
+<a name="line-9098"></a>  type_attr: "Targmax"
+<a name="line-9099"></a>}
+<a name="line-9100"></a>output_arg {
+<a name="line-9101"></a>  description: "Gradients w.r.t. the input of `max_pool`."
+<a name="line-9102"></a>  name: "output"
+<a name="line-9103"></a>  type_attr: "T"
+<a name="line-9104"></a>}
+<a name="line-9105"></a>-}</span>
+<a name="line-9106"></a>
+<a name="line-9107"></a><a name="checkNumerics"></a><span class='hs-comment'>-- | Checks a tensor for NaN and Inf values.</span>
+<a name="line-9108"></a><span class='hs-comment'>--</span>
+<a name="line-9109"></a><span class='hs-comment'>-- When run, reports an `InvalidArgument` error if `tensor` has any values</span>
+<a name="line-9110"></a><span class='hs-comment'>-- that are not a number (NaN) or infinity (Inf). Otherwise, passes `tensor` as-is.</span>
+<a name="line-9111"></a><span class='hs-definition'>checkNumerics</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-9112"></a>                                                      <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9113"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __tensor__</span>
+<a name="line-9114"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-9115"></a><span class='hs-definition'>checkNumerics</span> <span class='hs-varid'>tensor</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9116"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"CheckNumerics"</span>
+<a name="line-9117"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9118"></a>        <span class='hs-varid'>tensor</span>
+<a name="line-9119"></a><span class='hs-comment'>{-
+<a name="line-9120"></a>attr {
+<a name="line-9121"></a>  allowed_values {
+<a name="line-9122"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-9123"></a>  }
+<a name="line-9124"></a>  name: "T"
+<a name="line-9125"></a>  type: "type"
+<a name="line-9126"></a>}
+<a name="line-9127"></a>attr {
+<a name="line-9128"></a>  description: "Prefix of the error message."
+<a name="line-9129"></a>  name: "message"
+<a name="line-9130"></a>  type: "string"
+<a name="line-9131"></a>}
+<a name="line-9132"></a>input_arg { name: "tensor" type_attr: "T" }
+<a name="line-9133"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-9134"></a>-}</span>
+<a name="line-9135"></a>
+<a name="line-9136"></a><span class='hs-comment'>-- | Returns a tensor of zeros with the same shape and type as x.</span>
+<a name="line-9137"></a>
+<a name="line-9138"></a><a name="zerosLike"></a><span class='hs-definition'>zerosLike</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9139"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__: a tensor of type T.</span>
+<a name="line-9140"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__: a tensor of the same shape and type as x but filled with zeros.</span>
+<a name="line-9141"></a><span class='hs-definition'>zerosLike</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9142"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ZerosLike"</span>
+<a name="line-9143"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9144"></a>        <span class='hs-varid'>x</span>
+<a name="line-9145"></a><span class='hs-comment'>{-
+<a name="line-9146"></a>attr { name: "T" type: "type" }
+<a name="line-9147"></a>input_arg {
+<a name="line-9148"></a>  description: "a tensor of type T." name: "x" type_attr: "T"
+<a name="line-9149"></a>}
+<a name="line-9150"></a>output_arg {
+<a name="line-9151"></a>  description: "a tensor of the same shape and type as x but filled with zeros."
+<a name="line-9152"></a>  name: "y"
+<a name="line-9153"></a>  type_attr: "T"
+<a name="line-9154"></a>}
+<a name="line-9155"></a>-}</span>
+<a name="line-9156"></a>
+<a name="line-9157"></a><span class='hs-comment'>-- | Reads and outputs the entire contents of the input filename.</span>
+<a name="line-9158"></a>
+<a name="line-9159"></a><a name="readFile"></a><span class='hs-definition'>readFile</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __filename__</span>
+<a name="line-9160"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __contents__</span>
+<a name="line-9161"></a><span class='hs-definition'>readFile</span> <span class='hs-varid'>filename</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9162"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ReadFile"</span><span class='hs-layout'>)</span>
+<a name="line-9163"></a>        <span class='hs-varid'>filename</span>
+<a name="line-9164"></a><span class='hs-comment'>{-
+<a name="line-9165"></a>input_arg { name: "filename" type: DT_STRING }
+<a name="line-9166"></a>output_arg { name: "contents" type: DT_STRING }
+<a name="line-9167"></a>-}</span>
+<a name="line-9168"></a>
+<a name="line-9169"></a><a name="transpose"></a><span class='hs-comment'>-- | Shuffle dimensions of x according to a permutation.</span>
+<a name="line-9170"></a><span class='hs-comment'>--</span>
+<a name="line-9171"></a><span class='hs-comment'>-- The output `y` has the same rank as `x`. The shapes of `x` and `y` satisfy:</span>
+<a name="line-9172"></a><span class='hs-comment'>--   `y.shape[i] == x.shape[perm[i]] for i in [0, 1, ..., rank(x) - 1]`</span>
+<a name="line-9173"></a><span class='hs-definition'>transpose</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tperm</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tperm</span><span class='hs-layout'>,</span>
+<a name="line-9174"></a>                                     <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-9175"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tperm</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9176"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-9177"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tperm</span> <span class='hs-comment'>-- ^ __perm__</span>
+<a name="line-9178"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-9179"></a><span class='hs-definition'>transpose</span> <span class='hs-varid'>x</span> <span class='hs-varid'>perm</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9180"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Transpose"</span>
+<a name="line-9181"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-9182"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tperm"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tperm</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9183"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>perm</span>
+<a name="line-9184"></a><span class='hs-comment'>{-
+<a name="line-9185"></a>attr { name: "T" type: "type" }
+<a name="line-9186"></a>attr {
+<a name="line-9187"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-9188"></a>  default_value { type: DT_INT32 }
+<a name="line-9189"></a>  name: "Tperm"
+<a name="line-9190"></a>  type: "type"
+<a name="line-9191"></a>}
+<a name="line-9192"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-9193"></a>input_arg { name: "perm" type_attr: "Tperm" }
+<a name="line-9194"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-9195"></a>-}</span>
+<a name="line-9196"></a>
+<a name="line-9197"></a><span class='hs-comment'>-- | Transforms a serialized tensorflow.TensorProto proto into a Tensor.</span>
+<a name="line-9198"></a>
+<a name="line-9199"></a><a name="parseTensor"></a><span class='hs-definition'>parseTensor</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>out_type</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9200"></a>               <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __serialized__: A scalar string containing a serialized TensorProto proto.</span>
+<a name="line-9201"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>out_type</span> <span class='hs-comment'>-- ^ __output__: A Tensor of type `out_type`.</span>
+<a name="line-9202"></a><span class='hs-definition'>parseTensor</span> <span class='hs-varid'>serialized</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9203"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ParseTensor"</span>
+<a name="line-9204"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"out_type"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9205"></a>        <span class='hs-varid'>serialized</span>
+<a name="line-9206"></a><span class='hs-comment'>{-
+<a name="line-9207"></a>attr {
+<a name="line-9208"></a>  description: "The type of the serialized tensor.  The provided type must match the\ntype of the serialized tensor and no implicit conversion will take place."
+<a name="line-9209"></a>  name: "out_type"
+<a name="line-9210"></a>  type: "type"
+<a name="line-9211"></a>}
+<a name="line-9212"></a>input_arg {
+<a name="line-9213"></a>  description: "A scalar string containing a serialized TensorProto proto."
+<a name="line-9214"></a>  name: "serialized"
+<a name="line-9215"></a>  type: DT_STRING
+<a name="line-9216"></a>}
+<a name="line-9217"></a>output_arg {
+<a name="line-9218"></a>  description: "A Tensor of type `out_type`."
+<a name="line-9219"></a>  name: "output"
+<a name="line-9220"></a>  type_attr: "out_type"
+<a name="line-9221"></a>}
+<a name="line-9222"></a>-}</span>
+<a name="line-9223"></a>
+<a name="line-9224"></a><span class='hs-comment'>-- | Computes acos of x element-wise.</span>
+<a name="line-9225"></a>
+<a name="line-9226"></a><a name="acos"></a><span class='hs-definition'>acos</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-9227"></a>                                             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-9228"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-9229"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-9230"></a>                                             <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-9231"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-9232"></a><span class='hs-definition'>acos</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9233"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Acos"</span>
+<a name="line-9234"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9235"></a>        <span class='hs-varid'>x</span>
+<a name="line-9236"></a><span class='hs-comment'>{-
+<a name="line-9237"></a>attr {
+<a name="line-9238"></a>  allowed_values {
+<a name="line-9239"></a>    list {
+<a name="line-9240"></a>      type: DT_HALF
+<a name="line-9241"></a>      type: DT_FLOAT
+<a name="line-9242"></a>      type: DT_DOUBLE
+<a name="line-9243"></a>      type: DT_INT32
+<a name="line-9244"></a>      type: DT_INT64
+<a name="line-9245"></a>      type: DT_COMPLEX64
+<a name="line-9246"></a>      type: DT_COMPLEX128
+<a name="line-9247"></a>    }
+<a name="line-9248"></a>  }
+<a name="line-9249"></a>  name: "T"
+<a name="line-9250"></a>  type: "type"
+<a name="line-9251"></a>}
+<a name="line-9252"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-9253"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-9254"></a>-}</span>
+<a name="line-9255"></a>
+<a name="line-9256"></a><a name="bitcast"></a><span class='hs-comment'>-- | Bitcasts a tensor from one type to another without copying data.</span>
+<a name="line-9257"></a><span class='hs-comment'>--</span>
+<a name="line-9258"></a><span class='hs-comment'>-- Given a tensor `input`, this operation returns a tensor that has the same buffer</span>
+<a name="line-9259"></a><span class='hs-comment'>-- data as `input` with datatype `type`.</span>
+<a name="line-9260"></a><span class='hs-comment'>-- </span>
+<a name="line-9261"></a><span class='hs-comment'>-- If the input datatype `T` is larger than the output datatype `type` then the</span>
+<a name="line-9262"></a><span class='hs-comment'>-- shape changes from [...] to [..., sizeof(`T`)/sizeof(`type`)].</span>
+<a name="line-9263"></a><span class='hs-comment'>-- </span>
+<a name="line-9264"></a><span class='hs-comment'>-- If `T` is smaller than `type`, the operator requires that the rightmost</span>
+<a name="line-9265"></a><span class='hs-comment'>-- dimension be equal to sizeof(`type`)/sizeof(`T`). The shape then goes from</span>
+<a name="line-9266"></a><span class='hs-comment'>-- [..., sizeof(`type`)/sizeof(`T`)] to [...].</span>
+<a name="line-9267"></a><span class='hs-comment'>-- </span>
+<a name="line-9268"></a><span class='hs-comment'>-- *NOTE*: Bitcast is implemented as a low-level cast, so machines with different</span>
+<a name="line-9269"></a><span class='hs-comment'>-- endian orderings will give different results.</span>
+<a name="line-9270"></a><span class='hs-definition'>bitcast</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varid'>type'</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-9271"></a>                                <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-9272"></a>                                        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-9273"></a>                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-9274"></a>                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-9275"></a>                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-9276"></a>                                        <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>type'</span><span class='hs-layout'>,</span>
+<a name="line-9277"></a>                                <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-9278"></a>                                        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-9279"></a>                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-9280"></a>                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-9281"></a>                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-9282"></a>                                        <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>type'</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9283"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-9284"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>type'</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-9285"></a><span class='hs-definition'>bitcast</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9286"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Bitcast"</span>
+<a name="line-9287"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-9288"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"type"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>type'</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9289"></a>        <span class='hs-varid'>input</span>
+<a name="line-9290"></a><span class='hs-comment'>{-
+<a name="line-9291"></a>attr {
+<a name="line-9292"></a>  allowed_values {
+<a name="line-9293"></a>    list {
+<a name="line-9294"></a>      type: DT_FLOAT
+<a name="line-9295"></a>      type: DT_DOUBLE
+<a name="line-9296"></a>      type: DT_INT64
+<a name="line-9297"></a>      type: DT_INT32
+<a name="line-9298"></a>      type: DT_UINT8
+<a name="line-9299"></a>      type: DT_UINT16
+<a name="line-9300"></a>      type: DT_INT16
+<a name="line-9301"></a>      type: DT_INT8
+<a name="line-9302"></a>      type: DT_COMPLEX64
+<a name="line-9303"></a>      type: DT_COMPLEX128
+<a name="line-9304"></a>      type: DT_QINT8
+<a name="line-9305"></a>      type: DT_QUINT8
+<a name="line-9306"></a>      type: DT_QINT32
+<a name="line-9307"></a>      type: DT_HALF
+<a name="line-9308"></a>    }
+<a name="line-9309"></a>  }
+<a name="line-9310"></a>  name: "T"
+<a name="line-9311"></a>  type: "type"
+<a name="line-9312"></a>}
+<a name="line-9313"></a>attr {
+<a name="line-9314"></a>  allowed_values {
+<a name="line-9315"></a>    list {
+<a name="line-9316"></a>      type: DT_FLOAT
+<a name="line-9317"></a>      type: DT_DOUBLE
+<a name="line-9318"></a>      type: DT_INT64
+<a name="line-9319"></a>      type: DT_INT32
+<a name="line-9320"></a>      type: DT_UINT8
+<a name="line-9321"></a>      type: DT_UINT16
+<a name="line-9322"></a>      type: DT_INT16
+<a name="line-9323"></a>      type: DT_INT8
+<a name="line-9324"></a>      type: DT_COMPLEX64
+<a name="line-9325"></a>      type: DT_COMPLEX128
+<a name="line-9326"></a>      type: DT_QINT8
+<a name="line-9327"></a>      type: DT_QUINT8
+<a name="line-9328"></a>      type: DT_QINT32
+<a name="line-9329"></a>      type: DT_HALF
+<a name="line-9330"></a>    }
+<a name="line-9331"></a>  }
+<a name="line-9332"></a>  name: "type"
+<a name="line-9333"></a>  type: "type"
+<a name="line-9334"></a>}
+<a name="line-9335"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-9336"></a>output_arg { name: "output" type_attr: "type" }
+<a name="line-9337"></a>-}</span>
+<a name="line-9338"></a>
+<a name="line-9339"></a><a name="lookupTableImport"></a><span class='hs-comment'>-- | Replaces the contents of the table with the specified keys and values.</span>
+<a name="line-9340"></a><span class='hs-comment'>--</span>
+<a name="line-9341"></a><span class='hs-comment'>-- The tensor `keys` must be of the same type as the keys of the table.</span>
+<a name="line-9342"></a><span class='hs-comment'>-- The tensor `values` must be of the type of the table values.</span>
+<a name="line-9343"></a><span class='hs-definition'>lookupTableImport</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>tin</span> <span class='hs-varid'>tout</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>tin</span><span class='hs-layout'>,</span>
+<a name="line-9344"></a>                                                 <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tout</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9345"></a>                     <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __table_handle__: Handle to the table.</span>
+<a name="line-9346"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tin</span> <span class='hs-comment'>-- ^ __keys__: Any shape.  Keys to look up.</span>
+<a name="line-9347"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>tout</span> <span class='hs-comment'>-- ^ __values__: Values to associate with keys.</span>
+<a name="line-9348"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-9349"></a><span class='hs-definition'>lookupTableImport</span> <span class='hs-varid'>table_handle</span> <span class='hs-varid'>keys</span> <span class='hs-varid'>values</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9350"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"LookupTableImport"</span>
+<a name="line-9351"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tin"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tin</span><span class='hs-layout'>)</span>
+<a name="line-9352"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tout"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tout</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9353"></a>        <span class='hs-varid'>table_handle</span> <span class='hs-varid'>keys</span> <span class='hs-varid'>values</span>
+<a name="line-9354"></a><span class='hs-comment'>{-
+<a name="line-9355"></a>attr { name: "Tin" type: "type" }
+<a name="line-9356"></a>attr { name: "Tout" type: "type" }
+<a name="line-9357"></a>input_arg {
+<a name="line-9358"></a>  description: "Handle to the table."
+<a name="line-9359"></a>  is_ref: true
+<a name="line-9360"></a>  name: "table_handle"
+<a name="line-9361"></a>  type: DT_STRING
+<a name="line-9362"></a>}
+<a name="line-9363"></a>input_arg {
+<a name="line-9364"></a>  description: "Any shape.  Keys to look up."
+<a name="line-9365"></a>  name: "keys"
+<a name="line-9366"></a>  type_attr: "Tin"
+<a name="line-9367"></a>}
+<a name="line-9368"></a>input_arg {
+<a name="line-9369"></a>  description: "Values to associate with keys."
+<a name="line-9370"></a>  name: "values"
+<a name="line-9371"></a>  type_attr: "Tout"
+<a name="line-9372"></a>}
+<a name="line-9373"></a>-}</span>
+<a name="line-9374"></a>
+<a name="line-9375"></a><a name="biasAddGrad"></a><span class='hs-comment'>-- | The backward operation for "BiasAdd" on the "bias" tensor.</span>
+<a name="line-9376"></a><span class='hs-comment'>--</span>
+<a name="line-9377"></a><span class='hs-comment'>-- It accumulates all the values from out_backprop into the feature dimension.</span>
+<a name="line-9378"></a><span class='hs-comment'>-- For NHWC data format, the feature dimension is the last. For NCHW data format,</span>
+<a name="line-9379"></a><span class='hs-comment'>-- the feature dimension is the third-to-last.</span>
+<a name="line-9380"></a><span class='hs-definition'>biasAddGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-9381"></a>                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-9382"></a>                                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-9383"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-9384"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-9385"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-9386"></a>                                      <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9387"></a>               <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out_backprop__: Any number of dimensions.</span>
+<a name="line-9388"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 1-D with size the feature dimension of `out_backprop`.</span>
+<a name="line-9389"></a><span class='hs-definition'>biasAddGrad</span> <span class='hs-varid'>out_backprop</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9390"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BiasAddGrad"</span>
+<a name="line-9391"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9392"></a>        <span class='hs-varid'>out_backprop</span>
+<a name="line-9393"></a><span class='hs-comment'>{-
+<a name="line-9394"></a>attr {
+<a name="line-9395"></a>  allowed_values {
+<a name="line-9396"></a>    list {
+<a name="line-9397"></a>      type: DT_FLOAT
+<a name="line-9398"></a>      type: DT_DOUBLE
+<a name="line-9399"></a>      type: DT_INT64
+<a name="line-9400"></a>      type: DT_INT32
+<a name="line-9401"></a>      type: DT_UINT8
+<a name="line-9402"></a>      type: DT_UINT16
+<a name="line-9403"></a>      type: DT_INT16
+<a name="line-9404"></a>      type: DT_INT8
+<a name="line-9405"></a>      type: DT_COMPLEX64
+<a name="line-9406"></a>      type: DT_COMPLEX128
+<a name="line-9407"></a>      type: DT_QINT8
+<a name="line-9408"></a>      type: DT_QUINT8
+<a name="line-9409"></a>      type: DT_QINT32
+<a name="line-9410"></a>      type: DT_HALF
+<a name="line-9411"></a>    }
+<a name="line-9412"></a>  }
+<a name="line-9413"></a>  name: "T"
+<a name="line-9414"></a>  type: "type"
+<a name="line-9415"></a>}
+<a name="line-9416"></a>attr {
+<a name="line-9417"></a>  allowed_values { list { s: "NHWC" s: "NCHW" } }
+<a name="line-9418"></a>  default_value { s: "NHWC" }
+<a name="line-9419"></a>  description: "Specify the data format of the input and output data. With the\ndefault format \"NHWC\", the bias tensor will be added to the last dimension\nof the value tensor.\nAlternatively, the format could be \"NCHW\", the data storage order of:\n    [batch, in_channels, in_height, in_width].\nThe tensor will be added to \"in_channels\", the third-to-the-last\n    dimension."
+<a name="line-9420"></a>  name: "data_format"
+<a name="line-9421"></a>  type: "string"
+<a name="line-9422"></a>}
+<a name="line-9423"></a>input_arg {
+<a name="line-9424"></a>  description: "Any number of dimensions."
+<a name="line-9425"></a>  name: "out_backprop"
+<a name="line-9426"></a>  type_attr: "T"
+<a name="line-9427"></a>}
+<a name="line-9428"></a>output_arg {
+<a name="line-9429"></a>  description: "1-D with size the feature dimension of `out_backprop`."
+<a name="line-9430"></a>  name: "output"
+<a name="line-9431"></a>  type_attr: "T"
+<a name="line-9432"></a>}
+<a name="line-9433"></a>-}</span>
+<a name="line-9434"></a>
+<a name="line-9435"></a><span class='hs-comment'>-- | </span>
+<a name="line-9436"></a>
+<a name="line-9437"></a><a name="batchSelfAdjointEig"></a><span class='hs-definition'>batchSelfAdjointEig</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9438"></a>                       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-9439"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-9440"></a><span class='hs-definition'>batchSelfAdjointEig</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9441"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchSelfAdjointEig"</span>
+<a name="line-9442"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9443"></a>        <span class='hs-varid'>input</span>
+<a name="line-9444"></a><span class='hs-comment'>{-
+<a name="line-9445"></a>attr {
+<a name="line-9446"></a>  allowed_values { list { type: DT_DOUBLE type: DT_FLOAT } }
+<a name="line-9447"></a>  name: "T"
+<a name="line-9448"></a>  type: "type"
+<a name="line-9449"></a>}
+<a name="line-9450"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-9451"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-9452"></a>-}</span>
+<a name="line-9453"></a>
+<a name="line-9454"></a><a name="prod"></a><span class='hs-comment'>-- | Computes the product of elements across dimensions of a tensor.</span>
+<a name="line-9455"></a><span class='hs-comment'>--</span>
+<a name="line-9456"></a><span class='hs-comment'>-- Reduces `input` along the dimensions given in `reduction_indices`. Unless</span>
+<a name="line-9457"></a><span class='hs-comment'>-- `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in</span>
+<a name="line-9458"></a><span class='hs-comment'>-- `reduction_indices`. If `keep_dims` is true, the reduced dimensions are</span>
+<a name="line-9459"></a><span class='hs-comment'>-- retained with length 1.</span>
+<a name="line-9460"></a><span class='hs-definition'>prod</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-9461"></a>                               <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-9462"></a>                                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-9463"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-9464"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-9465"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-9466"></a>                                       <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span>
+<a name="line-9467"></a>                               <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9468"></a>        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: The tensor to reduce.</span>
+<a name="line-9469"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __reduction_indices__: The dimensions to reduce.</span>
+<a name="line-9470"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The reduced tensor.</span>
+<a name="line-9471"></a><span class='hs-definition'>prod</span> <span class='hs-varid'>input</span> <span class='hs-varid'>reduction_indices</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9472"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Prod"</span>
+<a name="line-9473"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-9474"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9475"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>reduction_indices</span>
+<a name="line-9476"></a><span class='hs-comment'>{-
+<a name="line-9477"></a>attr {
+<a name="line-9478"></a>  default_value { b: false }
+<a name="line-9479"></a>  description: "If true, retain reduced dimensions with length 1."
+<a name="line-9480"></a>  name: "keep_dims"
+<a name="line-9481"></a>  type: "bool"
+<a name="line-9482"></a>}
+<a name="line-9483"></a>attr {
+<a name="line-9484"></a>  allowed_values {
+<a name="line-9485"></a>    list {
+<a name="line-9486"></a>      type: DT_FLOAT
+<a name="line-9487"></a>      type: DT_DOUBLE
+<a name="line-9488"></a>      type: DT_INT64
+<a name="line-9489"></a>      type: DT_INT32
+<a name="line-9490"></a>      type: DT_UINT8
+<a name="line-9491"></a>      type: DT_UINT16
+<a name="line-9492"></a>      type: DT_INT16
+<a name="line-9493"></a>      type: DT_INT8
+<a name="line-9494"></a>      type: DT_COMPLEX64
+<a name="line-9495"></a>      type: DT_COMPLEX128
+<a name="line-9496"></a>      type: DT_QINT8
+<a name="line-9497"></a>      type: DT_QUINT8
+<a name="line-9498"></a>      type: DT_QINT32
+<a name="line-9499"></a>      type: DT_HALF
+<a name="line-9500"></a>    }
+<a name="line-9501"></a>  }
+<a name="line-9502"></a>  name: "T"
+<a name="line-9503"></a>  type: "type"
+<a name="line-9504"></a>}
+<a name="line-9505"></a>attr {
+<a name="line-9506"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-9507"></a>  default_value { type: DT_INT32 }
+<a name="line-9508"></a>  name: "Tidx"
+<a name="line-9509"></a>  type: "type"
+<a name="line-9510"></a>}
+<a name="line-9511"></a>input_arg {
+<a name="line-9512"></a>  description: "The tensor to reduce." name: "input" type_attr: "T"
+<a name="line-9513"></a>}
+<a name="line-9514"></a>input_arg {
+<a name="line-9515"></a>  description: "The dimensions to reduce."
+<a name="line-9516"></a>  name: "reduction_indices"
+<a name="line-9517"></a>  type_attr: "Tidx"
+<a name="line-9518"></a>}
+<a name="line-9519"></a>output_arg {
+<a name="line-9520"></a>  description: "The reduced tensor." name: "output" type_attr: "T"
+<a name="line-9521"></a>}
+<a name="line-9522"></a>-}</span>
+<a name="line-9523"></a>
+<a name="line-9524"></a><a name="resizeBilinear"></a><span class='hs-comment'>-- | Resize `images` to `size` using bilinear interpolation.</span>
+<a name="line-9525"></a><span class='hs-comment'>--</span>
+<a name="line-9526"></a><span class='hs-comment'>-- Input images can be of different types but output images are always float.</span>
+<a name="line-9527"></a><span class='hs-definition'>resizeBilinear</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-9528"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-9529"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-9530"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-9531"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-9532"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-9533"></a>                                                          <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9534"></a>                  <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __images__: 4-D with shape `[batch, height, width, channels]`.</span>
+<a name="line-9535"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __size__: = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The</span>
+<a name="line-9536"></a>                                              <span class='hs-comment'>-- new size for the images.</span>
+<a name="line-9537"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __resized_images__: 4-D with shape</span>
+<a name="line-9538"></a>                  <span class='hs-comment'>-- `[batch, new_height, new_width, channels]`.</span>
+<a name="line-9539"></a><span class='hs-definition'>resizeBilinear</span> <span class='hs-varid'>images</span> <span class='hs-varid'>size</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9540"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ResizeBilinear"</span>
+<a name="line-9541"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9542"></a>        <span class='hs-varid'>images</span> <span class='hs-varid'>size</span>
+<a name="line-9543"></a><span class='hs-comment'>{-
+<a name="line-9544"></a>attr {
+<a name="line-9545"></a>  allowed_values {
+<a name="line-9546"></a>    list {
+<a name="line-9547"></a>      type: DT_UINT8
+<a name="line-9548"></a>      type: DT_INT8
+<a name="line-9549"></a>      type: DT_INT16
+<a name="line-9550"></a>      type: DT_INT32
+<a name="line-9551"></a>      type: DT_INT64
+<a name="line-9552"></a>      type: DT_HALF
+<a name="line-9553"></a>      type: DT_FLOAT
+<a name="line-9554"></a>      type: DT_DOUBLE
+<a name="line-9555"></a>    }
+<a name="line-9556"></a>  }
+<a name="line-9557"></a>  name: "T"
+<a name="line-9558"></a>  type: "type"
+<a name="line-9559"></a>}
+<a name="line-9560"></a>attr {
+<a name="line-9561"></a>  default_value { b: false }
+<a name="line-9562"></a>  description: "If true, rescale input by (new_height - 1) / (height - 1), which\nexactly aligns the 4 corners of images and resized images. If false, rescale\nby new_height / height. Treat similarly the width dimension."
+<a name="line-9563"></a>  name: "align_corners"
+<a name="line-9564"></a>  type: "bool"
+<a name="line-9565"></a>}
+<a name="line-9566"></a>input_arg {
+<a name="line-9567"></a>  description: "4-D with shape `[batch, height, width, channels]`."
+<a name="line-9568"></a>  name: "images"
+<a name="line-9569"></a>  type_attr: "T"
+<a name="line-9570"></a>}
+<a name="line-9571"></a>input_arg {
+<a name="line-9572"></a>  description: "= A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The\nnew size for the images."
+<a name="line-9573"></a>  name: "size"
+<a name="line-9574"></a>  type: DT_INT32
+<a name="line-9575"></a>}
+<a name="line-9576"></a>output_arg {
+<a name="line-9577"></a>  description: "4-D with shape\n`[batch, new_height, new_width, channels]`."
+<a name="line-9578"></a>  name: "resized_images"
+<a name="line-9579"></a>  type: DT_FLOAT
+<a name="line-9580"></a>}
+<a name="line-9581"></a>-}</span>
+<a name="line-9582"></a>
+<a name="line-9583"></a><a name="tensorArrayUnpack"></a><span class='hs-comment'>-- | Unpack the data from the input value into TensorArray elements.</span>
+<a name="line-9584"></a><span class='hs-comment'>--</span>
+<a name="line-9585"></a><span class='hs-comment'>-- **WARNING: This op is deprecated.**</span>
+<a name="line-9586"></a><span class='hs-comment'>-- </span>
+<a name="line-9587"></a><span class='hs-comment'>-- Instead of this op, use `TensorArrayScatter` with</span>
+<a name="line-9588"></a><span class='hs-comment'>-- `indices = RangeOp(0, SizeOp(value)[0])`.</span>
+<a name="line-9589"></a><span class='hs-definition'>tensorArrayUnpack</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9590"></a>                     <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a TensorArray.</span>
+<a name="line-9591"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: The concatenated tensor to write to the TensorArray.</span>
+<a name="line-9592"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __flow_in__: A float scalar that enforces proper chaining of operations.</span>
+<a name="line-9593"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __flow_out__: A float scalar that enforces proper chaining of operations.</span>
+<a name="line-9594"></a><span class='hs-definition'>tensorArrayUnpack</span> <span class='hs-varid'>handle</span> <span class='hs-varid'>value</span> <span class='hs-varid'>flow_in</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9595"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TensorArrayUnpack"</span>
+<a name="line-9596"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9597"></a>        <span class='hs-varid'>handle</span> <span class='hs-varid'>value</span> <span class='hs-varid'>flow_in</span>
+<a name="line-9598"></a><span class='hs-comment'>{-
+<a name="line-9599"></a>attr { name: "T" type: "type" }
+<a name="line-9600"></a>input_arg {
+<a name="line-9601"></a>  description: "The handle to a TensorArray."
+<a name="line-9602"></a>  is_ref: true
+<a name="line-9603"></a>  name: "handle"
+<a name="line-9604"></a>  type: DT_STRING
+<a name="line-9605"></a>}
+<a name="line-9606"></a>input_arg {
+<a name="line-9607"></a>  description: "The concatenated tensor to write to the TensorArray."
+<a name="line-9608"></a>  name: "value"
+<a name="line-9609"></a>  type_attr: "T"
+<a name="line-9610"></a>}
+<a name="line-9611"></a>input_arg {
+<a name="line-9612"></a>  description: "A float scalar that enforces proper chaining of operations."
+<a name="line-9613"></a>  name: "flow_in"
+<a name="line-9614"></a>  type: DT_FLOAT
+<a name="line-9615"></a>}
+<a name="line-9616"></a>output_arg {
+<a name="line-9617"></a>  description: "A float scalar that enforces proper chaining of operations."
+<a name="line-9618"></a>  name: "flow_out"
+<a name="line-9619"></a>  type: DT_FLOAT
+<a name="line-9620"></a>}
+<a name="line-9621"></a>-}</span>
+<a name="line-9622"></a>
+<a name="line-9623"></a><span class='hs-comment'>-- | </span>
+<a name="line-9624"></a>
+<a name="line-9625"></a><a name="batchMatrixDeterminant"></a><span class='hs-definition'>batchMatrixDeterminant</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-9626"></a>                                                               <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9627"></a>                          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-9628"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-9629"></a><span class='hs-definition'>batchMatrixDeterminant</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9630"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchMatrixDeterminant"</span>
+<a name="line-9631"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9632"></a>        <span class='hs-varid'>input</span>
+<a name="line-9633"></a><span class='hs-comment'>{-
+<a name="line-9634"></a>attr {
+<a name="line-9635"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-9636"></a>  name: "T"
+<a name="line-9637"></a>  type: "type"
+<a name="line-9638"></a>}
+<a name="line-9639"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-9640"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-9641"></a>-}</span>
+<a name="line-9642"></a>
+<a name="line-9643"></a><a name="sum"></a><span class='hs-comment'>-- | Computes the sum of elements across dimensions of a tensor.</span>
+<a name="line-9644"></a><span class='hs-comment'>--</span>
+<a name="line-9645"></a><span class='hs-comment'>-- Reduces `input` along the dimensions given in `reduction_indices`. Unless</span>
+<a name="line-9646"></a><span class='hs-comment'>-- `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in</span>
+<a name="line-9647"></a><span class='hs-comment'>-- `reduction_indices`. If `keep_dims` is true, the reduced dimensions are</span>
+<a name="line-9648"></a><span class='hs-comment'>-- retained with length 1.</span>
+<a name="line-9649"></a><span class='hs-definition'>sum</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-9650"></a>                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-9651"></a>                                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-9652"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-9653"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-9654"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-9655"></a>                                      <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span>
+<a name="line-9656"></a>                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9657"></a>       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: The tensor to reduce.</span>
+<a name="line-9658"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __reduction_indices__: The dimensions to reduce.</span>
+<a name="line-9659"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The reduced tensor.</span>
+<a name="line-9660"></a><span class='hs-definition'>sum</span> <span class='hs-varid'>input</span> <span class='hs-varid'>reduction_indices</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9661"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Sum"</span>
+<a name="line-9662"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-9663"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9664"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>reduction_indices</span>
+<a name="line-9665"></a><span class='hs-comment'>{-
+<a name="line-9666"></a>attr {
+<a name="line-9667"></a>  default_value { b: false }
+<a name="line-9668"></a>  description: "If true, retain reduced dimensions with length 1."
+<a name="line-9669"></a>  name: "keep_dims"
+<a name="line-9670"></a>  type: "bool"
+<a name="line-9671"></a>}
+<a name="line-9672"></a>attr {
+<a name="line-9673"></a>  allowed_values {
+<a name="line-9674"></a>    list {
+<a name="line-9675"></a>      type: DT_FLOAT
+<a name="line-9676"></a>      type: DT_DOUBLE
+<a name="line-9677"></a>      type: DT_INT64
+<a name="line-9678"></a>      type: DT_INT32
+<a name="line-9679"></a>      type: DT_UINT8
+<a name="line-9680"></a>      type: DT_UINT16
+<a name="line-9681"></a>      type: DT_INT16
+<a name="line-9682"></a>      type: DT_INT8
+<a name="line-9683"></a>      type: DT_COMPLEX64
+<a name="line-9684"></a>      type: DT_COMPLEX128
+<a name="line-9685"></a>      type: DT_QINT8
+<a name="line-9686"></a>      type: DT_QUINT8
+<a name="line-9687"></a>      type: DT_QINT32
+<a name="line-9688"></a>      type: DT_HALF
+<a name="line-9689"></a>    }
+<a name="line-9690"></a>  }
+<a name="line-9691"></a>  name: "T"
+<a name="line-9692"></a>  type: "type"
+<a name="line-9693"></a>}
+<a name="line-9694"></a>attr {
+<a name="line-9695"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-9696"></a>  default_value { type: DT_INT32 }
+<a name="line-9697"></a>  name: "Tidx"
+<a name="line-9698"></a>  type: "type"
+<a name="line-9699"></a>}
+<a name="line-9700"></a>input_arg {
+<a name="line-9701"></a>  description: "The tensor to reduce." name: "input" type_attr: "T"
+<a name="line-9702"></a>}
+<a name="line-9703"></a>input_arg {
+<a name="line-9704"></a>  description: "The dimensions to reduce."
+<a name="line-9705"></a>  name: "reduction_indices"
+<a name="line-9706"></a>  type_attr: "Tidx"
+<a name="line-9707"></a>}
+<a name="line-9708"></a>output_arg {
+<a name="line-9709"></a>  description: "The reduced tensor." name: "output" type_attr: "T"
+<a name="line-9710"></a>}
+<a name="line-9711"></a>-}</span>
+<a name="line-9712"></a>
+<a name="line-9713"></a><a name="iFFT2D"></a><span class='hs-comment'>-- | Compute the inverse 2-dimensional discrete Fourier Transform over the inner-most</span>
+<a name="line-9714"></a><span class='hs-comment'>--</span>
+<a name="line-9715"></a><span class='hs-comment'>-- 2 dimensions of `input`.</span>
+<a name="line-9716"></a><span class='hs-definition'>iFFT2D</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __input__: A complex64 tensor.</span>
+<a name="line-9717"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __output__: A complex64 tensor of the same shape as `input`. The inner-most 2</span>
+<a name="line-9718"></a>          <span class='hs-comment'>-- dimensions of `input` are replaced with their inverse 2D Fourier Transform.</span>
+<a name="line-9719"></a><span class='hs-definition'>iFFT2D</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9720"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"IFFT2D"</span><span class='hs-layout'>)</span>
+<a name="line-9721"></a>        <span class='hs-varid'>input</span>
+<a name="line-9722"></a><span class='hs-comment'>{-
+<a name="line-9723"></a>input_arg {
+<a name="line-9724"></a>  description: "A complex64 tensor." name: "input" type: DT_COMPLEX64
+<a name="line-9725"></a>}
+<a name="line-9726"></a>output_arg {
+<a name="line-9727"></a>  description: "A complex64 tensor of the same shape as `input`. The inner-most 2\ndimensions of `input` are replaced with their inverse 2D Fourier Transform."
+<a name="line-9728"></a>  name: "output"
+<a name="line-9729"></a>  type: DT_COMPLEX64
+<a name="line-9730"></a>}
+<a name="line-9731"></a>-}</span>
+<a name="line-9732"></a>
+<a name="line-9733"></a><a name="fill"></a><span class='hs-comment'>-- | Creates a tensor filled with a scalar value.</span>
+<a name="line-9734"></a><span class='hs-comment'>--</span>
+<a name="line-9735"></a><span class='hs-comment'>-- This operation creates a tensor of shape `dims` and fills it with `value`.</span>
+<a name="line-9736"></a><span class='hs-comment'>-- </span>
+<a name="line-9737"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-9738"></a><span class='hs-comment'>-- </span>
+<a name="line-9739"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-9740"></a><span class='hs-comment'>-- # Output tensor has shape [2, 3].</span>
+<a name="line-9741"></a><span class='hs-comment'>-- fill([2, 3], 9) ==&gt; [[9, 9, 9]</span>
+<a name="line-9742"></a><span class='hs-comment'>--                      [9, 9, 9]]</span>
+<a name="line-9743"></a><span class='hs-comment'>-- ```</span>
+<a name="line-9744"></a><span class='hs-definition'>fill</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9745"></a>        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __dims__: 1-D. Represents the shape of the output tensor.</span>
+<a name="line-9746"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: 0-D (scalar). Value to fill the returned tensor.</span>
+<a name="line-9747"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-9748"></a><span class='hs-definition'>fill</span> <span class='hs-varid'>dims</span> <span class='hs-varid'>value</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9749"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Fill"</span>
+<a name="line-9750"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9751"></a>        <span class='hs-varid'>dims</span> <span class='hs-varid'>value</span>
+<a name="line-9752"></a><span class='hs-comment'>{-
+<a name="line-9753"></a>attr { name: "T" type: "type" }
+<a name="line-9754"></a>input_arg {
+<a name="line-9755"></a>  description: "1-D. Represents the shape of the output tensor."
+<a name="line-9756"></a>  name: "dims"
+<a name="line-9757"></a>  type: DT_INT32
+<a name="line-9758"></a>}
+<a name="line-9759"></a>input_arg {
+<a name="line-9760"></a>  description: "0-D (scalar). Value to fill the returned tensor."
+<a name="line-9761"></a>  name: "value"
+<a name="line-9762"></a>  type_attr: "T"
+<a name="line-9763"></a>}
+<a name="line-9764"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-9765"></a>-}</span>
+<a name="line-9766"></a>
+<a name="line-9767"></a><a name="fixedUnigramCandidateSampler"></a><span class='hs-comment'>-- | Generates labels for candidate sampling with a learned unigram distribution.</span>
+<a name="line-9768"></a><span class='hs-comment'>--</span>
+<a name="line-9769"></a><span class='hs-comment'>-- A unigram sampler could use a fixed unigram distribution read from a</span>
+<a name="line-9770"></a><span class='hs-comment'>-- file or passed in as an in-memory array instead of building up the distribution</span>
+<a name="line-9771"></a><span class='hs-comment'>-- from data on the fly. There is also an option to skew the distribution by</span>
+<a name="line-9772"></a><span class='hs-comment'>-- applying a distortion power to the weights.</span>
+<a name="line-9773"></a><span class='hs-comment'>-- </span>
+<a name="line-9774"></a><span class='hs-comment'>-- The vocabulary file should be in CSV-like format, with the last field</span>
+<a name="line-9775"></a><span class='hs-comment'>-- being the weight associated with the word.</span>
+<a name="line-9776"></a><span class='hs-comment'>-- </span>
+<a name="line-9777"></a><span class='hs-comment'>-- For each batch, this op picks a single set of sampled candidate labels.</span>
+<a name="line-9778"></a><span class='hs-comment'>-- </span>
+<a name="line-9779"></a><span class='hs-comment'>-- The advantages of sampling candidates per-batch are simplicity and the</span>
+<a name="line-9780"></a><span class='hs-comment'>-- possibility of efficient dense matrix multiplication. The disadvantage is that</span>
+<a name="line-9781"></a><span class='hs-comment'>-- the sampled candidates must be chosen independently of the context and of the</span>
+<a name="line-9782"></a><span class='hs-comment'>-- true labels.</span>
+<a name="line-9783"></a><span class='hs-definition'>fixedUnigramCandidateSampler</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_sampled__: Number of candidates to randomly sample per batch.</span>
+<a name="line-9784"></a>                                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_true__: Number of true labels per context.</span>
+<a name="line-9785"></a>                                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __range_max__: The sampler will sample integers from the interval [0, range_max).</span>
+<a name="line-9786"></a>                                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __unique__: If unique is true, we sample with rejection, so that all sampled</span>
+<a name="line-9787"></a>                                        <span class='hs-comment'>-- candidates in a batch are unique. This requires some approximation to</span>
+<a name="line-9788"></a>                                        <span class='hs-comment'>-- estimate the post-rejection sampling probabilities.</span>
+<a name="line-9789"></a>                                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __true_classes__: A batch_size * num_true matrix, in which each row contains the</span>
+<a name="line-9790"></a>                                                            <span class='hs-comment'>-- IDs of the num_true target_classes in the corresponding original label.</span>
+<a name="line-9791"></a>                                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-9792"></a>                                    <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span>
+<a name="line-9793"></a>                                <span class='hs-comment'>-- ^ (__sampled_candidates__, __true_expected_count__, __sampled_expected_count__)</span>
+<a name="line-9794"></a>                                <span class='hs-comment'>--</span>
+<a name="line-9795"></a>                                <span class='hs-comment'>-- * __sampled_candidates__: A vector of length num_sampled, in which each element is</span>
+<a name="line-9796"></a>                                <span class='hs-comment'>-- the ID of a sampled candidate.</span>
+<a name="line-9797"></a>                                <span class='hs-comment'>--</span>
+<a name="line-9798"></a>                                <span class='hs-comment'>-- * __true_expected_count__: A batch_size * num_true matrix, representing</span>
+<a name="line-9799"></a>                                <span class='hs-comment'>-- the number of times each candidate is expected to occur in a batch</span>
+<a name="line-9800"></a>                                <span class='hs-comment'>-- of sampled candidates. If unique=true, then this is a probability.</span>
+<a name="line-9801"></a>                                <span class='hs-comment'>--</span>
+<a name="line-9802"></a>                                <span class='hs-comment'>-- * __sampled_expected_count__: A vector of length num_sampled, for each sampled</span>
+<a name="line-9803"></a>                                <span class='hs-comment'>-- candidate representing the number of times the candidate is expected</span>
+<a name="line-9804"></a>                                <span class='hs-comment'>-- to occur in a batch of sampled candidates.  If unique=true, then this is a</span>
+<a name="line-9805"></a>                                <span class='hs-comment'>-- probability.</span>
+<a name="line-9806"></a><span class='hs-definition'>fixedUnigramCandidateSampler</span> <span class='hs-varid'>num_sampled</span> <span class='hs-varid'>num_true</span> <span class='hs-varid'>range_max</span> <span class='hs-varid'>unique</span>
+<a name="line-9807"></a>                             <span class='hs-varid'>true_classes</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9808"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"FixedUnigramCandidateSampler"</span>
+<a name="line-9809"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_sampled"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_sampled</span>
+<a name="line-9810"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_true"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_true</span>
+<a name="line-9811"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"range_max"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>range_max</span>
+<a name="line-9812"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"unique"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>unique</span><span class='hs-layout'>)</span>
+<a name="line-9813"></a>        <span class='hs-varid'>true_classes</span>
+<a name="line-9814"></a><span class='hs-comment'>{-
+<a name="line-9815"></a>attr {
+<a name="line-9816"></a>  description: "Number of true labels per context."
+<a name="line-9817"></a>  has_minimum: true
+<a name="line-9818"></a>  minimum: 1
+<a name="line-9819"></a>  name: "num_true"
+<a name="line-9820"></a>  type: "int"
+<a name="line-9821"></a>}
+<a name="line-9822"></a>attr {
+<a name="line-9823"></a>  description: "Number of candidates to randomly sample per batch."
+<a name="line-9824"></a>  has_minimum: true
+<a name="line-9825"></a>  minimum: 1
+<a name="line-9826"></a>  name: "num_sampled"
+<a name="line-9827"></a>  type: "int"
+<a name="line-9828"></a>}
+<a name="line-9829"></a>attr {
+<a name="line-9830"></a>  description: "If unique is true, we sample with rejection, so that all sampled\ncandidates in a batch are unique. This requires some approximation to\nestimate the post-rejection sampling probabilities."
+<a name="line-9831"></a>  name: "unique"
+<a name="line-9832"></a>  type: "bool"
+<a name="line-9833"></a>}
+<a name="line-9834"></a>attr {
+<a name="line-9835"></a>  description: "The sampler will sample integers from the interval [0, range_max)."
+<a name="line-9836"></a>  has_minimum: true
+<a name="line-9837"></a>  minimum: 1
+<a name="line-9838"></a>  name: "range_max"
+<a name="line-9839"></a>  type: "int"
+<a name="line-9840"></a>}
+<a name="line-9841"></a>attr {
+<a name="line-9842"></a>  default_value { s: "" }
+<a name="line-9843"></a>  description: "Each valid line in this file (which should have a CSV-like format)\ncorresponds to a valid word ID. IDs are in sequential order, starting from\nnum_reserved_ids. The last entry in each line is expected to be a value\ncorresponding to the count or relative probability. Exactly one of vocab_file\nand unigrams needs to be passed to this op."
+<a name="line-9844"></a>  name: "vocab_file"
+<a name="line-9845"></a>  type: "string"
+<a name="line-9846"></a>}
+<a name="line-9847"></a>attr {
+<a name="line-9848"></a>  default_value { f: 1.0 }
+<a name="line-9849"></a>  description: "The distortion is used to skew the unigram probability distribution.\nEach weight is first raised to the distortion\'s power before adding to the\ninternal unigram distribution. As a result, distortion = 1.0 gives regular\nunigram sampling (as defined by the vocab file), and distortion = 0.0 gives\na uniform distribution."
+<a name="line-9850"></a>  name: "distortion"
+<a name="line-9851"></a>  type: "float"
+<a name="line-9852"></a>}
+<a name="line-9853"></a>attr {
+<a name="line-9854"></a>  default_value { i: 0 }
+<a name="line-9855"></a>  description: "Optionally some reserved IDs can be added in the range [0,\n..., num_reserved_ids) by the users. One use case is that a special unknown\nword token is used as ID 0. These IDs will have a sampling probability of 0."
+<a name="line-9856"></a>  name: "num_reserved_ids"
+<a name="line-9857"></a>  type: "int"
+<a name="line-9858"></a>}
+<a name="line-9859"></a>attr {
+<a name="line-9860"></a>  default_value { i: 1 }
+<a name="line-9861"></a>  description: "A sampler can be used to sample from a subset of the original range\nin order to speed up the whole computation through parallelism. This parameter\n(together with \'shard\') indicates the number of partitions that are being\nused in the overall computation."
+<a name="line-9862"></a>  has_minimum: true
+<a name="line-9863"></a>  minimum: 1
+<a name="line-9864"></a>  name: "num_shards"
+<a name="line-9865"></a>  type: "int"
+<a name="line-9866"></a>}
+<a name="line-9867"></a>attr {
+<a name="line-9868"></a>  default_value { i: 0 }
+<a name="line-9869"></a>  description: "A sampler can be used to sample from a subset of the original range\nin order to speed up the whole computation through parallelism. This parameter\n(together with \'num_shards\') indicates the particular partition number of a\nsampler op, when partitioning is being used."
+<a name="line-9870"></a>  has_minimum: true
+<a name="line-9871"></a>  name: "shard"
+<a name="line-9872"></a>  type: "int"
+<a name="line-9873"></a>}
+<a name="line-9874"></a>attr {
+<a name="line-9875"></a>  default_value { list { } }
+<a name="line-9876"></a>  description: "A list of unigram counts or probabilities, one per ID in sequential\norder. Exactly one of vocab_file and unigrams should be passed to this op."
+<a name="line-9877"></a>  name: "unigrams"
+<a name="line-9878"></a>  type: "list(float)"
+<a name="line-9879"></a>}
+<a name="line-9880"></a>attr {
+<a name="line-9881"></a>  default_value { i: 0 }
+<a name="line-9882"></a>  description: "If either seed or seed2 are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-9883"></a>  name: "seed"
+<a name="line-9884"></a>  type: "int"
+<a name="line-9885"></a>}
+<a name="line-9886"></a>attr {
+<a name="line-9887"></a>  default_value { i: 0 }
+<a name="line-9888"></a>  description: "An second seed to avoid seed collision."
+<a name="line-9889"></a>  name: "seed2"
+<a name="line-9890"></a>  type: "int"
+<a name="line-9891"></a>}
+<a name="line-9892"></a>input_arg {
+<a name="line-9893"></a>  description: "A batch_size * num_true matrix, in which each row contains the\nIDs of the num_true target_classes in the corresponding original label."
+<a name="line-9894"></a>  name: "true_classes"
+<a name="line-9895"></a>  type: DT_INT64
+<a name="line-9896"></a>}
+<a name="line-9897"></a>output_arg {
+<a name="line-9898"></a>  description: "A vector of length num_sampled, in which each element is\nthe ID of a sampled candidate."
+<a name="line-9899"></a>  name: "sampled_candidates"
+<a name="line-9900"></a>  type: DT_INT64
+<a name="line-9901"></a>}
+<a name="line-9902"></a>output_arg {
+<a name="line-9903"></a>  description: "A batch_size * num_true matrix, representing\nthe number of times each candidate is expected to occur in a batch\nof sampled candidates. If unique=true, then this is a probability."
+<a name="line-9904"></a>  name: "true_expected_count"
+<a name="line-9905"></a>  type: DT_FLOAT
+<a name="line-9906"></a>}
+<a name="line-9907"></a>output_arg {
+<a name="line-9908"></a>  description: "A vector of length num_sampled, for each sampled\ncandidate representing the number of times the candidate is expected\nto occur in a batch of sampled candidates.  If unique=true, then this is a\nprobability."
+<a name="line-9909"></a>  name: "sampled_expected_count"
+<a name="line-9910"></a>  type: DT_FLOAT
+<a name="line-9911"></a>}
+<a name="line-9912"></a>-}</span>
+<a name="line-9913"></a>
+<a name="line-9914"></a><a name="dilation2D"></a><span class='hs-comment'>-- | Computes the grayscale dilation of 4-D `input` and 3-D `filter` tensors.</span>
+<a name="line-9915"></a><span class='hs-comment'>--</span>
+<a name="line-9916"></a><span class='hs-comment'>-- The `input` tensor has shape `[batch, in_height, in_width, depth]` and the</span>
+<a name="line-9917"></a><span class='hs-comment'>-- `filter` tensor has shape `[filter_height, filter_width, depth]`, i.e., each</span>
+<a name="line-9918"></a><span class='hs-comment'>-- input channel is processed independently of the others with its own structuring</span>
+<a name="line-9919"></a><span class='hs-comment'>-- function. The `output` tensor has shape</span>
+<a name="line-9920"></a><span class='hs-comment'>-- `[batch, out_height, out_width, depth]`. The spatial dimensions of the output</span>
+<a name="line-9921"></a><span class='hs-comment'>-- tensor depend on the `padding` algorithm. We currently only support the default</span>
+<a name="line-9922"></a><span class='hs-comment'>-- "NHWC" `data_format`.</span>
+<a name="line-9923"></a><span class='hs-comment'>-- </span>
+<a name="line-9924"></a><span class='hs-comment'>-- In detail, the grayscale morphological 2-D dilation is the max-sum correlation</span>
+<a name="line-9925"></a><span class='hs-comment'>-- (for consistency with `conv2d`, we use unmirrored filters):</span>
+<a name="line-9926"></a><span class='hs-comment'>-- </span>
+<a name="line-9927"></a><span class='hs-comment'>--     output[b, y, x, c] =</span>
+<a name="line-9928"></a><span class='hs-comment'>--        max_{dy, dx} input[b,</span>
+<a name="line-9929"></a><span class='hs-comment'>--                           strides[1] * y + rates[1] * dy,</span>
+<a name="line-9930"></a><span class='hs-comment'>--                           strides[2] * x + rates[2] * dx,</span>
+<a name="line-9931"></a><span class='hs-comment'>--                           c] +</span>
+<a name="line-9932"></a><span class='hs-comment'>--                     filter[dy, dx, c]</span>
+<a name="line-9933"></a><span class='hs-comment'>-- </span>
+<a name="line-9934"></a><span class='hs-comment'>-- Max-pooling is a special case when the filter has size equal to the pooling</span>
+<a name="line-9935"></a><span class='hs-comment'>-- kernel size and contains all zeros.</span>
+<a name="line-9936"></a><span class='hs-comment'>-- </span>
+<a name="line-9937"></a><span class='hs-comment'>-- Note on duality: The dilation of `input` by the `filter` is equal to the</span>
+<a name="line-9938"></a><span class='hs-comment'>-- negation of the erosion of `-input` by the reflected `filter`.</span>
+<a name="line-9939"></a><span class='hs-definition'>dilation2D</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-9940"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-9941"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-9942"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-9943"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-9944"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-9945"></a>                                                      <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-9946"></a>              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: 4-D with shape `[batch, in_height, in_width, depth]`.</span>
+<a name="line-9947"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __filter__: 3-D with shape `[filter_height, filter_width, depth]`.</span>
+<a name="line-9948"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 4-D with shape `[batch, out_height, out_width, depth]`.</span>
+<a name="line-9949"></a><span class='hs-definition'>dilation2D</span> <span class='hs-varid'>input</span> <span class='hs-varid'>filter</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-9950"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Dilation2D"</span>
+<a name="line-9951"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9952"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>filter</span>
+<a name="line-9953"></a><span class='hs-comment'>{-
+<a name="line-9954"></a>attr {
+<a name="line-9955"></a>  allowed_values {
+<a name="line-9956"></a>    list {
+<a name="line-9957"></a>      type: DT_FLOAT
+<a name="line-9958"></a>      type: DT_DOUBLE
+<a name="line-9959"></a>      type: DT_INT32
+<a name="line-9960"></a>      type: DT_INT64
+<a name="line-9961"></a>      type: DT_UINT8
+<a name="line-9962"></a>      type: DT_INT16
+<a name="line-9963"></a>      type: DT_INT8
+<a name="line-9964"></a>      type: DT_UINT16
+<a name="line-9965"></a>      type: DT_HALF
+<a name="line-9966"></a>    }
+<a name="line-9967"></a>  }
+<a name="line-9968"></a>  name: "T"
+<a name="line-9969"></a>  type: "type"
+<a name="line-9970"></a>}
+<a name="line-9971"></a>attr {
+<a name="line-9972"></a>  description: "The stride of the sliding window for each dimension of the input\ntensor. Must be: `[1, stride_height, stride_width, 1]`."
+<a name="line-9973"></a>  has_minimum: true
+<a name="line-9974"></a>  minimum: 4
+<a name="line-9975"></a>  name: "strides"
+<a name="line-9976"></a>  type: "list(int)"
+<a name="line-9977"></a>}
+<a name="line-9978"></a>attr {
+<a name="line-9979"></a>  description: "The input stride for atrous morphological dilation. Must be:\n`[1, rate_height, rate_width, 1]`."
+<a name="line-9980"></a>  has_minimum: true
+<a name="line-9981"></a>  minimum: 4
+<a name="line-9982"></a>  name: "rates"
+<a name="line-9983"></a>  type: "list(int)"
+<a name="line-9984"></a>}
+<a name="line-9985"></a>attr {
+<a name="line-9986"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-9987"></a>  description: "The type of padding algorithm to use."
+<a name="line-9988"></a>  name: "padding"
+<a name="line-9989"></a>  type: "string"
+<a name="line-9990"></a>}
+<a name="line-9991"></a>input_arg {
+<a name="line-9992"></a>  description: "4-D with shape `[batch, in_height, in_width, depth]`."
+<a name="line-9993"></a>  name: "input"
+<a name="line-9994"></a>  type_attr: "T"
+<a name="line-9995"></a>}
+<a name="line-9996"></a>input_arg {
+<a name="line-9997"></a>  description: "3-D with shape `[filter_height, filter_width, depth]`."
+<a name="line-9998"></a>  name: "filter"
+<a name="line-9999"></a>  type_attr: "T"
+<a name="line-10000"></a>}
+<a name="line-10001"></a>output_arg {
+<a name="line-10002"></a>  description: "4-D with shape `[batch, out_height, out_width, depth]`."
+<a name="line-10003"></a>  name: "output"
+<a name="line-10004"></a>  type_attr: "T"
+<a name="line-10005"></a>}
+<a name="line-10006"></a>-}</span>
+<a name="line-10007"></a>
+<a name="line-10008"></a><a name="polygamma"></a><span class='hs-comment'>-- | Compute the polygamma function \\(\psi^{(n)}(x)\\).</span>
+<a name="line-10009"></a><span class='hs-comment'>--</span>
+<a name="line-10010"></a><span class='hs-comment'>-- The polygamma function is defined as:</span>
+<a name="line-10011"></a><span class='hs-comment'>-- </span>
+<a name="line-10012"></a><span class='hs-comment'>-- ```</span>
+<a name="line-10013"></a><span class='hs-comment'>-- \psi^{(n)}(x) = \frac{d^n}{dx^n} \psi(x)</span>
+<a name="line-10014"></a><span class='hs-comment'>-- ```</span>
+<a name="line-10015"></a><span class='hs-comment'>-- where \\(\psi(x)\\) is the digamma function.</span>
+<a name="line-10016"></a><span class='hs-definition'>polygamma</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-10017"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __a__</span>
+<a name="line-10018"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-10019"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-10020"></a><span class='hs-definition'>polygamma</span> <span class='hs-varid'>a</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10021"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Polygamma"</span>
+<a name="line-10022"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-10023"></a>        <span class='hs-varid'>a</span> <span class='hs-varid'>x</span>
+<a name="line-10024"></a><span class='hs-comment'>{-
+<a name="line-10025"></a>attr {
+<a name="line-10026"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-10027"></a>  name: "T"
+<a name="line-10028"></a>  type: "type"
+<a name="line-10029"></a>}
+<a name="line-10030"></a>input_arg { name: "a" type_attr: "T" }
+<a name="line-10031"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-10032"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-10033"></a>-}</span>
+<a name="line-10034"></a>
+<a name="line-10035"></a><span class='hs-comment'>-- | Return the same ref tensor as the input ref tensor.</span>
+<a name="line-10036"></a>
+<a name="line-10037"></a><a name="refIdentity"></a><span class='hs-definition'>refIdentity</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-10038"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-10039"></a><span class='hs-definition'>refIdentity</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10040"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RefIdentity"</span>
+<a name="line-10041"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-10042"></a>        <span class='hs-varid'>input</span>
+<a name="line-10043"></a><span class='hs-comment'>{-
+<a name="line-10044"></a>attr { name: "T" type: "type" }
+<a name="line-10045"></a>input_arg { is_ref: true name: "input" type_attr: "T" }
+<a name="line-10046"></a>output_arg { is_ref: true name: "output" type_attr: "T" }
+<a name="line-10047"></a>-}</span>
+<a name="line-10048"></a>
+<a name="line-10049"></a><a name="encodePng"></a><span class='hs-comment'>-- | PNG-encode an image.</span>
+<a name="line-10050"></a><span class='hs-comment'>--</span>
+<a name="line-10051"></a><span class='hs-comment'>-- `image` is a 3-D uint8 or uint16 Tensor of shape `[height, width, channels]`</span>
+<a name="line-10052"></a><span class='hs-comment'>-- where `channels` is:</span>
+<a name="line-10053"></a><span class='hs-comment'>-- </span>
+<a name="line-10054"></a><span class='hs-comment'>-- *   1: for grayscale.</span>
+<a name="line-10055"></a><span class='hs-comment'>-- *   2: for grayscale + alpha.</span>
+<a name="line-10056"></a><span class='hs-comment'>-- *   3: for RGB.</span>
+<a name="line-10057"></a><span class='hs-comment'>-- *   4: for RGBA.</span>
+<a name="line-10058"></a><span class='hs-comment'>-- </span>
+<a name="line-10059"></a><span class='hs-comment'>-- The ZLIB compression level, `compression`, can be -1 for the PNG-encoder</span>
+<a name="line-10060"></a><span class='hs-comment'>-- default or a value from 0 to 9.  9 is the highest compression level, generating</span>
+<a name="line-10061"></a><span class='hs-comment'>-- the smallest output, but is slower.</span>
+<a name="line-10062"></a><span class='hs-definition'>encodePng</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-10063"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-10064"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __image__: 3-D with shape `[height, width, channels]`.</span>
+<a name="line-10065"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __contents__: 0-D. PNG-encoded image.</span>
+<a name="line-10066"></a><span class='hs-definition'>encodePng</span> <span class='hs-varid'>image</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10067"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"EncodePng"</span>
+<a name="line-10068"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-10069"></a>        <span class='hs-varid'>image</span>
+<a name="line-10070"></a><span class='hs-comment'>{-
+<a name="line-10071"></a>attr {
+<a name="line-10072"></a>  default_value { i: -1 }
+<a name="line-10073"></a>  description: "Compression level."
+<a name="line-10074"></a>  name: "compression"
+<a name="line-10075"></a>  type: "int"
+<a name="line-10076"></a>}
+<a name="line-10077"></a>attr {
+<a name="line-10078"></a>  allowed_values { list { type: DT_UINT8 type: DT_UINT16 } }
+<a name="line-10079"></a>  default_value { type: DT_UINT8 }
+<a name="line-10080"></a>  name: "T"
+<a name="line-10081"></a>  type: "type"
+<a name="line-10082"></a>}
+<a name="line-10083"></a>input_arg {
+<a name="line-10084"></a>  description: "3-D with shape `[height, width, channels]`."
+<a name="line-10085"></a>  name: "image"
+<a name="line-10086"></a>  type_attr: "T"
+<a name="line-10087"></a>}
+<a name="line-10088"></a>output_arg {
+<a name="line-10089"></a>  description: "0-D. PNG-encoded image."
+<a name="line-10090"></a>  name: "contents"
+<a name="line-10091"></a>  type: DT_STRING
+<a name="line-10092"></a>}
+<a name="line-10093"></a>-}</span>
+<a name="line-10094"></a>
+<a name="line-10095"></a><a name="lookupTableInsert"></a><span class='hs-comment'>-- | Updates the table to associates keys with values.</span>
+<a name="line-10096"></a><span class='hs-comment'>--</span>
+<a name="line-10097"></a><span class='hs-comment'>-- The tensor `keys` must be of the same type as the keys of the table.</span>
+<a name="line-10098"></a><span class='hs-comment'>-- The tensor `values` must be of the type of the table values.</span>
+<a name="line-10099"></a><span class='hs-definition'>lookupTableInsert</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>tin</span> <span class='hs-varid'>tout</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>tin</span><span class='hs-layout'>,</span>
+<a name="line-10100"></a>                                                 <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tout</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-10101"></a>                     <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __table_handle__: Handle to the table.</span>
+<a name="line-10102"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tin</span> <span class='hs-comment'>-- ^ __keys__: Any shape.  Keys to look up.</span>
+<a name="line-10103"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>tout</span> <span class='hs-comment'>-- ^ __values__: Values to associate with keys.</span>
+<a name="line-10104"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-10105"></a><span class='hs-definition'>lookupTableInsert</span> <span class='hs-varid'>table_handle</span> <span class='hs-varid'>keys</span> <span class='hs-varid'>values</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10106"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"LookupTableInsert"</span>
+<a name="line-10107"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tin"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tin</span><span class='hs-layout'>)</span>
+<a name="line-10108"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tout"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tout</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-10109"></a>        <span class='hs-varid'>table_handle</span> <span class='hs-varid'>keys</span> <span class='hs-varid'>values</span>
+<a name="line-10110"></a><span class='hs-comment'>{-
+<a name="line-10111"></a>attr { name: "Tin" type: "type" }
+<a name="line-10112"></a>attr { name: "Tout" type: "type" }
+<a name="line-10113"></a>input_arg {
+<a name="line-10114"></a>  description: "Handle to the table."
+<a name="line-10115"></a>  is_ref: true
+<a name="line-10116"></a>  name: "table_handle"
+<a name="line-10117"></a>  type: DT_STRING
+<a name="line-10118"></a>}
+<a name="line-10119"></a>input_arg {
+<a name="line-10120"></a>  description: "Any shape.  Keys to look up."
+<a name="line-10121"></a>  name: "keys"
+<a name="line-10122"></a>  type_attr: "Tin"
+<a name="line-10123"></a>}
+<a name="line-10124"></a>input_arg {
+<a name="line-10125"></a>  description: "Values to associate with keys."
+<a name="line-10126"></a>  name: "values"
+<a name="line-10127"></a>  type_attr: "Tout"
+<a name="line-10128"></a>}
+<a name="line-10129"></a>-}</span>
+<a name="line-10130"></a>
+<a name="line-10131"></a><span class='hs-comment'>-- | </span>
+<a name="line-10132"></a>
+<a name="line-10133"></a><a name="batchIFFT2D"></a><span class='hs-definition'>batchIFFT2D</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-10134"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-10135"></a><span class='hs-definition'>batchIFFT2D</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10136"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchIFFT2D"</span><span class='hs-layout'>)</span>
+<a name="line-10137"></a>        <span class='hs-varid'>input</span>
+<a name="line-10138"></a><span class='hs-comment'>{-
+<a name="line-10139"></a>input_arg { name: "input" type: DT_COMPLEX64 }
+<a name="line-10140"></a>output_arg { name: "output" type: DT_COMPLEX64 }
+<a name="line-10141"></a>-}</span>
+<a name="line-10142"></a>
+<a name="line-10143"></a><a name="uniqueWithCounts"></a><span class='hs-comment'>-- | Finds unique elements in a 1-D tensor.</span>
+<a name="line-10144"></a><span class='hs-comment'>--</span>
+<a name="line-10145"></a><span class='hs-comment'>-- This operation returns a tensor `y` containing all of the unique elements of `x`</span>
+<a name="line-10146"></a><span class='hs-comment'>-- sorted in the same order that they occur in `x`. This operation also returns a</span>
+<a name="line-10147"></a><span class='hs-comment'>-- tensor `idx` the same size as `x` that contains the index of each value of `x`</span>
+<a name="line-10148"></a><span class='hs-comment'>-- in the unique output `y`. Finally, it returns a third tensor `count` that</span>
+<a name="line-10149"></a><span class='hs-comment'>-- contains the count of each element of `y` in `x`. In other words:</span>
+<a name="line-10150"></a><span class='hs-comment'>-- </span>
+<a name="line-10151"></a><span class='hs-comment'>-- `y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`</span>
+<a name="line-10152"></a><span class='hs-comment'>-- </span>
+<a name="line-10153"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-10154"></a><span class='hs-comment'>-- </span>
+<a name="line-10155"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10156"></a><span class='hs-comment'>-- # tensor 'x' is [1, 1, 2, 4, 4, 4, 7, 8, 8]</span>
+<a name="line-10157"></a><span class='hs-comment'>-- y, idx, count = unique_with_counts(x)</span>
+<a name="line-10158"></a><span class='hs-comment'>-- y ==&gt; [1, 2, 4, 7, 8]</span>
+<a name="line-10159"></a><span class='hs-comment'>-- idx ==&gt; [0, 0, 1, 2, 2, 2, 3, 4, 4]</span>
+<a name="line-10160"></a><span class='hs-comment'>-- count ==&gt; [2, 1, 3, 1, 2]</span>
+<a name="line-10161"></a><span class='hs-comment'>-- ```</span>
+<a name="line-10162"></a><span class='hs-definition'>uniqueWithCounts</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varid'>out_idx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>out_idx</span><span class='hs-layout'>,</span>
+<a name="line-10163"></a>                                           <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-10164"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>out_idx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-10165"></a>                    <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__: 1-D.</span>
+<a name="line-10166"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>out_idx</span><span class='hs-layout'>,</span>
+<a name="line-10167"></a>                        <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>out_idx</span><span class='hs-layout'>)</span>
+<a name="line-10168"></a>                    <span class='hs-comment'>-- ^ (__y__, __idx__, __count__)</span>
+<a name="line-10169"></a>                    <span class='hs-comment'>--</span>
+<a name="line-10170"></a>                    <span class='hs-comment'>-- * __y__: 1-D.</span>
+<a name="line-10171"></a>                    <span class='hs-comment'>--</span>
+<a name="line-10172"></a>                    <span class='hs-comment'>-- * __idx__: 1-D.</span>
+<a name="line-10173"></a>                    <span class='hs-comment'>--</span>
+<a name="line-10174"></a>                    <span class='hs-comment'>-- * __count__: 1-D.</span>
+<a name="line-10175"></a><span class='hs-definition'>uniqueWithCounts</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10176"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"UniqueWithCounts"</span>
+<a name="line-10177"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-10178"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"out_idx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>out_idx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-10179"></a>        <span class='hs-varid'>x</span>
+<a name="line-10180"></a><span class='hs-comment'>{-
+<a name="line-10181"></a>attr { name: "T" type: "type" }
+<a name="line-10182"></a>attr {
+<a name="line-10183"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-10184"></a>  default_value { type: DT_INT32 }
+<a name="line-10185"></a>  name: "out_idx"
+<a name="line-10186"></a>  type: "type"
+<a name="line-10187"></a>}
+<a name="line-10188"></a>input_arg { description: "1-D." name: "x" type_attr: "T" }
+<a name="line-10189"></a>output_arg { description: "1-D." name: "y" type_attr: "T" }
+<a name="line-10190"></a>output_arg { description: "1-D." name: "idx" type_attr: "out_idx" }
+<a name="line-10191"></a>output_arg {
+<a name="line-10192"></a>  description: "1-D." name: "count" type_attr: "out_idx"
+<a name="line-10193"></a>}
+<a name="line-10194"></a>-}</span>
+<a name="line-10195"></a>
+<a name="line-10196"></a><a name="gatherNd"></a><span class='hs-comment'>-- | Gather values or slices from `params` according to `indices`.</span>
+<a name="line-10197"></a><span class='hs-comment'>--</span>
+<a name="line-10198"></a><span class='hs-comment'>-- `params` is a Tensor of rank `R` and `indices` is a Tensor of rank `M`.</span>
+<a name="line-10199"></a><span class='hs-comment'>-- </span>
+<a name="line-10200"></a><span class='hs-comment'>-- `indices` must be integer tensor, containing indices into `params`.</span>
+<a name="line-10201"></a><span class='hs-comment'>-- It must be shape `[d_0, ..., d_N, R]` where `0 &lt; R &lt;= M`.</span>
+<a name="line-10202"></a><span class='hs-comment'>-- </span>
+<a name="line-10203"></a><span class='hs-comment'>-- The innermost dimension of `indices` (with length `R`) corresponds to</span>
+<a name="line-10204"></a><span class='hs-comment'>-- indices into elements (if `R = M`) or slices (if `R &lt; M`) along the `N`th</span>
+<a name="line-10205"></a><span class='hs-comment'>-- dimension of `params`.</span>
+<a name="line-10206"></a><span class='hs-comment'>-- </span>
+<a name="line-10207"></a><span class='hs-comment'>-- Produces an output tensor with shape</span>
+<a name="line-10208"></a><span class='hs-comment'>-- </span>
+<a name="line-10209"></a><span class='hs-comment'>--     [d_0, ..., d_{n-1}, params.shape[R], ..., params.shape[M-1]].</span>
+<a name="line-10210"></a><span class='hs-comment'>-- </span>
+<a name="line-10211"></a><span class='hs-comment'>-- Some examples below.</span>
+<a name="line-10212"></a><span class='hs-comment'>-- </span>
+<a name="line-10213"></a><span class='hs-comment'>-- Simple indexing into a matrix:</span>
+<a name="line-10214"></a><span class='hs-comment'>-- </span>
+<a name="line-10215"></a><span class='hs-comment'>--     indices = [[0, 0], [1, 1]]</span>
+<a name="line-10216"></a><span class='hs-comment'>--     params = [['a', 'b'], ['c', 'd']]</span>
+<a name="line-10217"></a><span class='hs-comment'>--     output = ['a', 'd']</span>
+<a name="line-10218"></a><span class='hs-comment'>-- </span>
+<a name="line-10219"></a><span class='hs-comment'>-- Slice indexing into a matrix:</span>
+<a name="line-10220"></a><span class='hs-comment'>-- </span>
+<a name="line-10221"></a><span class='hs-comment'>--     indices = [[1], [0]]</span>
+<a name="line-10222"></a><span class='hs-comment'>--     params = [['a', 'b'], ['c', 'd']]</span>
+<a name="line-10223"></a><span class='hs-comment'>--     output = [['c', 'd'], ['a', 'b']]</span>
+<a name="line-10224"></a><span class='hs-comment'>-- </span>
+<a name="line-10225"></a><span class='hs-comment'>-- Indexing into a 3-tensor:</span>
+<a name="line-10226"></a><span class='hs-comment'>-- </span>
+<a name="line-10227"></a><span class='hs-comment'>--     indices = [[1]]</span>
+<a name="line-10228"></a><span class='hs-comment'>--     params = [[['a0', 'b0'], ['c0', 'd0']],</span>
+<a name="line-10229"></a><span class='hs-comment'>--               [['a1', 'b1'], ['c1', 'd1']]]</span>
+<a name="line-10230"></a><span class='hs-comment'>--     output = [[['a1', 'b1'], ['c1', 'd1']]]</span>
+<a name="line-10231"></a><span class='hs-comment'>-- </span>
+<a name="line-10232"></a><span class='hs-comment'>-- </span>
+<a name="line-10233"></a><span class='hs-comment'>--     indices = [[0, 1], [1, 0]]</span>
+<a name="line-10234"></a><span class='hs-comment'>--     params = [[['a0', 'b0'], ['c0', 'd0']],</span>
+<a name="line-10235"></a><span class='hs-comment'>--               [['a1', 'b1'], ['c1', 'd1']]]</span>
+<a name="line-10236"></a><span class='hs-comment'>--     output = [['c0', 'd0'], ['a1', 'b1']]</span>
+<a name="line-10237"></a><span class='hs-comment'>-- </span>
+<a name="line-10238"></a><span class='hs-comment'>-- </span>
+<a name="line-10239"></a><span class='hs-comment'>--     indices = [[0, 0, 1], [1, 0, 1]]</span>
+<a name="line-10240"></a><span class='hs-comment'>--     params = [[['a0', 'b0'], ['c0', 'd0']],</span>
+<a name="line-10241"></a><span class='hs-comment'>--               [['a1', 'b1'], ['c1', 'd1']]]</span>
+<a name="line-10242"></a><span class='hs-comment'>--     output = ['b0', 'b1']</span>
+<a name="line-10243"></a><span class='hs-comment'>-- </span>
+<a name="line-10244"></a><span class='hs-comment'>-- Batched indexing into a matrix:</span>
+<a name="line-10245"></a><span class='hs-comment'>-- </span>
+<a name="line-10246"></a><span class='hs-comment'>--     indices = [[[0, 0]], [[0, 1]]]</span>
+<a name="line-10247"></a><span class='hs-comment'>--     params = [['a', 'b'], ['c', 'd']]</span>
+<a name="line-10248"></a><span class='hs-comment'>--     output = [['a'], ['b']]</span>
+<a name="line-10249"></a><span class='hs-comment'>-- </span>
+<a name="line-10250"></a><span class='hs-comment'>-- Batched slice indexing into a matrix:</span>
+<a name="line-10251"></a><span class='hs-comment'>-- </span>
+<a name="line-10252"></a><span class='hs-comment'>--     indices = [[[1]], [[0]]]</span>
+<a name="line-10253"></a><span class='hs-comment'>--     params = [['a', 'b'], ['c', 'd']]</span>
+<a name="line-10254"></a><span class='hs-comment'>--     output = [[['c', 'd']], [['a', 'b']]]</span>
+<a name="line-10255"></a><span class='hs-comment'>-- </span>
+<a name="line-10256"></a><span class='hs-comment'>-- Batched indexing into a 3-tensor:</span>
+<a name="line-10257"></a><span class='hs-comment'>-- </span>
+<a name="line-10258"></a><span class='hs-comment'>--     indices = [[[1]], [[0]]]</span>
+<a name="line-10259"></a><span class='hs-comment'>--     params = [[['a0', 'b0'], ['c0', 'd0']],</span>
+<a name="line-10260"></a><span class='hs-comment'>--               [['a1', 'b1'], ['c1', 'd1']]]</span>
+<a name="line-10261"></a><span class='hs-comment'>--     output = [[[['a1', 'b1'], ['c1', 'd1']]],</span>
+<a name="line-10262"></a><span class='hs-comment'>--               [[['a0', 'b0'], ['c0', 'd0']]]]</span>
+<a name="line-10263"></a><span class='hs-comment'>-- </span>
+<a name="line-10264"></a><span class='hs-comment'>-- </span>
+<a name="line-10265"></a><span class='hs-comment'>--     indices = [[[0, 1], [1, 0]], [[0, 0], [1, 1]]]</span>
+<a name="line-10266"></a><span class='hs-comment'>--     params = [[['a0', 'b0'], ['c0', 'd0']],</span>
+<a name="line-10267"></a><span class='hs-comment'>--               [['a1', 'b1'], ['c1', 'd1']]]</span>
+<a name="line-10268"></a><span class='hs-comment'>--     output = [[['c0', 'd0'], ['a1', 'b1']],</span>
+<a name="line-10269"></a><span class='hs-comment'>--               [['a0', 'b0'], ['c1', 'd1']]]</span>
+<a name="line-10270"></a><span class='hs-comment'>-- </span>
+<a name="line-10271"></a><span class='hs-comment'>-- </span>
+<a name="line-10272"></a><span class='hs-comment'>--     indices = [[[0, 0, 1], [1, 0, 1]], [[0, 1, 1], [1, 1, 0]]]</span>
+<a name="line-10273"></a><span class='hs-comment'>--     params = [[['a0', 'b0'], ['c0', 'd0']],</span>
+<a name="line-10274"></a><span class='hs-comment'>--               [['a1', 'b1'], ['c1', 'd1']]]</span>
+<a name="line-10275"></a><span class='hs-comment'>--     output = [['b0', 'b1'], ['d0', 'c1']]</span>
+<a name="line-10276"></a><span class='hs-definition'>gatherNd</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tindices</span> <span class='hs-varid'>tparams</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-10277"></a>                                             <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-10278"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-10279"></a>                                             <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tparams</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-10280"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>tparams</span> <span class='hs-comment'>-- ^ __params__: `M-D`.  The tensor from which to gather values.</span>
+<a name="line-10281"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __indices__: `(N+1)-D`.  Index tensor having shape `[d_0, ..., d_N, R]`.</span>
+<a name="line-10282"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>tparams</span> <span class='hs-comment'>-- ^ __output__: `(N+M-R)-D`.  Values from `params` gathered from indices given by</span>
+<a name="line-10283"></a>            <span class='hs-comment'>-- `indices`.</span>
+<a name="line-10284"></a><span class='hs-definition'>gatherNd</span> <span class='hs-varid'>params</span> <span class='hs-varid'>indices</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10285"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"GatherNd"</span>
+<a name="line-10286"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span>
+<a name="line-10287"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tparams"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tparams</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-10288"></a>        <span class='hs-varid'>params</span> <span class='hs-varid'>indices</span>
+<a name="line-10289"></a><span class='hs-comment'>{-
+<a name="line-10290"></a>attr { name: "Tparams" type: "type" }
+<a name="line-10291"></a>attr {
+<a name="line-10292"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-10293"></a>  name: "Tindices"
+<a name="line-10294"></a>  type: "type"
+<a name="line-10295"></a>}
+<a name="line-10296"></a>input_arg {
+<a name="line-10297"></a>  description: "`M-D`.  The tensor from which to gather values."
+<a name="line-10298"></a>  name: "params"
+<a name="line-10299"></a>  type_attr: "Tparams"
+<a name="line-10300"></a>}
+<a name="line-10301"></a>input_arg {
+<a name="line-10302"></a>  description: "`(N+1)-D`.  Index tensor having shape `[d_0, ..., d_N, R]`."
+<a name="line-10303"></a>  name: "indices"
+<a name="line-10304"></a>  type_attr: "Tindices"
+<a name="line-10305"></a>}
+<a name="line-10306"></a>output_arg {
+<a name="line-10307"></a>  description: "`(N+M-R)-D`.  Values from `params` gathered from indices given by\n`indices`."
+<a name="line-10308"></a>  name: "output"
+<a name="line-10309"></a>  type_attr: "Tparams"
+<a name="line-10310"></a>}
+<a name="line-10311"></a>-}</span>
+<a name="line-10312"></a>
+<a name="line-10313"></a><span class='hs-comment'>-- | Read an element from the TensorArray into output `value`.</span>
+<a name="line-10314"></a>
+<a name="line-10315"></a><a name="tensorArrayRead"></a><span class='hs-definition'>tensorArrayRead</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-10316"></a>                   <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a TensorArray.</span>
+<a name="line-10317"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __index__</span>
+<a name="line-10318"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __flow_in__: A float scalar that enforces proper chaining of operations.</span>
+<a name="line-10319"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __value__: The tensor that is read from the TensorArray.</span>
+<a name="line-10320"></a><span class='hs-definition'>tensorArrayRead</span> <span class='hs-varid'>handle</span> <span class='hs-varid'>index</span> <span class='hs-varid'>flow_in</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10321"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TensorArrayRead"</span>
+<a name="line-10322"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-10323"></a>        <span class='hs-varid'>handle</span> <span class='hs-varid'>index</span> <span class='hs-varid'>flow_in</span>
+<a name="line-10324"></a><span class='hs-comment'>{-
+<a name="line-10325"></a>attr {
+<a name="line-10326"></a>  description: "The type of the elem that is returned."
+<a name="line-10327"></a>  name: "dtype"
+<a name="line-10328"></a>  type: "type"
+<a name="line-10329"></a>}
+<a name="line-10330"></a>input_arg {
+<a name="line-10331"></a>  description: "The handle to a TensorArray."
+<a name="line-10332"></a>  is_ref: true
+<a name="line-10333"></a>  name: "handle"
+<a name="line-10334"></a>  type: DT_STRING
+<a name="line-10335"></a>}
+<a name="line-10336"></a>input_arg { name: "index" type: DT_INT32 }
+<a name="line-10337"></a>input_arg {
+<a name="line-10338"></a>  description: "A float scalar that enforces proper chaining of operations."
+<a name="line-10339"></a>  name: "flow_in"
+<a name="line-10340"></a>  type: DT_FLOAT
+<a name="line-10341"></a>}
+<a name="line-10342"></a>output_arg {
+<a name="line-10343"></a>  description: "The tensor that is read from the TensorArray."
+<a name="line-10344"></a>  name: "value"
+<a name="line-10345"></a>  type_attr: "dtype"
+<a name="line-10346"></a>}
+<a name="line-10347"></a>-}</span>
+<a name="line-10348"></a>
+<a name="line-10349"></a><a name="readerReadUpTo"></a><span class='hs-comment'>-- | Returns up to `num_records` (key, value) pairs produced by a Reader.</span>
+<a name="line-10350"></a><span class='hs-comment'>--</span>
+<a name="line-10351"></a><span class='hs-comment'>-- Will dequeue from the input queue if necessary (e.g. when the</span>
+<a name="line-10352"></a><span class='hs-comment'>-- Reader needs to start reading from a new file since it has finished</span>
+<a name="line-10353"></a><span class='hs-comment'>-- with the previous file).</span>
+<a name="line-10354"></a><span class='hs-comment'>-- It may return less than `num_records` even before the last batch.</span>
+<a name="line-10355"></a><span class='hs-definition'>readerReadUpTo</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __reader_handle__: Handle to a `Reader`.</span>
+<a name="line-10356"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __queue_handle__: Handle to a `Queue`, with string work items.</span>
+<a name="line-10357"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_records__: number of records to read from `Reader`.</span>
+<a name="line-10358"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>,</span>
+<a name="line-10359"></a>                      <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>
+<a name="line-10360"></a>                  <span class='hs-comment'>-- ^ (__keys__, __values__)</span>
+<a name="line-10361"></a>                  <span class='hs-comment'>--</span>
+<a name="line-10362"></a>                  <span class='hs-comment'>-- * __keys__: A 1-D tensor.</span>
+<a name="line-10363"></a>                  <span class='hs-comment'>--</span>
+<a name="line-10364"></a>                  <span class='hs-comment'>-- * __values__: A 1-D tensor.</span>
+<a name="line-10365"></a><span class='hs-definition'>readerReadUpTo</span> <span class='hs-varid'>reader_handle</span> <span class='hs-varid'>queue_handle</span> <span class='hs-varid'>num_records</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10366"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ReaderReadUpTo"</span><span class='hs-layout'>)</span>
+<a name="line-10367"></a>        <span class='hs-varid'>reader_handle</span> <span class='hs-varid'>queue_handle</span> <span class='hs-varid'>num_records</span>
+<a name="line-10368"></a><span class='hs-comment'>{-
+<a name="line-10369"></a>input_arg {
+<a name="line-10370"></a>  description: "Handle to a `Reader`."
+<a name="line-10371"></a>  is_ref: true
+<a name="line-10372"></a>  name: "reader_handle"
+<a name="line-10373"></a>  type: DT_STRING
+<a name="line-10374"></a>}
+<a name="line-10375"></a>input_arg {
+<a name="line-10376"></a>  description: "Handle to a `Queue`, with string work items."
+<a name="line-10377"></a>  is_ref: true
+<a name="line-10378"></a>  name: "queue_handle"
+<a name="line-10379"></a>  type: DT_STRING
+<a name="line-10380"></a>}
+<a name="line-10381"></a>input_arg {
+<a name="line-10382"></a>  description: "number of records to read from `Reader`."
+<a name="line-10383"></a>  name: "num_records"
+<a name="line-10384"></a>  type: DT_INT64
+<a name="line-10385"></a>}
+<a name="line-10386"></a>output_arg {
+<a name="line-10387"></a>  description: "A 1-D tensor." name: "keys" type: DT_STRING
+<a name="line-10388"></a>}
+<a name="line-10389"></a>output_arg {
+<a name="line-10390"></a>  description: "A 1-D tensor." name: "values" type: DT_STRING
+<a name="line-10391"></a>}
+<a name="line-10392"></a>-}</span>
+<a name="line-10393"></a>
+<a name="line-10394"></a><a name="betainc"></a><span class='hs-comment'>-- | Compute the regularized incomplete beta integral \\(I_x(a, b)\\).</span>
+<a name="line-10395"></a><span class='hs-comment'>--</span>
+<a name="line-10396"></a><span class='hs-comment'>-- The regularized incomplete beta integral is defined as:</span>
+<a name="line-10397"></a><span class='hs-comment'>-- </span>
+<a name="line-10398"></a><span class='hs-comment'>-- ```</span>
+<a name="line-10399"></a><span class='hs-comment'>-- I_x(a, b) = \frac{B(x; a, b)}{B(a, b)}</span>
+<a name="line-10400"></a><span class='hs-comment'>-- ```</span>
+<a name="line-10401"></a><span class='hs-comment'>-- where</span>
+<a name="line-10402"></a><span class='hs-comment'>-- </span>
+<a name="line-10403"></a><span class='hs-comment'>-- ```</span>
+<a name="line-10404"></a><span class='hs-comment'>-- B(x; a, b) = \int_0^x t^{a-1} (1 - t)^{b-1} dt</span>
+<a name="line-10405"></a><span class='hs-comment'>-- ```</span>
+<a name="line-10406"></a><span class='hs-comment'>-- </span>
+<a name="line-10407"></a><span class='hs-comment'>-- is the incomplete beta function and \\(B(a, b)\\) is the *complete*</span>
+<a name="line-10408"></a><span class='hs-comment'>-- beta function.</span>
+<a name="line-10409"></a><span class='hs-definition'>betainc</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-10410"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __a__</span>
+<a name="line-10411"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __b__</span>
+<a name="line-10412"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-10413"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-10414"></a><span class='hs-definition'>betainc</span> <span class='hs-varid'>a</span> <span class='hs-varid'>b</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10415"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Betainc"</span>
+<a name="line-10416"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-10417"></a>        <span class='hs-varid'>a</span> <span class='hs-varid'>b</span> <span class='hs-varid'>x</span>
+<a name="line-10418"></a><span class='hs-comment'>{-
+<a name="line-10419"></a>attr {
+<a name="line-10420"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-10421"></a>  name: "T"
+<a name="line-10422"></a>  type: "type"
+<a name="line-10423"></a>}
+<a name="line-10424"></a>input_arg { name: "a" type_attr: "T" }
+<a name="line-10425"></a>input_arg { name: "b" type_attr: "T" }
+<a name="line-10426"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-10427"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-10428"></a>-}</span>
+<a name="line-10429"></a>
+<a name="line-10430"></a><span class='hs-comment'>-- | </span>
+<a name="line-10431"></a>
+<a name="line-10432"></a><a name="batchMatrixBandPart"></a><span class='hs-definition'>batchMatrixBandPart</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-10433"></a>                       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-10434"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_lower__</span>
+<a name="line-10435"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_upper__</span>
+<a name="line-10436"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __band__</span>
+<a name="line-10437"></a><span class='hs-definition'>batchMatrixBandPart</span> <span class='hs-varid'>input</span> <span class='hs-varid'>num_lower</span> <span class='hs-varid'>num_upper</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10438"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchMatrixBandPart"</span>
+<a name="line-10439"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-10440"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>num_lower</span> <span class='hs-varid'>num_upper</span>
+<a name="line-10441"></a><span class='hs-comment'>{-
+<a name="line-10442"></a>attr { name: "T" type: "type" }
+<a name="line-10443"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-10444"></a>input_arg { name: "num_lower" type: DT_INT64 }
+<a name="line-10445"></a>input_arg { name: "num_upper" type: DT_INT64 }
+<a name="line-10446"></a>output_arg { name: "band" type_attr: "T" }
+<a name="line-10447"></a>-}</span>
+<a name="line-10448"></a>
+<a name="line-10449"></a><span class='hs-comment'>-- | Computes the gradients of depthwise convolution with respect to the input.</span>
+<a name="line-10450"></a>
+<a name="line-10451"></a><a name="depthwiseConv2dNativeBackpropInput"></a><span class='hs-definition'>depthwiseConv2dNativeBackpropInput</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-10452"></a>                                                           <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-10453"></a>                                                                   <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-10454"></a>                                      <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __input_sizes__: An integer vector representing the shape of `input`,</span>
+<a name="line-10455"></a>                                                               <span class='hs-comment'>-- where `input` is a 4-D `[batch, height, width, channels]` tensor.</span>
+<a name="line-10456"></a>                                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __filter__: 4-D with shape</span>
+<a name="line-10457"></a>                                                     <span class='hs-comment'>-- `[filter_height, filter_width, in_channels, depthwise_multiplier]`.</span>
+<a name="line-10458"></a>                                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out_backprop__: 4-D with shape `[batch, out_height, out_width, out_channels]`.</span>
+<a name="line-10459"></a>                                                     <span class='hs-comment'>-- Gradients w.r.t. the output of the convolution.</span>
+<a name="line-10460"></a>                                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 4-D with shape `[batch, in_height, in_width, in_channels]`.  Gradient</span>
+<a name="line-10461"></a>                                      <span class='hs-comment'>-- w.r.t. the input of the convolution.</span>
+<a name="line-10462"></a><span class='hs-definition'>depthwiseConv2dNativeBackpropInput</span> <span class='hs-varid'>input_sizes</span> <span class='hs-varid'>filter</span>
+<a name="line-10463"></a>                                   <span class='hs-varid'>out_backprop</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10464"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DepthwiseConv2dNativeBackpropInput"</span>
+<a name="line-10465"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-10466"></a>        <span class='hs-varid'>input_sizes</span> <span class='hs-varid'>filter</span> <span class='hs-varid'>out_backprop</span>
+<a name="line-10467"></a><span class='hs-comment'>{-
+<a name="line-10468"></a>attr {
+<a name="line-10469"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-10470"></a>  name: "T"
+<a name="line-10471"></a>  type: "type"
+<a name="line-10472"></a>}
+<a name="line-10473"></a>attr {
+<a name="line-10474"></a>  description: "The stride of the sliding window for each dimension of the input\nof the convolution."
+<a name="line-10475"></a>  name: "strides"
+<a name="line-10476"></a>  type: "list(int)"
+<a name="line-10477"></a>}
+<a name="line-10478"></a>attr {
+<a name="line-10479"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-10480"></a>  description: "The type of padding algorithm to use."
+<a name="line-10481"></a>  name: "padding"
+<a name="line-10482"></a>  type: "string"
+<a name="line-10483"></a>}
+<a name="line-10484"></a>input_arg {
+<a name="line-10485"></a>  description: "An integer vector representing the shape of `input`,\nwhere `input` is a 4-D `[batch, height, width, channels]` tensor."
+<a name="line-10486"></a>  name: "input_sizes"
+<a name="line-10487"></a>  type: DT_INT32
+<a name="line-10488"></a>}
+<a name="line-10489"></a>input_arg {
+<a name="line-10490"></a>  description: "4-D with shape\n`[filter_height, filter_width, in_channels, depthwise_multiplier]`."
+<a name="line-10491"></a>  name: "filter"
+<a name="line-10492"></a>  type_attr: "T"
+<a name="line-10493"></a>}
+<a name="line-10494"></a>input_arg {
+<a name="line-10495"></a>  description: "4-D with shape `[batch, out_height, out_width, out_channels]`.\nGradients w.r.t. the output of the convolution."
+<a name="line-10496"></a>  name: "out_backprop"
+<a name="line-10497"></a>  type_attr: "T"
+<a name="line-10498"></a>}
+<a name="line-10499"></a>output_arg {
+<a name="line-10500"></a>  description: "4-D with shape `[batch, in_height, in_width, in_channels]`.  Gradient\nw.r.t. the input of the convolution."
+<a name="line-10501"></a>  name: "output"
+<a name="line-10502"></a>  type_attr: "T"
+<a name="line-10503"></a>}
+<a name="line-10504"></a>-}</span>
+<a name="line-10505"></a>
+<a name="line-10506"></a><span class='hs-comment'>-- | Forwards the `index`th element of `inputs` to `output`.</span>
+<a name="line-10507"></a>
+<a name="line-10508"></a><a name="refSelect"></a><span class='hs-definition'>refSelect</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-10509"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __index__: A scalar that determines the input that gets selected.</span>
+<a name="line-10510"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __inputs__: A list of ref tensors, one of which will be forwarded to `output`.</span>
+<a name="line-10511"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The forwarded tensor.</span>
+<a name="line-10512"></a><span class='hs-definition'>refSelect</span> <span class='hs-varid'>index</span> <span class='hs-varid'>inputs</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"N"</span><span class='hs-layout'>,</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"inputs"</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>inputs</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10513"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RefSelect"</span>
+<a name="line-10514"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-10515"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"N"</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>inputs</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-10516"></a>        <span class='hs-varid'>index</span> <span class='hs-varid'>inputs</span>
+<a name="line-10517"></a><span class='hs-comment'>{-
+<a name="line-10518"></a>attr { name: "T" type: "type" }
+<a name="line-10519"></a>attr { has_minimum: true minimum: 1 name: "N" type: "int" }
+<a name="line-10520"></a>input_arg {
+<a name="line-10521"></a>  description: "A scalar that determines the input that gets selected."
+<a name="line-10522"></a>  name: "index"
+<a name="line-10523"></a>  type: DT_INT32
+<a name="line-10524"></a>}
+<a name="line-10525"></a>input_arg {
+<a name="line-10526"></a>  description: "A list of ref tensors, one of which will be forwarded to `output`."
+<a name="line-10527"></a>  is_ref: true
+<a name="line-10528"></a>  name: "inputs"
+<a name="line-10529"></a>  number_attr: "N"
+<a name="line-10530"></a>  type_attr: "T"
+<a name="line-10531"></a>}
+<a name="line-10532"></a>output_arg {
+<a name="line-10533"></a>  description: "The forwarded tensor."
+<a name="line-10534"></a>  is_ref: true
+<a name="line-10535"></a>  name: "output"
+<a name="line-10536"></a>  type_attr: "T"
+<a name="line-10537"></a>}
+<a name="line-10538"></a>-}</span>
+<a name="line-10539"></a>
+<a name="line-10540"></a><a name="exit"></a><span class='hs-comment'>-- | Exits the current frame to its parent frame.</span>
+<a name="line-10541"></a><span class='hs-comment'>--</span>
+<a name="line-10542"></a><span class='hs-comment'>-- Exit makes its input `data` available to the parent frame.</span>
+<a name="line-10543"></a><span class='hs-definition'>exit</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-10544"></a>        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__: The tensor to be made available to the parent frame.</span>
+<a name="line-10545"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The same tensor as `data`.</span>
+<a name="line-10546"></a><span class='hs-definition'>exit</span> <span class='hs-varid'>data'</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10547"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Exit"</span>
+<a name="line-10548"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-10549"></a>        <span class='hs-varid'>data'</span>
+<a name="line-10550"></a><span class='hs-comment'>{-
+<a name="line-10551"></a>attr { name: "T" type: "type" }
+<a name="line-10552"></a>input_arg {
+<a name="line-10553"></a>  description: "The tensor to be made available to the parent frame."
+<a name="line-10554"></a>  name: "data"
+<a name="line-10555"></a>  type_attr: "T"
+<a name="line-10556"></a>}
+<a name="line-10557"></a>output_arg {
+<a name="line-10558"></a>  description: "The same tensor as `data`."
+<a name="line-10559"></a>  name: "output"
+<a name="line-10560"></a>  type_attr: "T"
+<a name="line-10561"></a>}
+<a name="line-10562"></a>-}</span>
+<a name="line-10563"></a>
+<a name="line-10564"></a><a name="lookupTableFind"></a><span class='hs-comment'>-- | Looks up keys in a table, outputs the corresponding values.</span>
+<a name="line-10565"></a><span class='hs-comment'>--</span>
+<a name="line-10566"></a><span class='hs-comment'>-- The tensor `keys` must of the same type as the keys of the table.</span>
+<a name="line-10567"></a><span class='hs-comment'>-- The output `values` is of the type of the table values.</span>
+<a name="line-10568"></a><span class='hs-comment'>-- </span>
+<a name="line-10569"></a><span class='hs-comment'>-- The scalar `default_value` is the value output for keys not present in the</span>
+<a name="line-10570"></a><span class='hs-comment'>-- table. It must also be of the same type as the table values.</span>
+<a name="line-10571"></a><span class='hs-definition'>lookupTableFind</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>tin</span> <span class='hs-varid'>tout</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>tin</span><span class='hs-layout'>,</span>
+<a name="line-10572"></a>                                               <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tout</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-10573"></a>                   <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __table_handle__: Handle to the table.</span>
+<a name="line-10574"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tin</span> <span class='hs-comment'>-- ^ __keys__: Any shape.  Keys to look up.</span>
+<a name="line-10575"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>tout</span> <span class='hs-comment'>-- ^ __default_value__</span>
+<a name="line-10576"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>tout</span> <span class='hs-comment'>-- ^ __values__: Same shape as `keys`.  Values found in the table, or `default_values`</span>
+<a name="line-10577"></a>                   <span class='hs-comment'>-- for missing keys.</span>
+<a name="line-10578"></a><span class='hs-definition'>lookupTableFind</span> <span class='hs-varid'>table_handle</span> <span class='hs-varid'>keys</span> <span class='hs-varid'>default_value</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10579"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"LookupTableFind"</span>
+<a name="line-10580"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tin"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tin</span><span class='hs-layout'>)</span>
+<a name="line-10581"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tout"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tout</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-10582"></a>        <span class='hs-varid'>table_handle</span> <span class='hs-varid'>keys</span> <span class='hs-varid'>default_value</span>
+<a name="line-10583"></a><span class='hs-comment'>{-
+<a name="line-10584"></a>attr { name: "Tin" type: "type" }
+<a name="line-10585"></a>attr { name: "Tout" type: "type" }
+<a name="line-10586"></a>input_arg {
+<a name="line-10587"></a>  description: "Handle to the table."
+<a name="line-10588"></a>  is_ref: true
+<a name="line-10589"></a>  name: "table_handle"
+<a name="line-10590"></a>  type: DT_STRING
+<a name="line-10591"></a>}
+<a name="line-10592"></a>input_arg {
+<a name="line-10593"></a>  description: "Any shape.  Keys to look up."
+<a name="line-10594"></a>  name: "keys"
+<a name="line-10595"></a>  type_attr: "Tin"
+<a name="line-10596"></a>}
+<a name="line-10597"></a>input_arg { name: "default_value" type_attr: "Tout" }
+<a name="line-10598"></a>output_arg {
+<a name="line-10599"></a>  description: "Same shape as `keys`.  Values found in the table, or `default_values`\nfor missing keys."
+<a name="line-10600"></a>  name: "values"
+<a name="line-10601"></a>  type_attr: "Tout"
+<a name="line-10602"></a>}
+<a name="line-10603"></a>-}</span>
+<a name="line-10604"></a>
+<a name="line-10605"></a><a name="squeeze"></a><span class='hs-comment'>-- | Removes dimensions of size 1 from the shape of a tensor.</span>
+<a name="line-10606"></a><span class='hs-comment'>--</span>
+<a name="line-10607"></a><span class='hs-comment'>-- Given a tensor `input`, this operation returns a tensor of the same type with</span>
+<a name="line-10608"></a><span class='hs-comment'>-- all dimensions of size 1 removed. If you don't want to remove all size 1</span>
+<a name="line-10609"></a><span class='hs-comment'>-- dimensions, you can remove specific size 1 dimensions by specifying</span>
+<a name="line-10610"></a><span class='hs-comment'>-- `squeeze_dims`.</span>
+<a name="line-10611"></a><span class='hs-comment'>-- </span>
+<a name="line-10612"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-10613"></a><span class='hs-comment'>-- </span>
+<a name="line-10614"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10615"></a><span class='hs-comment'>-- # 't' is a tensor of shape [1, 2, 1, 3, 1, 1]</span>
+<a name="line-10616"></a><span class='hs-comment'>-- shape(squeeze(t)) ==&gt; [2, 3]</span>
+<a name="line-10617"></a><span class='hs-comment'>-- ```</span>
+<a name="line-10618"></a><span class='hs-comment'>-- </span>
+<a name="line-10619"></a><span class='hs-comment'>-- Or, to remove specific size 1 dimensions:</span>
+<a name="line-10620"></a><span class='hs-comment'>-- </span>
+<a name="line-10621"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10622"></a><span class='hs-comment'>-- # 't' is a tensor of shape [1, 2, 1, 3, 1, 1]</span>
+<a name="line-10623"></a><span class='hs-comment'>-- shape(squeeze(t, [2, 4])) ==&gt; [1, 2, 3, 1]</span>
+<a name="line-10624"></a><span class='hs-comment'>-- ```</span>
+<a name="line-10625"></a><span class='hs-definition'>squeeze</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-10626"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: The `input` to squeeze.</span>
+<a name="line-10627"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Contains the same data as `input`, but has one or more dimensions of</span>
+<a name="line-10628"></a>           <span class='hs-comment'>-- size 1 removed.</span>
+<a name="line-10629"></a><span class='hs-definition'>squeeze</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10630"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Squeeze"</span>
+<a name="line-10631"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-10632"></a>        <span class='hs-varid'>input</span>
+<a name="line-10633"></a><span class='hs-comment'>{-
+<a name="line-10634"></a>attr { name: "T" type: "type" }
+<a name="line-10635"></a>attr {
+<a name="line-10636"></a>  default_value { list { } }
+<a name="line-10637"></a>  description: "If specified, only squeezes the dimensions listed. The dimension\nindex starts at 0. It is an error to squeeze a dimension that is not 1."
+<a name="line-10638"></a>  has_minimum: true
+<a name="line-10639"></a>  name: "squeeze_dims"
+<a name="line-10640"></a>  type: "list(int)"
+<a name="line-10641"></a>}
+<a name="line-10642"></a>input_arg {
+<a name="line-10643"></a>  description: "The `input` to squeeze." name: "input" type_attr: "T"
+<a name="line-10644"></a>}
+<a name="line-10645"></a>output_arg {
+<a name="line-10646"></a>  description: "Contains the same data as `input`, but has one or more dimensions of\nsize 1 removed."
+<a name="line-10647"></a>  name: "output"
+<a name="line-10648"></a>  type_attr: "T"
+<a name="line-10649"></a>}
+<a name="line-10650"></a>-}</span>
+<a name="line-10651"></a>
+<a name="line-10652"></a><a name="mean"></a><span class='hs-comment'>-- | Computes the mean of elements across dimensions of a tensor.</span>
+<a name="line-10653"></a><span class='hs-comment'>--</span>
+<a name="line-10654"></a><span class='hs-comment'>-- Reduces `input` along the dimensions given in `reduction_indices`. Unless</span>
+<a name="line-10655"></a><span class='hs-comment'>-- `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in</span>
+<a name="line-10656"></a><span class='hs-comment'>-- `reduction_indices`. If `keep_dims` is true, the reduced dimensions are</span>
+<a name="line-10657"></a><span class='hs-comment'>-- retained with length 1.</span>
+<a name="line-10658"></a><span class='hs-definition'>mean</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-10659"></a>                               <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-10660"></a>                                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-10661"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-10662"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-10663"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-10664"></a>                                       <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span>
+<a name="line-10665"></a>                               <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-10666"></a>        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: The tensor to reduce.</span>
+<a name="line-10667"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __reduction_indices__: The dimensions to reduce.</span>
+<a name="line-10668"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The reduced tensor.</span>
+<a name="line-10669"></a><span class='hs-definition'>mean</span> <span class='hs-varid'>input</span> <span class='hs-varid'>reduction_indices</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10670"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Mean"</span>
+<a name="line-10671"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-10672"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-10673"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>reduction_indices</span>
+<a name="line-10674"></a><span class='hs-comment'>{-
+<a name="line-10675"></a>attr {
+<a name="line-10676"></a>  default_value { b: false }
+<a name="line-10677"></a>  description: "If true, retain reduced dimensions with length 1."
+<a name="line-10678"></a>  name: "keep_dims"
+<a name="line-10679"></a>  type: "bool"
+<a name="line-10680"></a>}
+<a name="line-10681"></a>attr {
+<a name="line-10682"></a>  allowed_values {
+<a name="line-10683"></a>    list {
+<a name="line-10684"></a>      type: DT_FLOAT
+<a name="line-10685"></a>      type: DT_DOUBLE
+<a name="line-10686"></a>      type: DT_INT64
+<a name="line-10687"></a>      type: DT_INT32
+<a name="line-10688"></a>      type: DT_UINT8
+<a name="line-10689"></a>      type: DT_UINT16
+<a name="line-10690"></a>      type: DT_INT16
+<a name="line-10691"></a>      type: DT_INT8
+<a name="line-10692"></a>      type: DT_COMPLEX64
+<a name="line-10693"></a>      type: DT_COMPLEX128
+<a name="line-10694"></a>      type: DT_QINT8
+<a name="line-10695"></a>      type: DT_QUINT8
+<a name="line-10696"></a>      type: DT_QINT32
+<a name="line-10697"></a>      type: DT_HALF
+<a name="line-10698"></a>    }
+<a name="line-10699"></a>  }
+<a name="line-10700"></a>  name: "T"
+<a name="line-10701"></a>  type: "type"
+<a name="line-10702"></a>}
+<a name="line-10703"></a>attr {
+<a name="line-10704"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-10705"></a>  default_value { type: DT_INT32 }
+<a name="line-10706"></a>  name: "Tidx"
+<a name="line-10707"></a>  type: "type"
+<a name="line-10708"></a>}
+<a name="line-10709"></a>input_arg {
+<a name="line-10710"></a>  description: "The tensor to reduce." name: "input" type_attr: "T"
+<a name="line-10711"></a>}
+<a name="line-10712"></a>input_arg {
+<a name="line-10713"></a>  description: "The dimensions to reduce."
+<a name="line-10714"></a>  name: "reduction_indices"
+<a name="line-10715"></a>  type_attr: "Tidx"
+<a name="line-10716"></a>}
+<a name="line-10717"></a>output_arg {
+<a name="line-10718"></a>  description: "The reduced tensor." name: "output" type_attr: "T"
+<a name="line-10719"></a>}
+<a name="line-10720"></a>-}</span>
+<a name="line-10721"></a>
+<a name="line-10722"></a><a name="spaceToBatchND"></a><span class='hs-comment'>-- | SpaceToBatch for N-D tensors of type T.</span>
+<a name="line-10723"></a><span class='hs-comment'>--</span>
+<a name="line-10724"></a><span class='hs-comment'>-- This operation divides "spatial" dimensions `[1, ..., M]` of the input into a</span>
+<a name="line-10725"></a><span class='hs-comment'>-- grid of blocks of shape `block_shape`, and interleaves these blocks with the</span>
+<a name="line-10726"></a><span class='hs-comment'>-- "batch" dimension (0) such that in the output, the spatial dimensions</span>
+<a name="line-10727"></a><span class='hs-comment'>-- `[1, ..., M]` correspond to the position within the grid, and the batch</span>
+<a name="line-10728"></a><span class='hs-comment'>-- dimension combines both the position within a spatial block and the original</span>
+<a name="line-10729"></a><span class='hs-comment'>-- batch position.  Prior to division into blocks, the spatial dimensions of the</span>
+<a name="line-10730"></a><span class='hs-comment'>-- input are optionally zero padded according to `paddings`.  See below for a</span>
+<a name="line-10731"></a><span class='hs-comment'>-- precise description.</span>
+<a name="line-10732"></a><span class='hs-definition'>spaceToBatchND</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tblock_shape</span> <span class='hs-varid'>tpaddings</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-10733"></a>                                                              <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tblock_shape</span><span class='hs-layout'>,</span>
+<a name="line-10734"></a>                                                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-10735"></a>                                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tblock_shape</span><span class='hs-layout'>,</span>
+<a name="line-10736"></a>                                                              <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tpaddings</span><span class='hs-layout'>,</span>
+<a name="line-10737"></a>                                                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-10738"></a>                                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tpaddings</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-10739"></a>                  <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: N-D with shape `input_shape = [batch] + spatial_shape + remaining_shape`,</span>
+<a name="line-10740"></a>                              <span class='hs-comment'>-- where spatial_shape has `M` dimensions.</span>
+<a name="line-10741"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tblock_shape</span> <span class='hs-comment'>-- ^ __block_shape__: 1-D with shape `[M]`, all values must be &gt;= 1.</span>
+<a name="line-10742"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>tpaddings</span> <span class='hs-comment'>-- ^ __paddings__: 2-D with shape `[M, 2]`, all values must be &gt;= 0.</span>
+<a name="line-10743"></a>                                         <span class='hs-comment'>--   `paddings[i] = [pad_start, pad_end]` specifies the padding for input dimension</span>
+<a name="line-10744"></a>                                         <span class='hs-comment'>--   `i + 1`, which corresponds to spatial dimension `i`.  It is required that</span>
+<a name="line-10745"></a>                                         <span class='hs-comment'>--   `block_shape[i]` divides `input_shape[i + 1] + pad_start + pad_end`.</span>
+<a name="line-10746"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10747"></a>                                         <span class='hs-comment'>-- This operation is equivalent to the following steps:</span>
+<a name="line-10748"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10749"></a>                                         <span class='hs-comment'>-- 1. Zero-pad the start and end of dimensions `[1, ..., M]` of the</span>
+<a name="line-10750"></a>                                         <span class='hs-comment'>--    input according to `paddings` to produce `padded` of shape `padded_shape`.</span>
+<a name="line-10751"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10752"></a>                                         <span class='hs-comment'>-- 2. Reshape `padded` to `reshaped_padded` of shape:</span>
+<a name="line-10753"></a>                                         <span class='hs-comment'>--      [batch] +</span>
+<a name="line-10754"></a>                                         <span class='hs-comment'>--      [padded_shape[1] / block_shape[0],</span>
+<a name="line-10755"></a>                                         <span class='hs-comment'>--        block_shape[0],</span>
+<a name="line-10756"></a>                                         <span class='hs-comment'>--       ...,</span>
+<a name="line-10757"></a>                                         <span class='hs-comment'>--       padded_shape[M] / block_shape[M-1],</span>
+<a name="line-10758"></a>                                         <span class='hs-comment'>--       block_shape[M-1]] +</span>
+<a name="line-10759"></a>                                         <span class='hs-comment'>--      remaining_shape</span>
+<a name="line-10760"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10761"></a>                                         <span class='hs-comment'>-- 3. Permute dimensions of `reshaped_padded` to produce</span>
+<a name="line-10762"></a>                                         <span class='hs-comment'>--    `permuted_reshaped_padded` of shape:</span>
+<a name="line-10763"></a>                                         <span class='hs-comment'>--      block_shape +</span>
+<a name="line-10764"></a>                                         <span class='hs-comment'>--      [batch] +</span>
+<a name="line-10765"></a>                                         <span class='hs-comment'>--      [padded_shape[1] / block_shape[0],</span>
+<a name="line-10766"></a>                                         <span class='hs-comment'>--       ...,</span>
+<a name="line-10767"></a>                                         <span class='hs-comment'>--       padded_shape[M] / block_shape[M-1]] +</span>
+<a name="line-10768"></a>                                         <span class='hs-comment'>--      remaining_shape</span>
+<a name="line-10769"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10770"></a>                                         <span class='hs-comment'>-- 4. Reshape `permuted_reshaped_padded` to flatten `block_shape` into the batch</span>
+<a name="line-10771"></a>                                         <span class='hs-comment'>--    dimension, producing an output tensor of shape:</span>
+<a name="line-10772"></a>                                         <span class='hs-comment'>--      [batch * prod(block_shape)] +</span>
+<a name="line-10773"></a>                                         <span class='hs-comment'>--      [padded_shape[1] / block_shape[0],</span>
+<a name="line-10774"></a>                                         <span class='hs-comment'>--       ...,</span>
+<a name="line-10775"></a>                                         <span class='hs-comment'>--       padded_shape[M] / block_shape[M-1]] +</span>
+<a name="line-10776"></a>                                         <span class='hs-comment'>--      remaining_shape</span>
+<a name="line-10777"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10778"></a>                                         <span class='hs-comment'>-- Some examples:</span>
+<a name="line-10779"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10780"></a>                                         <span class='hs-comment'>-- (1) For the following input of shape `[1, 2, 2, 1]`, `block_shape = [2, 2]`, and</span>
+<a name="line-10781"></a>                                         <span class='hs-comment'>--     `paddings = [[0, 0], [0, 0]]`:</span>
+<a name="line-10782"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10783"></a>                                         <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10784"></a>                                         <span class='hs-comment'>-- x = [[[[1], [2]], [[3], [4]]]]</span>
+<a name="line-10785"></a>                                         <span class='hs-comment'>-- ```</span>
+<a name="line-10786"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10787"></a>                                         <span class='hs-comment'>-- The output tensor has shape `[4, 1, 1, 1]` and value:</span>
+<a name="line-10788"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10789"></a>                                         <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10790"></a>                                         <span class='hs-comment'>-- [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]</span>
+<a name="line-10791"></a>                                         <span class='hs-comment'>-- ```</span>
+<a name="line-10792"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10793"></a>                                         <span class='hs-comment'>-- (2) For the following input of shape `[1, 2, 2, 3]`, `block_shape = [2, 2]`, and</span>
+<a name="line-10794"></a>                                         <span class='hs-comment'>--     `paddings = [[0, 0], [0, 0]]`:</span>
+<a name="line-10795"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10796"></a>                                         <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10797"></a>                                         <span class='hs-comment'>-- x = [[[[1, 2, 3], [4, 5, 6]],</span>
+<a name="line-10798"></a>                                         <span class='hs-comment'>--       [[7, 8, 9], [10, 11, 12]]]]</span>
+<a name="line-10799"></a>                                         <span class='hs-comment'>-- ```</span>
+<a name="line-10800"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10801"></a>                                         <span class='hs-comment'>-- The output tensor has shape `[4, 1, 1, 3]` and value:</span>
+<a name="line-10802"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10803"></a>                                         <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10804"></a>                                         <span class='hs-comment'>-- [[[1, 2, 3]], [[4, 5, 6]], [[7, 8, 9]], [[10, 11, 12]]]</span>
+<a name="line-10805"></a>                                         <span class='hs-comment'>-- ```</span>
+<a name="line-10806"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10807"></a>                                         <span class='hs-comment'>-- (3) For the following input of shape `[1, 4, 4, 1]`, `block_shape = [2, 2]`, and</span>
+<a name="line-10808"></a>                                         <span class='hs-comment'>--     `paddings = [[0, 0], [0, 0]]`:</span>
+<a name="line-10809"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10810"></a>                                         <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10811"></a>                                         <span class='hs-comment'>-- x = [[[[1],   [2],  [3],  [4]],</span>
+<a name="line-10812"></a>                                         <span class='hs-comment'>--       [[5],   [6],  [7],  [8]],</span>
+<a name="line-10813"></a>                                         <span class='hs-comment'>--       [[9],  [10], [11],  [12]],</span>
+<a name="line-10814"></a>                                         <span class='hs-comment'>--       [[13], [14], [15],  [16]]]]</span>
+<a name="line-10815"></a>                                         <span class='hs-comment'>-- ```</span>
+<a name="line-10816"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10817"></a>                                         <span class='hs-comment'>-- The output tensor has shape `[4, 2, 2, 1]` and value:</span>
+<a name="line-10818"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10819"></a>                                         <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10820"></a>                                         <span class='hs-comment'>-- x = [[[[1], [3]], [[5], [7]]],</span>
+<a name="line-10821"></a>                                         <span class='hs-comment'>--      [[[2], [4]], [[10], [12]]],</span>
+<a name="line-10822"></a>                                         <span class='hs-comment'>--      [[[5], [7]], [[13], [15]]],</span>
+<a name="line-10823"></a>                                         <span class='hs-comment'>--      [[[6], [8]], [[14], [16]]]]</span>
+<a name="line-10824"></a>                                         <span class='hs-comment'>-- ```</span>
+<a name="line-10825"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10826"></a>                                         <span class='hs-comment'>-- (4) For the following input of shape `[2, 2, 4, 1]`, block_shape = `[2, 2]`, and</span>
+<a name="line-10827"></a>                                         <span class='hs-comment'>--     paddings = `[[0, 0], [2, 0]]`:</span>
+<a name="line-10828"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10829"></a>                                         <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10830"></a>                                         <span class='hs-comment'>-- x = [[[[1],   [2],  [3],  [4]],</span>
+<a name="line-10831"></a>                                         <span class='hs-comment'>--       [[5],   [6],  [7],  [8]]],</span>
+<a name="line-10832"></a>                                         <span class='hs-comment'>--      [[[9],  [10], [11],  [12]],</span>
+<a name="line-10833"></a>                                         <span class='hs-comment'>--       [[13], [14], [15],  [16]]]]</span>
+<a name="line-10834"></a>                                         <span class='hs-comment'>-- ```</span>
+<a name="line-10835"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10836"></a>                                         <span class='hs-comment'>-- The output tensor has shape `[8, 1, 3, 1]` and value:</span>
+<a name="line-10837"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10838"></a>                                         <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10839"></a>                                         <span class='hs-comment'>-- x = [[[[0], [1], [3]]], [[[0], [9], [11]]],</span>
+<a name="line-10840"></a>                                         <span class='hs-comment'>--      [[[0], [2], [4]]], [[[0], [10], [12]]],</span>
+<a name="line-10841"></a>                                         <span class='hs-comment'>--      [[[0], [5], [7]]], [[[0], [13], [15]]],</span>
+<a name="line-10842"></a>                                         <span class='hs-comment'>--      [[[0], [6], [8]]], [[[0], [14], [16]]]]</span>
+<a name="line-10843"></a>                                         <span class='hs-comment'>-- ```</span>
+<a name="line-10844"></a>                                         <span class='hs-comment'>-- </span>
+<a name="line-10845"></a>                                         <span class='hs-comment'>-- Among others, this operation is useful for reducing atrous convolution into</span>
+<a name="line-10846"></a>                                         <span class='hs-comment'>-- regular convolution.</span>
+<a name="line-10847"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-10848"></a><span class='hs-definition'>spaceToBatchND</span> <span class='hs-varid'>input</span> <span class='hs-varid'>block_shape</span> <span class='hs-varid'>paddings</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10849"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SpaceToBatchND"</span>
+<a name="line-10850"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-10851"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tblock_shape"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tblock_shape</span><span class='hs-layout'>)</span>
+<a name="line-10852"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tpaddings"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tpaddings</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-10853"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>block_shape</span> <span class='hs-varid'>paddings</span>
+<a name="line-10854"></a><span class='hs-comment'>{-
+<a name="line-10855"></a>attr { name: "T" type: "type" }
+<a name="line-10856"></a>attr {
+<a name="line-10857"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-10858"></a>  default_value { type: DT_INT32 }
+<a name="line-10859"></a>  name: "Tblock_shape"
+<a name="line-10860"></a>  type: "type"
+<a name="line-10861"></a>}
+<a name="line-10862"></a>attr {
+<a name="line-10863"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-10864"></a>  default_value { type: DT_INT32 }
+<a name="line-10865"></a>  name: "Tpaddings"
+<a name="line-10866"></a>  type: "type"
+<a name="line-10867"></a>}
+<a name="line-10868"></a>input_arg {
+<a name="line-10869"></a>  description: "N-D with shape `input_shape = [batch] + spatial_shape + remaining_shape`,\nwhere spatial_shape has `M` dimensions."
+<a name="line-10870"></a>  name: "input"
+<a name="line-10871"></a>  type_attr: "T"
+<a name="line-10872"></a>}
+<a name="line-10873"></a>input_arg {
+<a name="line-10874"></a>  description: "1-D with shape `[M]`, all values must be &gt;= 1."
+<a name="line-10875"></a>  name: "block_shape"
+<a name="line-10876"></a>  type_attr: "Tblock_shape"
+<a name="line-10877"></a>}
+<a name="line-10878"></a>input_arg {
+<a name="line-10879"></a>  description: "2-D with shape `[M, 2]`, all values must be &gt;= 0.\n  `paddings[i] = [pad_start, pad_end]` specifies the padding for input dimension\n  `i + 1`, which corresponds to spatial dimension `i`.  It is required that\n  `block_shape[i]` divides `input_shape[i + 1] + pad_start + pad_end`.\n\nThis operation is equivalent to the following steps:\n\n1. Zero-pad the start and end of dimensions `[1, ..., M]` of the\n   input according to `paddings` to produce `padded` of shape `padded_shape`.\n\n2. Reshape `padded` to `reshaped_padded` of shape:\n     [batch] +\n     [padded_shape[1] / block_shape[0],\n       block_shape[0],\n      ...,\n      padded_shape[M] / block_shape[M-1],\n      block_shape[M-1]] +\n     remaining_shape\n\n3. Permute dimensions of `reshaped_padded` to produce\n   `permuted_reshaped_padded` of shape:\n     block_shape +\n     [batch] +\n     [padded_shape[1] / block_shape[0],\n      ...,\n      padded_shape[M] / block_shape[M-1]] +\n     remaining_shape\n\n4. Reshape `permuted_reshaped_padded` to flatten `block_shape` into the batch\n   dimension, producing an output tensor of shape:\n     [batch * prod(block_shape)] +\n     [padded_shape[1] / block_shape[0],\n      ...,\n      padded_shape[M] / block_shape[M-1]] +\n     remaining_shape\n\nSome examples:\n\n(1) For the following input of shape `[1, 2, 2, 1]`, `block_shape = [2, 2]`, and\n    `paddings = [[0, 0], [0, 0]]`:\n\n```prettyprint\nx = [[[[1], [2]], [[3], [4]]]]\n```\n\nThe output tensor has shape `[4, 1, 1, 1]` and value:\n\n```prettyprint\n[[[[1]]], [[[2]]], [[[3]]], [[[4]]]]\n```\n\n(2) For the following input of shape `[1, 2, 2, 3]`, `block_shape = [2, 2]`, and\n    `paddings = [[0, 0], [0, 0]]`:\n\n```prettyprint\nx = [[[[1, 2, 3], [4, 5, 6]],\n      [[7, 8, 9], [10, 11, 12]]]]\n```\n\nThe output tensor has shape `[4, 1, 1, 3]` and value:\n\n```prettyprint\n[[[1, 2, 3]], [[4, 5, 6]], [[7, 8, 9]], [[10, 11, 12]]]\n```\n\n(3) For the following input of shape `[1, 4, 4, 1]`, `block_shape = [2, 2]`, and\n    `paddings = [[0, 0], [0, 0]]`:\n\n```prettyprint\nx = [[[[1],   [2],  [3],  [4]],\n      [[5],   [6],  [7],  [8]],\n      [[9],  [10], [11],  [12]],\n      [[13], [14], [15],  [16]]]]\n```\n\nThe output tensor has shape `[4, 2, 2, 1]` and value:\n\n```prettyprint\nx = [[[[1], [3]], [[5], [7]]],\n     [[[2], [4]], [[10], [12]]],\n     [[[5], [7]], [[13], [15]]],\n     [[[6], [8]], [[14], [16]]]]\n```\n\n(4) For the following input of shape `[2, 2, 4, 1]`, block_shape = `[2, 2]`, and\n    paddings = `[[0, 0], [2, 0]]`:\n\n```prettyprint\nx = [[[[1],   [2],  [3],  [4]],\n      [[5],   [6],  [7],  [8]]],\n     [[[9],  [10], [11],  [12]],\n      [[13], [14], [15],  [16]]]]\n```\n\nThe output tensor has shape `[8, 1, 3, 1]` and value:\n\n```prettyprint\nx = [[[[0], [1], [3]]], [[[0], [9], [11]]],\n     [[[0], [2], [4]]], [[[0], [10], [12]]],\n     [[[0], [5], [7]]], [[[0], [13], [15]]],\n     [[[0], [6], [8]]], [[[0], [14], [16]]]]\n```\n\nAmong others, this operation is useful for reducing atrous convolution into\nregular convolution."
+<a name="line-10880"></a>  name: "paddings"
+<a name="line-10881"></a>  type_attr: "Tpaddings"
+<a name="line-10882"></a>}
+<a name="line-10883"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-10884"></a>-}</span>
+<a name="line-10885"></a>
+<a name="line-10886"></a><a name="spaceToBatch"></a><span class='hs-comment'>-- | SpaceToBatch for 4-D tensors of type T.</span>
+<a name="line-10887"></a><span class='hs-comment'>--</span>
+<a name="line-10888"></a><span class='hs-comment'>-- This is a legacy version of the more general SpaceToBatchND.</span>
+<a name="line-10889"></a><span class='hs-comment'>-- </span>
+<a name="line-10890"></a><span class='hs-comment'>-- Zero-pads and then rearranges (permutes) blocks of spatial data into batch.</span>
+<a name="line-10891"></a><span class='hs-comment'>-- More specifically, this op outputs a copy of the input tensor where values from</span>
+<a name="line-10892"></a><span class='hs-comment'>-- the `height` and `width` dimensions are moved to the `batch` dimension. After</span>
+<a name="line-10893"></a><span class='hs-comment'>-- the zero-padding, both `height` and `width` of the input must be divisible by the</span>
+<a name="line-10894"></a><span class='hs-comment'>-- block size.</span>
+<a name="line-10895"></a><span class='hs-definition'>spaceToBatch</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tpaddings</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tpaddings</span><span class='hs-layout'>,</span>
+<a name="line-10896"></a>                                            <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-10897"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tpaddings</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-10898"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __block_size__</span>
+<a name="line-10899"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: 4-D with shape `[batch, height, width, depth]`.</span>
+<a name="line-10900"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tpaddings</span> <span class='hs-comment'>-- ^ __paddings__: 2-D tensor of non-negative integers with shape `[2, 2]`. It specifies</span>
+<a name="line-10901"></a>                                       <span class='hs-comment'>--   the padding of the input with zeros across the spatial dimensions as follows:</span>
+<a name="line-10902"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10903"></a>                                       <span class='hs-comment'>--       paddings = [[pad_top, pad_bottom], [pad_left, pad_right]]</span>
+<a name="line-10904"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10905"></a>                                       <span class='hs-comment'>--   The effective spatial dimensions of the zero-padded input tensor will be:</span>
+<a name="line-10906"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10907"></a>                                       <span class='hs-comment'>--       height_pad = pad_top + height + pad_bottom</span>
+<a name="line-10908"></a>                                       <span class='hs-comment'>--       width_pad = pad_left + width + pad_right</span>
+<a name="line-10909"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10910"></a>                                       <span class='hs-comment'>-- The attr `block_size` must be greater than one. It indicates the block size.</span>
+<a name="line-10911"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10912"></a>                                       <span class='hs-comment'>--   * Non-overlapping blocks of size `block_size x block size` in the height and</span>
+<a name="line-10913"></a>                                       <span class='hs-comment'>--     width dimensions are rearranged into the batch dimension at each location.</span>
+<a name="line-10914"></a>                                       <span class='hs-comment'>--   * The batch of the output tensor is `batch * block_size * block_size`.</span>
+<a name="line-10915"></a>                                       <span class='hs-comment'>--   * Both height_pad and width_pad must be divisible by block_size.</span>
+<a name="line-10916"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10917"></a>                                       <span class='hs-comment'>-- The shape of the output will be:</span>
+<a name="line-10918"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10919"></a>                                       <span class='hs-comment'>--     [batch*block_size*block_size, height_pad/block_size, width_pad/block_size,</span>
+<a name="line-10920"></a>                                       <span class='hs-comment'>--      depth]</span>
+<a name="line-10921"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10922"></a>                                       <span class='hs-comment'>-- Some examples:</span>
+<a name="line-10923"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10924"></a>                                       <span class='hs-comment'>-- (1) For the following input of shape `[1, 2, 2, 1]` and block_size of 2:</span>
+<a name="line-10925"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10926"></a>                                       <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10927"></a>                                       <span class='hs-comment'>-- x = [[[[1], [2]], [[3], [4]]]]</span>
+<a name="line-10928"></a>                                       <span class='hs-comment'>-- ```</span>
+<a name="line-10929"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10930"></a>                                       <span class='hs-comment'>-- The output tensor has shape `[4, 1, 1, 1]` and value:</span>
+<a name="line-10931"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10932"></a>                                       <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10933"></a>                                       <span class='hs-comment'>-- [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]</span>
+<a name="line-10934"></a>                                       <span class='hs-comment'>-- ```</span>
+<a name="line-10935"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10936"></a>                                       <span class='hs-comment'>-- (2) For the following input of shape `[1, 2, 2, 3]` and block_size of 2:</span>
+<a name="line-10937"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10938"></a>                                       <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10939"></a>                                       <span class='hs-comment'>-- x = [[[[1, 2, 3], [4, 5, 6]],</span>
+<a name="line-10940"></a>                                       <span class='hs-comment'>--       [[7, 8, 9], [10, 11, 12]]]]</span>
+<a name="line-10941"></a>                                       <span class='hs-comment'>-- ```</span>
+<a name="line-10942"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10943"></a>                                       <span class='hs-comment'>-- The output tensor has shape `[4, 1, 1, 3]` and value:</span>
+<a name="line-10944"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10945"></a>                                       <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10946"></a>                                       <span class='hs-comment'>-- [[[1, 2, 3]], [[4, 5, 6]], [[7, 8, 9]], [[10, 11, 12]]]</span>
+<a name="line-10947"></a>                                       <span class='hs-comment'>-- ```</span>
+<a name="line-10948"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10949"></a>                                       <span class='hs-comment'>-- (3) For the following input of shape `[1, 4, 4, 1]` and block_size of 2:</span>
+<a name="line-10950"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10951"></a>                                       <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10952"></a>                                       <span class='hs-comment'>-- x = [[[[1],   [2],  [3],  [4]],</span>
+<a name="line-10953"></a>                                       <span class='hs-comment'>--       [[5],   [6],  [7],  [8]],</span>
+<a name="line-10954"></a>                                       <span class='hs-comment'>--       [[9],  [10], [11],  [12]],</span>
+<a name="line-10955"></a>                                       <span class='hs-comment'>--       [[13], [14], [15],  [16]]]]</span>
+<a name="line-10956"></a>                                       <span class='hs-comment'>-- ```</span>
+<a name="line-10957"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10958"></a>                                       <span class='hs-comment'>-- The output tensor has shape `[4, 2, 2, 1]` and value:</span>
+<a name="line-10959"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10960"></a>                                       <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10961"></a>                                       <span class='hs-comment'>-- x = [[[[1], [3]], [[5], [7]]],</span>
+<a name="line-10962"></a>                                       <span class='hs-comment'>--      [[[2], [4]], [[10], [12]]],</span>
+<a name="line-10963"></a>                                       <span class='hs-comment'>--      [[[5], [7]], [[13], [15]]],</span>
+<a name="line-10964"></a>                                       <span class='hs-comment'>--      [[[6], [8]], [[14], [16]]]]</span>
+<a name="line-10965"></a>                                       <span class='hs-comment'>-- ```</span>
+<a name="line-10966"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10967"></a>                                       <span class='hs-comment'>-- (4) For the following input of shape `[2, 2, 4, 1]` and block_size of 2:</span>
+<a name="line-10968"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10969"></a>                                       <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10970"></a>                                       <span class='hs-comment'>-- x = [[[[1],   [2],  [3],  [4]],</span>
+<a name="line-10971"></a>                                       <span class='hs-comment'>--       [[5],   [6],  [7],  [8]]],</span>
+<a name="line-10972"></a>                                       <span class='hs-comment'>--      [[[9],  [10], [11],  [12]],</span>
+<a name="line-10973"></a>                                       <span class='hs-comment'>--       [[13], [14], [15],  [16]]]]</span>
+<a name="line-10974"></a>                                       <span class='hs-comment'>-- ```</span>
+<a name="line-10975"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10976"></a>                                       <span class='hs-comment'>-- The output tensor has shape `[8, 1, 2, 1]` and value:</span>
+<a name="line-10977"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10978"></a>                                       <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-10979"></a>                                       <span class='hs-comment'>-- x = [[[[1], [3]]], [[[9], [11]]], [[[2], [4]]], [[[10], [12]]],</span>
+<a name="line-10980"></a>                                       <span class='hs-comment'>--      [[[5], [7]]], [[[13], [15]]], [[[6], [8]]], [[[14], [16]]]]</span>
+<a name="line-10981"></a>                                       <span class='hs-comment'>-- ```</span>
+<a name="line-10982"></a>                                       <span class='hs-comment'>-- </span>
+<a name="line-10983"></a>                                       <span class='hs-comment'>-- Among others, this operation is useful for reducing atrous convolution into</span>
+<a name="line-10984"></a>                                       <span class='hs-comment'>-- regular convolution.</span>
+<a name="line-10985"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-10986"></a><span class='hs-definition'>spaceToBatch</span> <span class='hs-varid'>block_size</span> <span class='hs-varid'>input</span> <span class='hs-varid'>paddings</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-10987"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SpaceToBatch"</span>
+<a name="line-10988"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-10989"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tpaddings"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tpaddings</span><span class='hs-layout'>)</span>
+<a name="line-10990"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"block_size"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>block_size</span><span class='hs-layout'>)</span>
+<a name="line-10991"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>paddings</span>
+<a name="line-10992"></a><span class='hs-comment'>{-
+<a name="line-10993"></a>attr { name: "T" type: "type" }
+<a name="line-10994"></a>attr {
+<a name="line-10995"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-10996"></a>  default_value { type: DT_INT32 }
+<a name="line-10997"></a>  name: "Tpaddings"
+<a name="line-10998"></a>  type: "type"
+<a name="line-10999"></a>}
+<a name="line-11000"></a>attr {
+<a name="line-11001"></a>  has_minimum: true minimum: 2 name: "block_size" type: "int"
+<a name="line-11002"></a>}
+<a name="line-11003"></a>input_arg {
+<a name="line-11004"></a>  description: "4-D with shape `[batch, height, width, depth]`."
+<a name="line-11005"></a>  name: "input"
+<a name="line-11006"></a>  type_attr: "T"
+<a name="line-11007"></a>}
+<a name="line-11008"></a>input_arg {
+<a name="line-11009"></a>  description: "2-D tensor of non-negative integers with shape `[2, 2]`. It specifies\n  the padding of the input with zeros across the spatial dimensions as follows:\n\n      paddings = [[pad_top, pad_bottom], [pad_left, pad_right]]\n\n  The effective spatial dimensions of the zero-padded input tensor will be:\n\n      height_pad = pad_top + height + pad_bottom\n      width_pad = pad_left + width + pad_right\n\nThe attr `block_size` must be greater than one. It indicates the block size.\n\n  * Non-overlapping blocks of size `block_size x block size` in the height and\n    width dimensions are rearranged into the batch dimension at each location.\n  * The batch of the output tensor is `batch * block_size * block_size`.\n  * Both height_pad and width_pad must be divisible by block_size.\n\nThe shape of the output will be:\n\n    [batch*block_size*block_size, height_pad/block_size, width_pad/block_size,\n     depth]\n\nSome examples:\n\n(1) For the following input of shape `[1, 2, 2, 1]` and block_size of 2:\n\n```prettyprint\nx = [[[[1], [2]], [[3], [4]]]]\n```\n\nThe output tensor has shape `[4, 1, 1, 1]` and value:\n\n```prettyprint\n[[[[1]]], [[[2]]], [[[3]]], [[[4]]]]\n```\n\n(2) For the following input of shape `[1, 2, 2, 3]` and block_size of 2:\n\n```prettyprint\nx = [[[[1, 2, 3], [4, 5, 6]],\n      [[7, 8, 9], [10, 11, 12]]]]\n```\n\nThe output tensor has shape `[4, 1, 1, 3]` and value:\n\n```prettyprint\n[[[1, 2, 3]], [[4, 5, 6]], [[7, 8, 9]], [[10, 11, 12]]]\n```\n\n(3) For the following input of shape `[1, 4, 4, 1]` and block_size of 2:\n\n```prettyprint\nx = [[[[1],   [2],  [3],  [4]],\n      [[5],   [6],  [7],  [8]],\n      [[9],  [10], [11],  [12]],\n      [[13], [14], [15],  [16]]]]\n```\n\nThe output tensor has shape `[4, 2, 2, 1]` and value:\n\n```prettyprint\nx = [[[[1], [3]], [[5], [7]]],\n     [[[2], [4]], [[10], [12]]],\n     [[[5], [7]], [[13], [15]]],\n     [[[6], [8]], [[14], [16]]]]\n```\n\n(4) For the following input of shape `[2, 2, 4, 1]` and block_size of 2:\n\n```prettyprint\nx = [[[[1],   [2],  [3],  [4]],\n      [[5],   [6],  [7],  [8]]],\n     [[[9],  [10], [11],  [12]],\n      [[13], [14], [15],  [16]]]]\n```\n\nThe output tensor has shape `[8, 1, 2, 1]` and value:\n\n```prettyprint\nx = [[[[1], [3]]], [[[9], [11]]], [[[2], [4]]], [[[10], [12]]],\n     [[[5], [7]]], [[[13], [15]]], [[[6], [8]]], [[[14], [16]]]]\n```\n\nAmong others, this operation is useful for reducing atrous convolution into\nregular convolution."
+<a name="line-11010"></a>  name: "paddings"
+<a name="line-11011"></a>  type_attr: "Tpaddings"
+<a name="line-11012"></a>}
+<a name="line-11013"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-11014"></a>-}</span>
+<a name="line-11015"></a>
+<a name="line-11016"></a><a name="cTCGreedyDecoder"></a><span class='hs-comment'>-- | Performs greedy decoding on the logits given in inputs.</span>
+<a name="line-11017"></a><span class='hs-comment'>--</span>
+<a name="line-11018"></a><span class='hs-comment'>-- A note about the attribute merge_repeated: if enabled, when</span>
+<a name="line-11019"></a><span class='hs-comment'>-- consecutive logits' maximum indices are the same, only the first of</span>
+<a name="line-11020"></a><span class='hs-comment'>-- these is emitted.  Labeling the blank '*', the sequence "A B B * B B"</span>
+<a name="line-11021"></a><span class='hs-comment'>-- becomes "A B" if merge_repeated = True and "A B B B B" if</span>
+<a name="line-11022"></a><span class='hs-comment'>-- merge_repeated = False.</span>
+<a name="line-11023"></a><span class='hs-comment'>-- </span>
+<a name="line-11024"></a><span class='hs-comment'>-- Regardless of the value of merge_repeated, if the maximum index of a given</span>
+<a name="line-11025"></a><span class='hs-comment'>-- time and batch corresponds to the blank, index `(num_classes - 1)`, no new</span>
+<a name="line-11026"></a><span class='hs-comment'>-- element is emitted.</span>
+<a name="line-11027"></a><span class='hs-definition'>cTCGreedyDecoder</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __inputs__: 3-D, shape: `(max_time x batch_size x num_classes)`, the logits.</span>
+<a name="line-11028"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __sequence_length__: A vector containing sequence lengths, size `(batch_size)`.</span>
+<a name="line-11029"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-11030"></a>                        <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-11031"></a>                        <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span>
+<a name="line-11032"></a>                    <span class='hs-comment'>-- ^ (__decoded_indices__, __decoded_values__, __decoded_shape__, __log_probability__)</span>
+<a name="line-11033"></a>                    <span class='hs-comment'>--</span>
+<a name="line-11034"></a>                    <span class='hs-comment'>-- * __decoded_indices__: Indices matrix, size `(total_decoded_outputs x 2)`,</span>
+<a name="line-11035"></a>                    <span class='hs-comment'>-- of a `SparseTensor&lt;int64, 2&gt;`.  The rows store: [batch, time].</span>
+<a name="line-11036"></a>                    <span class='hs-comment'>--</span>
+<a name="line-11037"></a>                    <span class='hs-comment'>-- * __decoded_values__: Values vector, size: `(total_decoded_outputs)`,</span>
+<a name="line-11038"></a>                    <span class='hs-comment'>-- of a `SparseTensor&lt;int64, 2&gt;`.  The vector stores the decoded classes.</span>
+<a name="line-11039"></a>                    <span class='hs-comment'>--</span>
+<a name="line-11040"></a>                    <span class='hs-comment'>-- * __decoded_shape__: Shape vector, size `(2)`, of the decoded SparseTensor.</span>
+<a name="line-11041"></a>                    <span class='hs-comment'>-- Values are: `[batch_size, max_decoded_length]`.</span>
+<a name="line-11042"></a>                    <span class='hs-comment'>--</span>
+<a name="line-11043"></a>                    <span class='hs-comment'>-- * __log_probability__: Matrix, size `(batch_size x 1)`, containing sequence</span>
+<a name="line-11044"></a>                    <span class='hs-comment'>-- log-probabilities.</span>
+<a name="line-11045"></a><span class='hs-definition'>cTCGreedyDecoder</span> <span class='hs-varid'>inputs</span> <span class='hs-varid'>sequence_length</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-11046"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"CTCGreedyDecoder"</span><span class='hs-layout'>)</span>
+<a name="line-11047"></a>        <span class='hs-varid'>inputs</span> <span class='hs-varid'>sequence_length</span>
+<a name="line-11048"></a><span class='hs-comment'>{-
+<a name="line-11049"></a>attr {
+<a name="line-11050"></a>  default_value { b: false }
+<a name="line-11051"></a>  description: "If True, merge repeated classes in output."
+<a name="line-11052"></a>  name: "merge_repeated"
+<a name="line-11053"></a>  type: "bool"
+<a name="line-11054"></a>}
+<a name="line-11055"></a>input_arg {
+<a name="line-11056"></a>  description: "3-D, shape: `(max_time x batch_size x num_classes)`, the logits."
+<a name="line-11057"></a>  name: "inputs"
+<a name="line-11058"></a>  type: DT_FLOAT
+<a name="line-11059"></a>}
+<a name="line-11060"></a>input_arg {
+<a name="line-11061"></a>  description: "A vector containing sequence lengths, size `(batch_size)`."
+<a name="line-11062"></a>  name: "sequence_length"
+<a name="line-11063"></a>  type: DT_INT32
+<a name="line-11064"></a>}
+<a name="line-11065"></a>output_arg {
+<a name="line-11066"></a>  description: "Indices matrix, size `(total_decoded_outputs x 2)`,\nof a `SparseTensor&lt;int64, 2&gt;`.  The rows store: [batch, time]."
+<a name="line-11067"></a>  name: "decoded_indices"
+<a name="line-11068"></a>  type: DT_INT64
+<a name="line-11069"></a>}
+<a name="line-11070"></a>output_arg {
+<a name="line-11071"></a>  description: "Values vector, size: `(total_decoded_outputs)`,\nof a `SparseTensor&lt;int64, 2&gt;`.  The vector stores the decoded classes."
+<a name="line-11072"></a>  name: "decoded_values"
+<a name="line-11073"></a>  type: DT_INT64
+<a name="line-11074"></a>}
+<a name="line-11075"></a>output_arg {
+<a name="line-11076"></a>  description: "Shape vector, size `(2)`, of the decoded SparseTensor.\nValues are: `[batch_size, max_decoded_length]`."
+<a name="line-11077"></a>  name: "decoded_shape"
+<a name="line-11078"></a>  type: DT_INT64
+<a name="line-11079"></a>}
+<a name="line-11080"></a>output_arg {
+<a name="line-11081"></a>  description: "Matrix, size `(batch_size x 1)`, containing sequence\nlog-probabilities."
+<a name="line-11082"></a>  name: "log_probability"
+<a name="line-11083"></a>  type: DT_FLOAT
+<a name="line-11084"></a>}
+<a name="line-11085"></a>-}</span>
+<a name="line-11086"></a>
+<a name="line-11087"></a><a name="batchToSpaceND"></a><span class='hs-comment'>-- | BatchToSpace for N-D tensors of type T.</span>
+<a name="line-11088"></a><span class='hs-comment'>--</span>
+<a name="line-11089"></a><span class='hs-comment'>-- This operation reshapes the "batch" dimension 0 into `M + 1` dimensions of shape</span>
+<a name="line-11090"></a><span class='hs-comment'>-- `block_shape + [batch]`, interleaves these blocks back into the grid defined by</span>
+<a name="line-11091"></a><span class='hs-comment'>-- the spatial dimensions `[1, ..., M]`, to obtain a result with the same rank as</span>
+<a name="line-11092"></a><span class='hs-comment'>-- the input.  The spatial dimensions of this intermediate result are then</span>
+<a name="line-11093"></a><span class='hs-comment'>-- optionally cropped according to `crops` to produce the output.  This is the</span>
+<a name="line-11094"></a><span class='hs-comment'>-- reverse of SpaceToBatch.  See below for a precise description.</span>
+<a name="line-11095"></a><span class='hs-definition'>batchToSpaceND</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tblock_shape</span> <span class='hs-varid'>tcrops</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-11096"></a>                                                           <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tblock_shape</span><span class='hs-layout'>,</span>
+<a name="line-11097"></a>                                                           <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-11098"></a>                                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tblock_shape</span><span class='hs-layout'>,</span>
+<a name="line-11099"></a>                                                           <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tcrops</span><span class='hs-layout'>,</span>
+<a name="line-11100"></a>                                                           <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-11101"></a>                                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tcrops</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-11102"></a>                  <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: N-D with shape `input_shape = [batch] + spatial_shape + remaining_shape`,</span>
+<a name="line-11103"></a>                              <span class='hs-comment'>-- where spatial_shape has M dimensions.</span>
+<a name="line-11104"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tblock_shape</span> <span class='hs-comment'>-- ^ __block_shape__: 1-D with shape `[M]`, all values must be &gt;= 1.</span>
+<a name="line-11105"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>tcrops</span> <span class='hs-comment'>-- ^ __crops__: 2-D with shape `[M, 2]`, all values must be &gt;= 0.</span>
+<a name="line-11106"></a>                                      <span class='hs-comment'>--   `crops[i] = [crop_start, crop_end]` specifies the amount to crop from input</span>
+<a name="line-11107"></a>                                      <span class='hs-comment'>--   dimension `i + 1`, which corresponds to spatial dimension `i`.  It is</span>
+<a name="line-11108"></a>                                      <span class='hs-comment'>--   required that</span>
+<a name="line-11109"></a>                                      <span class='hs-comment'>--   `crop_start[i] + crop_end[i] &lt;= block_shape[i] * input_shape[i + 1]`.</span>
+<a name="line-11110"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11111"></a>                                      <span class='hs-comment'>-- This operation is equivalent to the following steps:</span>
+<a name="line-11112"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11113"></a>                                      <span class='hs-comment'>-- 1. Reshape `input` to `reshaped` of shape:</span>
+<a name="line-11114"></a>                                      <span class='hs-comment'>--      [block_shape[0], ..., block_shape[M-1],</span>
+<a name="line-11115"></a>                                      <span class='hs-comment'>--       batch / prod(block_shape),</span>
+<a name="line-11116"></a>                                      <span class='hs-comment'>--       input_shape[1], ..., input_shape[N-1]]</span>
+<a name="line-11117"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11118"></a>                                      <span class='hs-comment'>-- 2. Permute dimensions of `reshaped` to produce `permuted` of shape</span>
+<a name="line-11119"></a>                                      <span class='hs-comment'>--      [batch / prod(block_shape),</span>
+<a name="line-11120"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11121"></a>                                      <span class='hs-comment'>--       input_shape[1], block_shape[0],</span>
+<a name="line-11122"></a>                                      <span class='hs-comment'>--       ...,</span>
+<a name="line-11123"></a>                                      <span class='hs-comment'>--       input_shape[M], block_shape[M-1],</span>
+<a name="line-11124"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11125"></a>                                      <span class='hs-comment'>--       input_shape[M+1], ..., input_shape[N-1]]</span>
+<a name="line-11126"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11127"></a>                                      <span class='hs-comment'>-- 3. Reshape `permuted` to produce `reshaped_permuted` of shape</span>
+<a name="line-11128"></a>                                      <span class='hs-comment'>--      [batch / prod(block_shape),</span>
+<a name="line-11129"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11130"></a>                                      <span class='hs-comment'>--       input_shape[1] * block_shape[0],</span>
+<a name="line-11131"></a>                                      <span class='hs-comment'>--       ...,</span>
+<a name="line-11132"></a>                                      <span class='hs-comment'>--       input_shape[M] * block_shape[M-1],</span>
+<a name="line-11133"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11134"></a>                                      <span class='hs-comment'>--       input_shape[M+1],</span>
+<a name="line-11135"></a>                                      <span class='hs-comment'>--       ...,</span>
+<a name="line-11136"></a>                                      <span class='hs-comment'>--       input_shape[N-1]]</span>
+<a name="line-11137"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11138"></a>                                      <span class='hs-comment'>-- 4. Crop the start and end of dimensions `[1, ..., M]` of</span>
+<a name="line-11139"></a>                                      <span class='hs-comment'>--    `reshaped_permuted` according to `crops` to produce the output of shape:</span>
+<a name="line-11140"></a>                                      <span class='hs-comment'>--      [batch / prod(block_shape),</span>
+<a name="line-11141"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11142"></a>                                      <span class='hs-comment'>--       input_shape[1] * block_shape[0] - crops[0,0] - crops[0,1],</span>
+<a name="line-11143"></a>                                      <span class='hs-comment'>--       ...,</span>
+<a name="line-11144"></a>                                      <span class='hs-comment'>--       input_shape[M] * block_shape[M-1] - crops[M-1,0] - crops[M-1,1],</span>
+<a name="line-11145"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11146"></a>                                      <span class='hs-comment'>--       input_shape[M+1], ..., input_shape[N-1]]</span>
+<a name="line-11147"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11148"></a>                                      <span class='hs-comment'>-- Some examples:</span>
+<a name="line-11149"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11150"></a>                                      <span class='hs-comment'>-- (1) For the following input of shape `[4, 1, 1, 1]`, `block_shape = [2, 2]`, and</span>
+<a name="line-11151"></a>                                      <span class='hs-comment'>--     `crops = [[0, 0], [0, 0]]`:</span>
+<a name="line-11152"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11153"></a>                                      <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-11154"></a>                                      <span class='hs-comment'>-- [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]</span>
+<a name="line-11155"></a>                                      <span class='hs-comment'>-- ```</span>
+<a name="line-11156"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11157"></a>                                      <span class='hs-comment'>-- The output tensor has shape `[1, 2, 2, 1]` and value:</span>
+<a name="line-11158"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11159"></a>                                      <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-11160"></a>                                      <span class='hs-comment'>-- x = [[[[1], [2]], [[3], [4]]]]</span>
+<a name="line-11161"></a>                                      <span class='hs-comment'>-- ```</span>
+<a name="line-11162"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11163"></a>                                      <span class='hs-comment'>-- (2) For the following input of shape `[4, 1, 1, 3]`, `block_shape = [2, 2]`, and</span>
+<a name="line-11164"></a>                                      <span class='hs-comment'>--     `crops = [[0, 0], [0, 0]]`:</span>
+<a name="line-11165"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11166"></a>                                      <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-11167"></a>                                      <span class='hs-comment'>-- [[[1, 2, 3]], [[4, 5, 6]], [[7, 8, 9]], [[10, 11, 12]]]</span>
+<a name="line-11168"></a>                                      <span class='hs-comment'>-- ```</span>
+<a name="line-11169"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11170"></a>                                      <span class='hs-comment'>-- The output tensor has shape `[1, 2, 2, 3]` and value:</span>
+<a name="line-11171"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11172"></a>                                      <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-11173"></a>                                      <span class='hs-comment'>-- x = [[[[1, 2, 3], [4, 5, 6]],</span>
+<a name="line-11174"></a>                                      <span class='hs-comment'>--       [[7, 8, 9], [10, 11, 12]]]]</span>
+<a name="line-11175"></a>                                      <span class='hs-comment'>-- ```</span>
+<a name="line-11176"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11177"></a>                                      <span class='hs-comment'>-- (3) For the following input of shape `[4, 2, 2, 1]`, `block_shape = [2, 2]`, and</span>
+<a name="line-11178"></a>                                      <span class='hs-comment'>--     `crops = [[0, 0], [0, 0]]`:</span>
+<a name="line-11179"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11180"></a>                                      <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-11181"></a>                                      <span class='hs-comment'>-- x = [[[[1], [3]], [[5], [7]]],</span>
+<a name="line-11182"></a>                                      <span class='hs-comment'>--      [[[2], [4]], [[10], [12]]],</span>
+<a name="line-11183"></a>                                      <span class='hs-comment'>--      [[[5], [7]], [[13], [15]]],</span>
+<a name="line-11184"></a>                                      <span class='hs-comment'>--      [[[6], [8]], [[14], [16]]]]</span>
+<a name="line-11185"></a>                                      <span class='hs-comment'>-- ```</span>
+<a name="line-11186"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11187"></a>                                      <span class='hs-comment'>-- The output tensor has shape `[1, 4, 4, 1]` and value:</span>
+<a name="line-11188"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11189"></a>                                      <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-11190"></a>                                      <span class='hs-comment'>-- x = [[[1],   [2],  [3],  [4]],</span>
+<a name="line-11191"></a>                                      <span class='hs-comment'>--      [[5],   [6],  [7],  [8]],</span>
+<a name="line-11192"></a>                                      <span class='hs-comment'>--      [[9],  [10], [11],  [12]],</span>
+<a name="line-11193"></a>                                      <span class='hs-comment'>--      [[13], [14], [15],  [16]]]</span>
+<a name="line-11194"></a>                                      <span class='hs-comment'>-- ```</span>
+<a name="line-11195"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11196"></a>                                      <span class='hs-comment'>-- (4) For the following input of shape `[8, 1, 3, 1]`, `block_shape = [2, 2]`, and</span>
+<a name="line-11197"></a>                                      <span class='hs-comment'>--     `crops = [[0, 0], [2, 0]]`:</span>
+<a name="line-11198"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11199"></a>                                      <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-11200"></a>                                      <span class='hs-comment'>-- x = [[[[0], [1], [3]]], [[[0], [9], [11]]],</span>
+<a name="line-11201"></a>                                      <span class='hs-comment'>--      [[[0], [2], [4]]], [[[0], [10], [12]]],</span>
+<a name="line-11202"></a>                                      <span class='hs-comment'>--      [[[0], [5], [7]]], [[[0], [13], [15]]],</span>
+<a name="line-11203"></a>                                      <span class='hs-comment'>--      [[[0], [6], [8]]], [[[0], [14], [16]]]]</span>
+<a name="line-11204"></a>                                      <span class='hs-comment'>-- ```</span>
+<a name="line-11205"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11206"></a>                                      <span class='hs-comment'>-- The output tensor has shape `[2, 2, 4, 1]` and value:</span>
+<a name="line-11207"></a>                                      <span class='hs-comment'>-- </span>
+<a name="line-11208"></a>                                      <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-11209"></a>                                      <span class='hs-comment'>-- x = [[[[1],   [2],  [3],  [4]],</span>
+<a name="line-11210"></a>                                      <span class='hs-comment'>--       [[5],   [6],  [7],  [8]]],</span>
+<a name="line-11211"></a>                                      <span class='hs-comment'>--      [[[9],  [10], [11],  [12]],</span>
+<a name="line-11212"></a>                                      <span class='hs-comment'>--       [[13], [14], [15],  [16]]]]</span>
+<a name="line-11213"></a>                                      <span class='hs-comment'>-- ```</span>
+<a name="line-11214"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-11215"></a><span class='hs-definition'>batchToSpaceND</span> <span class='hs-varid'>input</span> <span class='hs-varid'>block_shape</span> <span class='hs-varid'>crops</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-11216"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchToSpaceND"</span>
+<a name="line-11217"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-11218"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tblock_shape"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tblock_shape</span><span class='hs-layout'>)</span>
+<a name="line-11219"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tcrops"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tcrops</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-11220"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>block_shape</span> <span class='hs-varid'>crops</span>
+<a name="line-11221"></a><span class='hs-comment'>{-
+<a name="line-11222"></a>attr { name: "T" type: "type" }
+<a name="line-11223"></a>attr {
+<a name="line-11224"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-11225"></a>  default_value { type: DT_INT32 }
+<a name="line-11226"></a>  name: "Tblock_shape"
+<a name="line-11227"></a>  type: "type"
+<a name="line-11228"></a>}
+<a name="line-11229"></a>attr {
+<a name="line-11230"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-11231"></a>  default_value { type: DT_INT32 }
+<a name="line-11232"></a>  name: "Tcrops"
+<a name="line-11233"></a>  type: "type"
+<a name="line-11234"></a>}
+<a name="line-11235"></a>input_arg {
+<a name="line-11236"></a>  description: "N-D with shape `input_shape = [batch] + spatial_shape + remaining_shape`,\nwhere spatial_shape has M dimensions."
+<a name="line-11237"></a>  name: "input"
+<a name="line-11238"></a>  type_attr: "T"
+<a name="line-11239"></a>}
+<a name="line-11240"></a>input_arg {
+<a name="line-11241"></a>  description: "1-D with shape `[M]`, all values must be &gt;= 1."
+<a name="line-11242"></a>  name: "block_shape"
+<a name="line-11243"></a>  type_attr: "Tblock_shape"
+<a name="line-11244"></a>}
+<a name="line-11245"></a>input_arg {
+<a name="line-11246"></a>  description: "2-D with shape `[M, 2]`, all values must be &gt;= 0.\n  `crops[i] = [crop_start, crop_end]` specifies the amount to crop from input\n  dimension `i + 1`, which corresponds to spatial dimension `i`.  It is\n  required that\n  `crop_start[i] + crop_end[i] &lt;= block_shape[i] * input_shape[i + 1]`.\n\nThis operation is equivalent to the following steps:\n\n1. Reshape `input` to `reshaped` of shape:\n     [block_shape[0], ..., block_shape[M-1],\n      batch / prod(block_shape),\n      input_shape[1], ..., input_shape[N-1]]\n\n2. Permute dimensions of `reshaped` to produce `permuted` of shape\n     [batch / prod(block_shape),\n\n      input_shape[1], block_shape[0],\n      ...,\n      input_shape[M], block_shape[M-1],\n\n      input_shape[M+1], ..., input_shape[N-1]]\n\n3. Reshape `permuted` to produce `reshaped_permuted` of shape\n     [batch / prod(block_shape),\n\n      input_shape[1] * block_shape[0],\n      ...,\n      input_shape[M] * block_shape[M-1],\n\n      input_shape[M+1],\n      ...,\n      input_shape[N-1]]\n\n4. Crop the start and end of dimensions `[1, ..., M]` of\n   `reshaped_permuted` according to `crops` to produce the output of shape:\n     [batch / prod(block_shape),\n\n      input_shape[1] * block_shape[0] - crops[0,0] - crops[0,1],\n      ...,\n      input_shape[M] * block_shape[M-1] - crops[M-1,0] - crops[M-1,1],\n\n      input_shape[M+1], ..., input_shape[N-1]]\n\nSome examples:\n\n(1) For the following input of shape `[4, 1, 1, 1]`, `block_shape = [2, 2]`, and\n    `crops = [[0, 0], [0, 0]]`:\n\n```prettyprint\n[[[[1]]], [[[2]]], [[[3]]], [[[4]]]]\n```\n\nThe output tensor has shape `[1, 2, 2, 1]` and value:\n\n```prettyprint\nx = [[[[1], [2]], [[3], [4]]]]\n```\n\n(2) For the following input of shape `[4, 1, 1, 3]`, `block_shape = [2, 2]`, and\n    `crops = [[0, 0], [0, 0]]`:\n\n```prettyprint\n[[[1, 2, 3]], [[4, 5, 6]], [[7, 8, 9]], [[10, 11, 12]]]\n```\n\nThe output tensor has shape `[1, 2, 2, 3]` and value:\n\n```prettyprint\nx = [[[[1, 2, 3], [4, 5, 6]],\n      [[7, 8, 9], [10, 11, 12]]]]\n```\n\n(3) For the following input of shape `[4, 2, 2, 1]`, `block_shape = [2, 2]`, and\n    `crops = [[0, 0], [0, 0]]`:\n\n```prettyprint\nx = [[[[1], [3]], [[5], [7]]],\n     [[[2], [4]], [[10], [12]]],\n     [[[5], [7]], [[13], [15]]],\n     [[[6], [8]], [[14], [16]]]]\n```\n\nThe output tensor has shape `[1, 4, 4, 1]` and value:\n\n```prettyprint\nx = [[[1],   [2],  [3],  [4]],\n     [[5],   [6],  [7],  [8]],\n     [[9],  [10], [11],  [12]],\n     [[13], [14], [15],  [16]]]\n```\n\n(4) For the following input of shape `[8, 1, 3, 1]`, `block_shape = [2, 2]`, and\n    `crops = [[0, 0], [2, 0]]`:\n\n```prettyprint\nx = [[[[0], [1], [3]]], [[[0], [9], [11]]],\n     [[[0], [2], [4]]], [[[0], [10], [12]]],\n     [[[0], [5], [7]]], [[[0], [13], [15]]],\n     [[[0], [6], [8]]], [[[0], [14], [16]]]]\n```\n\nThe output tensor has shape `[2, 2, 4, 1]` and value:\n\n```prettyprint\nx = [[[[1],   [2],  [3],  [4]],\n      [[5],   [6],  [7],  [8]]],\n     [[[9],  [10], [11],  [12]],\n      [[13], [14], [15],  [16]]]]\n```"
+<a name="line-11247"></a>  name: "crops"
+<a name="line-11248"></a>  type_attr: "Tcrops"
+<a name="line-11249"></a>}
+<a name="line-11250"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-11251"></a>-}</span>
+<a name="line-11252"></a>
+<a name="line-11253"></a><a name="pack"></a><span class='hs-comment'>-- | Packs a list of `N` rank-`R` tensors into one rank-`(R+1)` tensor.</span>
+<a name="line-11254"></a><span class='hs-comment'>--</span>
+<a name="line-11255"></a><span class='hs-comment'>-- Packs the `N` tensors in `values` into a tensor with rank one higher than each</span>
+<a name="line-11256"></a><span class='hs-comment'>-- tensor in `values`, by packing them along the `axis` dimension.</span>
+<a name="line-11257"></a><span class='hs-comment'>-- Given a list of tensors of shape `(A, B, C)`;</span>
+<a name="line-11258"></a><span class='hs-comment'>-- </span>
+<a name="line-11259"></a><span class='hs-comment'>-- if `axis == 0` then the `output` tensor will have the shape `(N, A, B, C)`.</span>
+<a name="line-11260"></a><span class='hs-comment'>-- if `axis == 1` then the `output` tensor will have the shape `(A, N, B, C)`.</span>
+<a name="line-11261"></a><span class='hs-comment'>-- Etc.</span>
+<a name="line-11262"></a><span class='hs-comment'>-- </span>
+<a name="line-11263"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-11264"></a><span class='hs-comment'>-- </span>
+<a name="line-11265"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-11266"></a><span class='hs-comment'>-- # 'x' is [1, 4]</span>
+<a name="line-11267"></a><span class='hs-comment'>-- # 'y' is [2, 5]</span>
+<a name="line-11268"></a><span class='hs-comment'>-- # 'z' is [3, 6]</span>
+<a name="line-11269"></a><span class='hs-comment'>-- pack([x, y, z]) =&gt; [[1, 4], [2, 5], [3, 6]]  # Pack along first dim.</span>
+<a name="line-11270"></a><span class='hs-comment'>-- pack([x, y, z], axis=1) =&gt; [[1, 2, 3], [4, 5, 6]]</span>
+<a name="line-11271"></a><span class='hs-comment'>-- ```</span>
+<a name="line-11272"></a><span class='hs-comment'>-- </span>
+<a name="line-11273"></a><span class='hs-comment'>-- This is the opposite of `unpack`.</span>
+<a name="line-11274"></a><span class='hs-definition'>pack</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-11275"></a>        <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __values__: Must be of same shape and type.</span>
+<a name="line-11276"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The packed tensor.</span>
+<a name="line-11277"></a><span class='hs-definition'>pack</span> <span class='hs-varid'>values</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"N"</span><span class='hs-layout'>,</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"values"</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>values</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-11278"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Pack"</span>
+<a name="line-11279"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-11280"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"N"</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>values</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-11281"></a>        <span class='hs-varid'>values</span>
+<a name="line-11282"></a><span class='hs-comment'>{-
+<a name="line-11283"></a>attr { has_minimum: true minimum: 1 name: "N" type: "int" }
+<a name="line-11284"></a>attr { name: "T" type: "type" }
+<a name="line-11285"></a>attr {
+<a name="line-11286"></a>  default_value { i: 0 }
+<a name="line-11287"></a>  description: "Dimension along which to pack.  Negative values wrap around, so the\nvalid range is `[-(R+1), R+1)`."
+<a name="line-11288"></a>  name: "axis"
+<a name="line-11289"></a>  type: "int"
+<a name="line-11290"></a>}
+<a name="line-11291"></a>input_arg {
+<a name="line-11292"></a>  description: "Must be of same shape and type."
+<a name="line-11293"></a>  name: "values"
+<a name="line-11294"></a>  number_attr: "N"
+<a name="line-11295"></a>  type_attr: "T"
+<a name="line-11296"></a>}
+<a name="line-11297"></a>output_arg {
+<a name="line-11298"></a>  description: "The packed tensor." name: "output" type_attr: "T"
+<a name="line-11299"></a>}
+<a name="line-11300"></a>-}</span>
+<a name="line-11301"></a>
+<a name="line-11302"></a><a name="oneHot"></a><span class='hs-comment'>-- | Returns a one-hot tensor.</span>
+<a name="line-11303"></a><span class='hs-comment'>--</span>
+<a name="line-11304"></a><span class='hs-comment'>-- The locations represented by indices in `indices` take value `on_value`,</span>
+<a name="line-11305"></a><span class='hs-comment'>-- while all other locations take value `off_value`.</span>
+<a name="line-11306"></a><span class='hs-comment'>-- </span>
+<a name="line-11307"></a><span class='hs-comment'>-- If the input `indices` is rank `N`, the output will have rank `N+1`,</span>
+<a name="line-11308"></a><span class='hs-comment'>-- The new axis is created at dimension `axis` (default: the new axis is</span>
+<a name="line-11309"></a><span class='hs-comment'>-- appended at the end).</span>
+<a name="line-11310"></a><span class='hs-comment'>-- </span>
+<a name="line-11311"></a><span class='hs-comment'>-- If `indices` is a scalar the output shape will be a vector of length `depth`.</span>
+<a name="line-11312"></a><span class='hs-comment'>-- </span>
+<a name="line-11313"></a><span class='hs-comment'>-- If `indices` is a vector of length `features`, the output shape will be:</span>
+<a name="line-11314"></a><span class='hs-comment'>-- ```</span>
+<a name="line-11315"></a><span class='hs-comment'>--   features x depth if axis == -1</span>
+<a name="line-11316"></a><span class='hs-comment'>--   depth x features if axis == 0</span>
+<a name="line-11317"></a><span class='hs-comment'>-- ```</span>
+<a name="line-11318"></a><span class='hs-comment'>-- </span>
+<a name="line-11319"></a><span class='hs-comment'>-- If `indices` is a matrix (batch) with shape `[batch, features]`,</span>
+<a name="line-11320"></a><span class='hs-comment'>-- the output shape will be:</span>
+<a name="line-11321"></a><span class='hs-comment'>-- ```</span>
+<a name="line-11322"></a><span class='hs-comment'>--   batch x features x depth if axis == -1</span>
+<a name="line-11323"></a><span class='hs-comment'>--   batch x depth x features if axis == 1</span>
+<a name="line-11324"></a><span class='hs-comment'>--   depth x batch x features if axis == 0</span>
+<a name="line-11325"></a><span class='hs-comment'>-- ```</span>
+<a name="line-11326"></a><span class='hs-comment'>-- </span>
+<a name="line-11327"></a><span class='hs-comment'>-- </span>
+<a name="line-11328"></a><span class='hs-comment'>-- Examples</span>
+<a name="line-11329"></a><span class='hs-comment'>-- =========</span>
+<a name="line-11330"></a><span class='hs-comment'>-- </span>
+<a name="line-11331"></a><span class='hs-comment'>-- Suppose that</span>
+<a name="line-11332"></a><span class='hs-comment'>-- </span>
+<a name="line-11333"></a><span class='hs-comment'>-- ```</span>
+<a name="line-11334"></a><span class='hs-comment'>--   indices = [0, 2, -1, 1]</span>
+<a name="line-11335"></a><span class='hs-comment'>--   depth = 3</span>
+<a name="line-11336"></a><span class='hs-comment'>--   on_value = 5.0</span>
+<a name="line-11337"></a><span class='hs-comment'>--   off_value = 0.0</span>
+<a name="line-11338"></a><span class='hs-comment'>--   axis = -1</span>
+<a name="line-11339"></a><span class='hs-comment'>-- ```</span>
+<a name="line-11340"></a><span class='hs-comment'>-- </span>
+<a name="line-11341"></a><span class='hs-comment'>-- Then output is `[4 x 3]`:</span>
+<a name="line-11342"></a><span class='hs-comment'>-- </span>
+<a name="line-11343"></a><span class='hs-comment'>--     ```output =</span>
+<a name="line-11344"></a><span class='hs-comment'>--       [5.0 0.0 0.0]  // one_hot(0)</span>
+<a name="line-11345"></a><span class='hs-comment'>--       [0.0 0.0 5.0]  // one_hot(2)</span>
+<a name="line-11346"></a><span class='hs-comment'>--       [0.0 0.0 0.0]  // one_hot(-1)</span>
+<a name="line-11347"></a><span class='hs-comment'>--       [0.0 5.0 0.0]  // one_hot(1)</span>
+<a name="line-11348"></a><span class='hs-comment'>--     ```</span>
+<a name="line-11349"></a><span class='hs-comment'>-- </span>
+<a name="line-11350"></a><span class='hs-comment'>-- Suppose that</span>
+<a name="line-11351"></a><span class='hs-comment'>-- </span>
+<a name="line-11352"></a><span class='hs-comment'>-- ```</span>
+<a name="line-11353"></a><span class='hs-comment'>--   indices = [0, 2, -1, 1]</span>
+<a name="line-11354"></a><span class='hs-comment'>--   depth = 3</span>
+<a name="line-11355"></a><span class='hs-comment'>--   on_value = 0.0</span>
+<a name="line-11356"></a><span class='hs-comment'>--   off_value = 3.0</span>
+<a name="line-11357"></a><span class='hs-comment'>--   axis = 0</span>
+<a name="line-11358"></a><span class='hs-comment'>-- ```</span>
+<a name="line-11359"></a><span class='hs-comment'>-- </span>
+<a name="line-11360"></a><span class='hs-comment'>-- Then output is `[3 x 4]`:</span>
+<a name="line-11361"></a><span class='hs-comment'>-- </span>
+<a name="line-11362"></a><span class='hs-comment'>--     ```output =</span>
+<a name="line-11363"></a><span class='hs-comment'>--       [0.0 3.0 3.0 3.0]</span>
+<a name="line-11364"></a><span class='hs-comment'>--       [3.0 3.0 3.0 0.0]</span>
+<a name="line-11365"></a><span class='hs-comment'>--       [3.0 3.0 3.0 3.0]</span>
+<a name="line-11366"></a><span class='hs-comment'>--       [3.0 0.0 3.0 3.0]</span>
+<a name="line-11367"></a><span class='hs-comment'>--     //  ^                one_hot(0)</span>
+<a name="line-11368"></a><span class='hs-comment'>--     //      ^            one_hot(2)</span>
+<a name="line-11369"></a><span class='hs-comment'>--     //          ^        one_hot(-1)</span>
+<a name="line-11370"></a><span class='hs-comment'>--     //              ^    one_hot(1)</span>
+<a name="line-11371"></a><span class='hs-comment'>--     ```</span>
+<a name="line-11372"></a><span class='hs-comment'>-- Suppose that</span>
+<a name="line-11373"></a><span class='hs-comment'>-- </span>
+<a name="line-11374"></a><span class='hs-comment'>-- ```</span>
+<a name="line-11375"></a><span class='hs-comment'>--   indices = [[0, 2], [1, -1]]</span>
+<a name="line-11376"></a><span class='hs-comment'>--   depth = 3</span>
+<a name="line-11377"></a><span class='hs-comment'>--   on_value = 1.0</span>
+<a name="line-11378"></a><span class='hs-comment'>--   off_value = 0.0</span>
+<a name="line-11379"></a><span class='hs-comment'>--   axis = -1</span>
+<a name="line-11380"></a><span class='hs-comment'>-- ```</span>
+<a name="line-11381"></a><span class='hs-comment'>-- </span>
+<a name="line-11382"></a><span class='hs-comment'>-- Then output is `[2 x 2 x 3]`:</span>
+<a name="line-11383"></a><span class='hs-comment'>-- </span>
+<a name="line-11384"></a><span class='hs-comment'>--     ```output =</span>
+<a name="line-11385"></a><span class='hs-comment'>--       [</span>
+<a name="line-11386"></a><span class='hs-comment'>--         [1.0, 0.0, 0.0]  // one_hot(0)</span>
+<a name="line-11387"></a><span class='hs-comment'>--         [0.0, 0.0, 1.0]  // one_hot(2)</span>
+<a name="line-11388"></a><span class='hs-comment'>--       ][</span>
+<a name="line-11389"></a><span class='hs-comment'>--         [0.0, 1.0, 0.0]  // one_hot(1)</span>
+<a name="line-11390"></a><span class='hs-comment'>--         [0.0, 0.0, 0.0]  // one_hot(-1)</span>
+<a name="line-11391"></a><span class='hs-comment'>--       ]```</span>
+<a name="line-11392"></a><span class='hs-definition'>oneHot</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tI</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tI</span><span class='hs-layout'>,</span>
+<a name="line-11393"></a>                                     <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-11394"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tI</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-11395"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>tI</span> <span class='hs-comment'>-- ^ __indices__: A tensor of indices.</span>
+<a name="line-11396"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __depth__: A scalar defining the depth of the one hot dimension.</span>
+<a name="line-11397"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __on_value__: A scalar defining the value to fill in output when `indices[j] = i`.</span>
+<a name="line-11398"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __off_value__: A scalar defining the value to fill in output when `indices[j] != i`.</span>
+<a name="line-11399"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The one-hot tensor.</span>
+<a name="line-11400"></a><span class='hs-definition'>oneHot</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>depth</span> <span class='hs-varid'>on_value</span> <span class='hs-varid'>off_value</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-11401"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"OneHot"</span>
+<a name="line-11402"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-11403"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"TI"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tI</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-11404"></a>        <span class='hs-varid'>indices</span> <span class='hs-varid'>depth</span> <span class='hs-varid'>on_value</span> <span class='hs-varid'>off_value</span>
+<a name="line-11405"></a><span class='hs-comment'>{-
+<a name="line-11406"></a>attr {
+<a name="line-11407"></a>  default_value { i: -1 }
+<a name="line-11408"></a>  description: "The axis to fill (default: -1, a new inner-most axis)."
+<a name="line-11409"></a>  name: "axis"
+<a name="line-11410"></a>  type: "int"
+<a name="line-11411"></a>}
+<a name="line-11412"></a>attr { name: "T" type: "type" }
+<a name="line-11413"></a>attr {
+<a name="line-11414"></a>  allowed_values {
+<a name="line-11415"></a>    list { type: DT_UINT8 type: DT_INT32 type: DT_INT64 }
+<a name="line-11416"></a>  }
+<a name="line-11417"></a>  default_value { type: DT_INT64 }
+<a name="line-11418"></a>  name: "TI"
+<a name="line-11419"></a>  type: "type"
+<a name="line-11420"></a>}
+<a name="line-11421"></a>input_arg {
+<a name="line-11422"></a>  description: "A tensor of indices." name: "indices" type_attr: "TI"
+<a name="line-11423"></a>}
+<a name="line-11424"></a>input_arg {
+<a name="line-11425"></a>  description: "A scalar defining the depth of the one hot dimension."
+<a name="line-11426"></a>  name: "depth"
+<a name="line-11427"></a>  type: DT_INT32
+<a name="line-11428"></a>}
+<a name="line-11429"></a>input_arg {
+<a name="line-11430"></a>  description: "A scalar defining the value to fill in output when `indices[j] = i`."
+<a name="line-11431"></a>  name: "on_value"
+<a name="line-11432"></a>  type_attr: "T"
+<a name="line-11433"></a>}
+<a name="line-11434"></a>input_arg {
+<a name="line-11435"></a>  description: "A scalar defining the value to fill in output when `indices[j] != i`."
+<a name="line-11436"></a>  name: "off_value"
+<a name="line-11437"></a>  type_attr: "T"
+<a name="line-11438"></a>}
+<a name="line-11439"></a>output_arg {
+<a name="line-11440"></a>  description: "The one-hot tensor." name: "output" type_attr: "T"
+<a name="line-11441"></a>}
+<a name="line-11442"></a>-}</span>
+<a name="line-11443"></a>
+<a name="line-11444"></a><a name="broadcastGradientArgs"></a><span class='hs-comment'>-- | Return the reduction indices for computing gradients of s0 op s1 with broadcast.</span>
+<a name="line-11445"></a><span class='hs-comment'>--</span>
+<a name="line-11446"></a><span class='hs-comment'>-- This is typically used by gradient computations for a broadcasting operation.</span>
+<a name="line-11447"></a><span class='hs-definition'>broadcastGradientArgs</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-11448"></a>                                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-11449"></a>                         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __s0__</span>
+<a name="line-11450"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __s1__</span>
+<a name="line-11451"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-11452"></a>                         <span class='hs-comment'>-- ^ (__r0__, __r1__)</span>
+<a name="line-11453"></a>                         <span class='hs-comment'>--</span>
+<a name="line-11454"></a>                         <span class='hs-comment'>-- * __r0__</span>
+<a name="line-11455"></a>                         <span class='hs-comment'>--</span>
+<a name="line-11456"></a>                         <span class='hs-comment'>-- * __r1__</span>
+<a name="line-11457"></a><span class='hs-definition'>broadcastGradientArgs</span> <span class='hs-varid'>s0</span> <span class='hs-varid'>s1</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-11458"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BroadcastGradientArgs"</span>
+<a name="line-11459"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-11460"></a>        <span class='hs-varid'>s0</span> <span class='hs-varid'>s1</span>
+<a name="line-11461"></a><span class='hs-comment'>{-
+<a name="line-11462"></a>attr {
+<a name="line-11463"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-11464"></a>  default_value { type: DT_INT32 }
+<a name="line-11465"></a>  name: "T"
+<a name="line-11466"></a>  type: "type"
+<a name="line-11467"></a>}
+<a name="line-11468"></a>input_arg { name: "s0" type_attr: "T" }
+<a name="line-11469"></a>input_arg { name: "s1" type_attr: "T" }
+<a name="line-11470"></a>output_arg { name: "r0" type_attr: "T" }
+<a name="line-11471"></a>output_arg { name: "r1" type_attr: "T" }
+<a name="line-11472"></a>-}</span>
+<a name="line-11473"></a>
+<a name="line-11474"></a><a name="matrixSetDiag"></a><span class='hs-comment'>-- | Returns a batched matrix tensor with new batched diagonal values.</span>
+<a name="line-11475"></a><span class='hs-comment'>--</span>
+<a name="line-11476"></a><span class='hs-comment'>-- Given `input` and `diagonal`, this operation returns a tensor with the</span>
+<a name="line-11477"></a><span class='hs-comment'>-- same shape and values as `input`, except for the diagonals of the innermost</span>
+<a name="line-11478"></a><span class='hs-comment'>-- matrices.  These will be overwritten by the values in `diagonal`.</span>
+<a name="line-11479"></a><span class='hs-comment'>-- The batched matrices must be square.</span>
+<a name="line-11480"></a><span class='hs-comment'>-- </span>
+<a name="line-11481"></a><span class='hs-comment'>-- The output is computed as follows:</span>
+<a name="line-11482"></a><span class='hs-comment'>-- </span>
+<a name="line-11483"></a><span class='hs-comment'>-- Assume `input` has `k+1` dimensions `[I, J, K, ..., N, N]` and `diagonal` has</span>
+<a name="line-11484"></a><span class='hs-comment'>-- `k` dimensions `[I, J, K, ..., N]`.  Then the output is a</span>
+<a name="line-11485"></a><span class='hs-comment'>-- tensor of rank `k+1` with dimensions [I, J, K, ..., N, N]` where:</span>
+<a name="line-11486"></a><span class='hs-comment'>-- </span>
+<a name="line-11487"></a><span class='hs-comment'>--   * `output[i, j, k, ..., m, n] = diagonal[i, j, k, ..., n]` for `m == n`.</span>
+<a name="line-11488"></a><span class='hs-comment'>--   * `output[i, j, k, ..., m, n] = input[i, j, k, ..., m, n]` for `m != n`.</span>
+<a name="line-11489"></a><span class='hs-definition'>matrixSetDiag</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-11490"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Rank `k+1`, where `k &gt;= 1`.</span>
+<a name="line-11491"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __diagonal__: Rank `k`, where `k &gt;= 1`.</span>
+<a name="line-11492"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Rank `k+1`, with `output.shape = input.shape`.</span>
+<a name="line-11493"></a><span class='hs-definition'>matrixSetDiag</span> <span class='hs-varid'>input</span> <span class='hs-varid'>diagonal</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-11494"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MatrixSetDiag"</span>
+<a name="line-11495"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-11496"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>diagonal</span>
+<a name="line-11497"></a><span class='hs-comment'>{-
+<a name="line-11498"></a>attr { name: "T" type: "type" }
+<a name="line-11499"></a>input_arg {
+<a name="line-11500"></a>  description: "Rank `k+1`, where `k &gt;= 1`."
+<a name="line-11501"></a>  name: "input"
+<a name="line-11502"></a>  type_attr: "T"
+<a name="line-11503"></a>}
+<a name="line-11504"></a>input_arg {
+<a name="line-11505"></a>  description: "Rank `k`, where `k &gt;= 1`."
+<a name="line-11506"></a>  name: "diagonal"
+<a name="line-11507"></a>  type_attr: "T"
+<a name="line-11508"></a>}
+<a name="line-11509"></a>output_arg {
+<a name="line-11510"></a>  description: "Rank `k+1`, with `output.shape = input.shape`."
+<a name="line-11511"></a>  name: "output"
+<a name="line-11512"></a>  type_attr: "T"
+<a name="line-11513"></a>}
+<a name="line-11514"></a>-}</span>
+<a name="line-11515"></a>
+<a name="line-11516"></a><a name="applyRMSProp"></a><span class='hs-comment'>-- | Update '*var' according to the RMSProp algorithm.</span>
+<a name="line-11517"></a><span class='hs-comment'>--</span>
+<a name="line-11518"></a><span class='hs-comment'>-- Note that in dense implement of this algorithm, ms and mom will</span>
+<a name="line-11519"></a><span class='hs-comment'>-- update even if the grad is zero, but in this sparse implement, ms</span>
+<a name="line-11520"></a><span class='hs-comment'>-- and mom will not update in iterations the grad is zero.</span>
+<a name="line-11521"></a><span class='hs-comment'>-- </span>
+<a name="line-11522"></a><span class='hs-comment'>-- mean_square = decay * mean_square + (1-decay) * gradient ** 2</span>
+<a name="line-11523"></a><span class='hs-comment'>-- Delta = learning_rate * gradient / sqrt(mean_square + epsilon)</span>
+<a name="line-11524"></a><span class='hs-comment'>-- </span>
+<a name="line-11525"></a><span class='hs-comment'>-- ms &lt;- rho * ms_{t-1} + (1-rho) * grad * grad</span>
+<a name="line-11526"></a><span class='hs-comment'>-- mom &lt;- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)</span>
+<a name="line-11527"></a><span class='hs-comment'>-- var &lt;- var - mom</span>
+<a name="line-11528"></a><span class='hs-definition'>applyRMSProp</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>v8</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-11529"></a>                                                    <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-11530"></a>                                                            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-11531"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-11532"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-11533"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-11534"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-11535"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-11536"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-11537"></a>                                                            <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-11538"></a>                <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-11539"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __ms__: Should be from a Variable().</span>
+<a name="line-11540"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __mom__: Should be from a Variable().</span>
+<a name="line-11541"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr__: Scaling factor. Must be a scalar.</span>
+<a name="line-11542"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __rho__: Decay rate. Must be a scalar.</span>
+<a name="line-11543"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __momentum__</span>
+<a name="line-11544"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __epsilon__: Ridge term. Must be a scalar.</span>
+<a name="line-11545"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v8</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: The gradient.</span>
+<a name="line-11546"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-11547"></a><span class='hs-definition'>applyRMSProp</span> <span class='hs-varid'>var</span> <span class='hs-varid'>ms</span> <span class='hs-varid'>mom</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>rho</span> <span class='hs-varid'>momentum</span> <span class='hs-varid'>epsilon</span> <span class='hs-varid'>grad</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-11548"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ApplyRMSProp"</span>
+<a name="line-11549"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-11550"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>ms</span> <span class='hs-varid'>mom</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>rho</span> <span class='hs-varid'>momentum</span> <span class='hs-varid'>epsilon</span> <span class='hs-varid'>grad</span>
+<a name="line-11551"></a><span class='hs-comment'>{-
+<a name="line-11552"></a>attr {
+<a name="line-11553"></a>  allowed_values {
+<a name="line-11554"></a>    list {
+<a name="line-11555"></a>      type: DT_FLOAT
+<a name="line-11556"></a>      type: DT_DOUBLE
+<a name="line-11557"></a>      type: DT_INT64
+<a name="line-11558"></a>      type: DT_INT32
+<a name="line-11559"></a>      type: DT_UINT8
+<a name="line-11560"></a>      type: DT_UINT16
+<a name="line-11561"></a>      type: DT_INT16
+<a name="line-11562"></a>      type: DT_INT8
+<a name="line-11563"></a>      type: DT_COMPLEX64
+<a name="line-11564"></a>      type: DT_COMPLEX128
+<a name="line-11565"></a>      type: DT_QINT8
+<a name="line-11566"></a>      type: DT_QUINT8
+<a name="line-11567"></a>      type: DT_QINT32
+<a name="line-11568"></a>      type: DT_HALF
+<a name="line-11569"></a>    }
+<a name="line-11570"></a>  }
+<a name="line-11571"></a>  name: "T"
+<a name="line-11572"></a>  type: "type"
+<a name="line-11573"></a>}
+<a name="line-11574"></a>attr {
+<a name="line-11575"></a>  default_value { b: false }
+<a name="line-11576"></a>  description: "If `True`, updating of the var, m, and v tensors will be protected\nby a lock; otherwise the behavior is undefined, but may exhibit less\ncontention."
+<a name="line-11577"></a>  name: "use_locking"
+<a name="line-11578"></a>  type: "bool"
+<a name="line-11579"></a>}
+<a name="line-11580"></a>input_arg {
+<a name="line-11581"></a>  description: "Should be from a Variable()."
+<a name="line-11582"></a>  is_ref: true
+<a name="line-11583"></a>  name: "var"
+<a name="line-11584"></a>  type_attr: "T"
+<a name="line-11585"></a>}
+<a name="line-11586"></a>input_arg {
+<a name="line-11587"></a>  description: "Should be from a Variable()."
+<a name="line-11588"></a>  is_ref: true
+<a name="line-11589"></a>  name: "ms"
+<a name="line-11590"></a>  type_attr: "T"
+<a name="line-11591"></a>}
+<a name="line-11592"></a>input_arg {
+<a name="line-11593"></a>  description: "Should be from a Variable()."
+<a name="line-11594"></a>  is_ref: true
+<a name="line-11595"></a>  name: "mom"
+<a name="line-11596"></a>  type_attr: "T"
+<a name="line-11597"></a>}
+<a name="line-11598"></a>input_arg {
+<a name="line-11599"></a>  description: "Scaling factor. Must be a scalar."
+<a name="line-11600"></a>  name: "lr"
+<a name="line-11601"></a>  type_attr: "T"
+<a name="line-11602"></a>}
+<a name="line-11603"></a>input_arg {
+<a name="line-11604"></a>  description: "Decay rate. Must be a scalar."
+<a name="line-11605"></a>  name: "rho"
+<a name="line-11606"></a>  type_attr: "T"
+<a name="line-11607"></a>}
+<a name="line-11608"></a>input_arg { name: "momentum" type_attr: "T" }
+<a name="line-11609"></a>input_arg {
+<a name="line-11610"></a>  description: "Ridge term. Must be a scalar."
+<a name="line-11611"></a>  name: "epsilon"
+<a name="line-11612"></a>  type_attr: "T"
+<a name="line-11613"></a>}
+<a name="line-11614"></a>input_arg {
+<a name="line-11615"></a>  description: "The gradient." name: "grad" type_attr: "T"
+<a name="line-11616"></a>}
+<a name="line-11617"></a>output_arg {
+<a name="line-11618"></a>  description: "Same as \"var\"."
+<a name="line-11619"></a>  is_ref: true
+<a name="line-11620"></a>  name: "out"
+<a name="line-11621"></a>  type_attr: "T"
+<a name="line-11622"></a>}
+<a name="line-11623"></a>-}</span>
+<a name="line-11624"></a>
+<a name="line-11625"></a><span class='hs-comment'>-- | Returns a constant tensor.</span>
+<a name="line-11626"></a>
+<a name="line-11627"></a><a name="const"></a><span class='hs-definition'>const</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-11628"></a><span class='hs-definition'>const</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-11629"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Const"</span>
+<a name="line-11630"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-11631"></a>        
+<a name="line-11632"></a><span class='hs-comment'>{-
+<a name="line-11633"></a>attr {
+<a name="line-11634"></a>  description: "Attr `value` is the tensor to return."
+<a name="line-11635"></a>  name: "value"
+<a name="line-11636"></a>  type: "tensor"
+<a name="line-11637"></a>}
+<a name="line-11638"></a>attr { name: "dtype" type: "type" }
+<a name="line-11639"></a>output_arg { name: "output" type_attr: "dtype" }
+<a name="line-11640"></a>-}</span>
+<a name="line-11641"></a>
+<a name="line-11642"></a><a name="enter"></a><span class='hs-comment'>-- | Creates or finds a child frame, and makes `data` available to the child frame.</span>
+<a name="line-11643"></a><span class='hs-comment'>--</span>
+<a name="line-11644"></a><span class='hs-comment'>-- This op is used together with `Exit` to create loops in the graph.</span>
+<a name="line-11645"></a><span class='hs-comment'>-- The unique `frame_name` is used by the `Executor` to identify frames. If</span>
+<a name="line-11646"></a><span class='hs-comment'>-- `is_constant` is true, `output` is a constant in the child frame; otherwise</span>
+<a name="line-11647"></a><span class='hs-comment'>-- it may be changed in the child frame. At most `parallel_iterations` iterations</span>
+<a name="line-11648"></a><span class='hs-comment'>-- are run in parallel in the child frame.</span>
+<a name="line-11649"></a><span class='hs-definition'>enter</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-11650"></a>         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__: The tensor to be made available to the child frame.</span>
+<a name="line-11651"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The same tensor as `data`.</span>
+<a name="line-11652"></a><span class='hs-definition'>enter</span> <span class='hs-varid'>data'</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-11653"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Enter"</span>
+<a name="line-11654"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-11655"></a>        <span class='hs-varid'>data'</span>
+<a name="line-11656"></a><span class='hs-comment'>{-
+<a name="line-11657"></a>attr { name: "T" type: "type" }
+<a name="line-11658"></a>attr {
+<a name="line-11659"></a>  description: "The name of the child frame."
+<a name="line-11660"></a>  name: "frame_name"
+<a name="line-11661"></a>  type: "string"
+<a name="line-11662"></a>}
+<a name="line-11663"></a>attr {
+<a name="line-11664"></a>  default_value { b: false }
+<a name="line-11665"></a>  description: "If true, the output is constant within the child frame."
+<a name="line-11666"></a>  name: "is_constant"
+<a name="line-11667"></a>  type: "bool"
+<a name="line-11668"></a>}
+<a name="line-11669"></a>attr {
+<a name="line-11670"></a>  default_value { i: 10 }
+<a name="line-11671"></a>  description: "The number of iterations allowed to run in parallel."
+<a name="line-11672"></a>  name: "parallel_iterations"
+<a name="line-11673"></a>  type: "int"
+<a name="line-11674"></a>}
+<a name="line-11675"></a>input_arg {
+<a name="line-11676"></a>  description: "The tensor to be made available to the child frame."
+<a name="line-11677"></a>  name: "data"
+<a name="line-11678"></a>  type_attr: "T"
+<a name="line-11679"></a>}
+<a name="line-11680"></a>output_arg {
+<a name="line-11681"></a>  description: "The same tensor as `data`."
+<a name="line-11682"></a>  name: "output"
+<a name="line-11683"></a>  type_attr: "T"
+<a name="line-11684"></a>}
+<a name="line-11685"></a>-}</span>
+<a name="line-11686"></a>
+<a name="line-11687"></a><a name="debugIdentity"></a><span class='hs-comment'>-- | Debug Identity Op.</span>
+<a name="line-11688"></a><span class='hs-comment'>--</span>
+<a name="line-11689"></a><span class='hs-comment'>-- Provides an identity mapping of the non-Ref type input tensor for debugging.</span>
+<a name="line-11690"></a><span class='hs-definition'>debugIdentity</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-11691"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Input tensor, non-Reference type.</span>
+<a name="line-11692"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Output tensor that equals the input tensor.</span>
+<a name="line-11693"></a><span class='hs-definition'>debugIdentity</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-11694"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DebugIdentity"</span>
+<a name="line-11695"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-11696"></a>        <span class='hs-varid'>input</span>
+<a name="line-11697"></a><span class='hs-comment'>{-
+<a name="line-11698"></a>attr { name: "T" type: "type" }
+<a name="line-11699"></a>attr {
+<a name="line-11700"></a>  default_value { s: "" }
+<a name="line-11701"></a>  description: "Name of the input tensor."
+<a name="line-11702"></a>  name: "tensor_name"
+<a name="line-11703"></a>  type: "string"
+<a name="line-11704"></a>}
+<a name="line-11705"></a>attr {
+<a name="line-11706"></a>  default_value { list { } }
+<a name="line-11707"></a>  description: "List of URLs to debug targets, e.g.,\nfile:///foo/tfdbg_dump, grpc:://localhost:11011"
+<a name="line-11708"></a>  name: "debug_urls"
+<a name="line-11709"></a>  type: "list(string)"
+<a name="line-11710"></a>}
+<a name="line-11711"></a>input_arg {
+<a name="line-11712"></a>  description: "Input tensor, non-Reference type."
+<a name="line-11713"></a>  name: "input"
+<a name="line-11714"></a>  type_attr: "T"
+<a name="line-11715"></a>}
+<a name="line-11716"></a>output_arg {
+<a name="line-11717"></a>  description: "Output tensor that equals the input tensor."
+<a name="line-11718"></a>  name: "output"
+<a name="line-11719"></a>  type_attr: "T"
+<a name="line-11720"></a>}
+<a name="line-11721"></a>-}</span>
+<a name="line-11722"></a>
+<a name="line-11723"></a><a name="debugNanCount"></a><span class='hs-comment'>-- | Debug NaN Value Counter Op</span>
+<a name="line-11724"></a><span class='hs-comment'>--</span>
+<a name="line-11725"></a><span class='hs-comment'>-- Counts number of NaNs in the input tensor, for debugging.</span>
+<a name="line-11726"></a><span class='hs-definition'>debugNanCount</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-11727"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Input tensor, non-Reference type.</span>
+<a name="line-11728"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __output__: An integer output tensor that is the number of NaNs in the input.</span>
+<a name="line-11729"></a><span class='hs-definition'>debugNanCount</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-11730"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DebugNanCount"</span>
+<a name="line-11731"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-11732"></a>        <span class='hs-varid'>input</span>
+<a name="line-11733"></a><span class='hs-comment'>{-
+<a name="line-11734"></a>attr { name: "T" type: "type" }
+<a name="line-11735"></a>attr {
+<a name="line-11736"></a>  default_value { s: "" }
+<a name="line-11737"></a>  description: "Name of the input tensor."
+<a name="line-11738"></a>  name: "tensor_name"
+<a name="line-11739"></a>  type: "string"
+<a name="line-11740"></a>}
+<a name="line-11741"></a>attr {
+<a name="line-11742"></a>  default_value { list { } }
+<a name="line-11743"></a>  description: "List of URLs to debug targets, e.g.,\nfile:///foo/tfdbg_dump, grpc:://localhost:11011"
+<a name="line-11744"></a>  name: "debug_urls"
+<a name="line-11745"></a>  type: "list(string)"
+<a name="line-11746"></a>}
+<a name="line-11747"></a>input_arg {
+<a name="line-11748"></a>  description: "Input tensor, non-Reference type."
+<a name="line-11749"></a>  name: "input"
+<a name="line-11750"></a>  type_attr: "T"
+<a name="line-11751"></a>}
+<a name="line-11752"></a>output_arg {
+<a name="line-11753"></a>  description: "An integer output tensor that is the number of NaNs in the input."
+<a name="line-11754"></a>  name: "output"
+<a name="line-11755"></a>  type: DT_INT64
+<a name="line-11756"></a>}
+<a name="line-11757"></a>-}</span>
+<a name="line-11758"></a>
+<a name="line-11759"></a><a name="batchNormWithGlobalNormalization"></a><span class='hs-comment'>-- | Batch normalization.</span>
+<a name="line-11760"></a><span class='hs-comment'>--</span>
+<a name="line-11761"></a><span class='hs-comment'>-- This op is deprecated. Prefer `tf.nn.batch_normalization`.</span>
+<a name="line-11762"></a><span class='hs-definition'>batchNormWithGlobalNormalization</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-11763"></a>                                                               <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-11764"></a>                                                                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-11765"></a>                                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-11766"></a>                                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-11767"></a>                                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-11768"></a>                                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-11769"></a>                                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-11770"></a>                                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-11771"></a>                                                                       <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-11772"></a>                                                                       <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-11773"></a>                                    <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __scale_after_normalization__: A bool indicating whether the resulted tensor</span>
+<a name="line-11774"></a>                                         <span class='hs-comment'>-- needs to be multiplied with gamma.</span>
+<a name="line-11775"></a>                                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __variance_epsilon__: A small float number to avoid dividing by 0.</span>
+<a name="line-11776"></a>                                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __t__: A 4D input Tensor.</span>
+<a name="line-11777"></a>                                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __m__: A 1D mean Tensor with size matching the last dimension of t.</span>
+<a name="line-11778"></a>                                                   <span class='hs-comment'>-- This is the first output from tf.nn.moments,</span>
+<a name="line-11779"></a>                                                   <span class='hs-comment'>-- or a saved moving average thereof.</span>
+<a name="line-11780"></a>                                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __v__: A 1D variance Tensor with size matching the last dimension of t.</span>
+<a name="line-11781"></a>                                                   <span class='hs-comment'>-- This is the second output from tf.nn.moments,</span>
+<a name="line-11782"></a>                                                   <span class='hs-comment'>-- or a saved moving average thereof.</span>
+<a name="line-11783"></a>                                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __beta__: A 1D beta Tensor with size matching the last dimension of t.</span>
+<a name="line-11784"></a>                                                   <span class='hs-comment'>-- An offset to be added to the normalized tensor.</span>
+<a name="line-11785"></a>                                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __gamma__: A 1D gamma Tensor with size matching the last dimension of t.</span>
+<a name="line-11786"></a>                                                   <span class='hs-comment'>-- If "scale_after_normalization" is true, this tensor will be multiplied</span>
+<a name="line-11787"></a>                                                   <span class='hs-comment'>-- with the normalized tensor.</span>
+<a name="line-11788"></a>                                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __result__</span>
+<a name="line-11789"></a><span class='hs-definition'>batchNormWithGlobalNormalization</span> <span class='hs-varid'>scale_after_normalization</span> <span class='hs-varid'>variance_epsilon</span> <span class='hs-varid'>t</span> <span class='hs-varid'>m</span>
+<a name="line-11790"></a>                                 <span class='hs-varid'>v</span> <span class='hs-varid'>beta</span> <span class='hs-varid'>gamma</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-11791"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchNormWithGlobalNormalization"</span>
+<a name="line-11792"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-11793"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"scale_after_normalization"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>scale_after_normalization</span>
+<a name="line-11794"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"variance_epsilon"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>variance_epsilon</span><span class='hs-layout'>)</span>
+<a name="line-11795"></a>        <span class='hs-varid'>t</span> <span class='hs-varid'>m</span> <span class='hs-varid'>v</span> <span class='hs-varid'>beta</span> <span class='hs-varid'>gamma</span>
+<a name="line-11796"></a><span class='hs-comment'>{-
+<a name="line-11797"></a>attr {
+<a name="line-11798"></a>  allowed_values {
+<a name="line-11799"></a>    list {
+<a name="line-11800"></a>      type: DT_FLOAT
+<a name="line-11801"></a>      type: DT_DOUBLE
+<a name="line-11802"></a>      type: DT_INT64
+<a name="line-11803"></a>      type: DT_INT32
+<a name="line-11804"></a>      type: DT_UINT8
+<a name="line-11805"></a>      type: DT_UINT16
+<a name="line-11806"></a>      type: DT_INT16
+<a name="line-11807"></a>      type: DT_INT8
+<a name="line-11808"></a>      type: DT_COMPLEX64
+<a name="line-11809"></a>      type: DT_COMPLEX128
+<a name="line-11810"></a>      type: DT_QINT8
+<a name="line-11811"></a>      type: DT_QUINT8
+<a name="line-11812"></a>      type: DT_QINT32
+<a name="line-11813"></a>      type: DT_HALF
+<a name="line-11814"></a>    }
+<a name="line-11815"></a>  }
+<a name="line-11816"></a>  name: "T"
+<a name="line-11817"></a>  type: "type"
+<a name="line-11818"></a>}
+<a name="line-11819"></a>attr {
+<a name="line-11820"></a>  description: "A small float number to avoid dividing by 0."
+<a name="line-11821"></a>  name: "variance_epsilon"
+<a name="line-11822"></a>  type: "float"
+<a name="line-11823"></a>}
+<a name="line-11824"></a>attr {
+<a name="line-11825"></a>  description: "A bool indicating whether the resulted tensor\nneeds to be multiplied with gamma."
+<a name="line-11826"></a>  name: "scale_after_normalization"
+<a name="line-11827"></a>  type: "bool"
+<a name="line-11828"></a>}
+<a name="line-11829"></a>input_arg {
+<a name="line-11830"></a>  description: "A 4D input Tensor." name: "t" type_attr: "T"
+<a name="line-11831"></a>}
+<a name="line-11832"></a>input_arg {
+<a name="line-11833"></a>  description: "A 1D mean Tensor with size matching the last dimension of t.\nThis is the first output from tf.nn.moments,\nor a saved moving average thereof."
+<a name="line-11834"></a>  name: "m"
+<a name="line-11835"></a>  type_attr: "T"
+<a name="line-11836"></a>}
+<a name="line-11837"></a>input_arg {
+<a name="line-11838"></a>  description: "A 1D variance Tensor with size matching the last dimension of t.\nThis is the second output from tf.nn.moments,\nor a saved moving average thereof."
+<a name="line-11839"></a>  name: "v"
+<a name="line-11840"></a>  type_attr: "T"
+<a name="line-11841"></a>}
+<a name="line-11842"></a>input_arg {
+<a name="line-11843"></a>  description: "A 1D beta Tensor with size matching the last dimension of t.\nAn offset to be added to the normalized tensor."
+<a name="line-11844"></a>  name: "beta"
+<a name="line-11845"></a>  type_attr: "T"
+<a name="line-11846"></a>}
+<a name="line-11847"></a>input_arg {
+<a name="line-11848"></a>  description: "A 1D gamma Tensor with size matching the last dimension of t.\nIf \"scale_after_normalization\" is true, this tensor will be multiplied\nwith the normalized tensor."
+<a name="line-11849"></a>  name: "gamma"
+<a name="line-11850"></a>  type_attr: "T"
+<a name="line-11851"></a>}
+<a name="line-11852"></a>output_arg { name: "result" type_attr: "T" }
+<a name="line-11853"></a>-}</span>
+<a name="line-11854"></a>
+<a name="line-11855"></a><span class='hs-comment'>-- | </span>
+<a name="line-11856"></a>
+<a name="line-11857"></a><a name="batchMatrixDiag"></a><span class='hs-definition'>batchMatrixDiag</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __diagonal__</span>
+<a name="line-11858"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-11859"></a><span class='hs-definition'>batchMatrixDiag</span> <span class='hs-varid'>diagonal</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-11860"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchMatrixDiag"</span>
+<a name="line-11861"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-11862"></a>        <span class='hs-varid'>diagonal</span>
+<a name="line-11863"></a><span class='hs-comment'>{-
+<a name="line-11864"></a>attr { name: "T" type: "type" }
+<a name="line-11865"></a>input_arg { name: "diagonal" type_attr: "T" }
+<a name="line-11866"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-11867"></a>-}</span>
+<a name="line-11868"></a>
+<a name="line-11869"></a><a name="unpack"></a><span class='hs-comment'>-- | Unpacks a given dimension of a rank-`R` tensor into `num` rank-`(R-1)` tensors.</span>
+<a name="line-11870"></a><span class='hs-comment'>--</span>
+<a name="line-11871"></a><span class='hs-comment'>-- Unpacks `num` tensors from `value` by chipping it along the `axis` dimension.</span>
+<a name="line-11872"></a><span class='hs-comment'>-- For example, given a tensor of shape `(A, B, C, D)`;</span>
+<a name="line-11873"></a><span class='hs-comment'>-- </span>
+<a name="line-11874"></a><span class='hs-comment'>-- If `axis == 0` then the i'th tensor in `output` is the slice `value[i, :, :, :]`</span>
+<a name="line-11875"></a><span class='hs-comment'>--   and each tensor in `output` will have shape `(B, C, D)`. (Note that the</span>
+<a name="line-11876"></a><span class='hs-comment'>--   dimension unpacked along is gone, unlike `split`).</span>
+<a name="line-11877"></a><span class='hs-comment'>-- </span>
+<a name="line-11878"></a><span class='hs-comment'>-- If `axis == 1` then the i'th tensor in `output` is the slice `value[:, i, :, :]`</span>
+<a name="line-11879"></a><span class='hs-comment'>--   and each tensor in `output` will have shape `(A, C, D)`.</span>
+<a name="line-11880"></a><span class='hs-comment'>-- Etc.</span>
+<a name="line-11881"></a><span class='hs-comment'>-- </span>
+<a name="line-11882"></a><span class='hs-comment'>-- This is the opposite of `pack`.</span>
+<a name="line-11883"></a><span class='hs-definition'>unpack</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num__</span>
+<a name="line-11884"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: 1-D or higher, with `axis` dimension size equal to `num`.</span>
+<a name="line-11885"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __output__: The list of tensors unpacked from `value`.</span>
+<a name="line-11886"></a><span class='hs-definition'>unpack</span> <span class='hs-varid'>num</span> <span class='hs-varid'>value</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-11887"></a>    <span class='hs-varid'>buildListOp</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>num</span><span class='hs-keyglyph'>]</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Unpack"</span>
+<a name="line-11888"></a>                       <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-11889"></a>                       <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num</span><span class='hs-layout'>)</span>
+<a name="line-11890"></a>        <span class='hs-varid'>value</span>
+<a name="line-11891"></a><span class='hs-comment'>{-
+<a name="line-11892"></a>attr { has_minimum: true name: "num" type: "int" }
+<a name="line-11893"></a>attr { name: "T" type: "type" }
+<a name="line-11894"></a>attr {
+<a name="line-11895"></a>  default_value { i: 0 }
+<a name="line-11896"></a>  description: "Dimension along which to unpack.  Negative values wrap around, so the\nvalid range is `[-R, R)`."
+<a name="line-11897"></a>  name: "axis"
+<a name="line-11898"></a>  type: "int"
+<a name="line-11899"></a>}
+<a name="line-11900"></a>input_arg {
+<a name="line-11901"></a>  description: "1-D or higher, with `axis` dimension size equal to `num`."
+<a name="line-11902"></a>  name: "value"
+<a name="line-11903"></a>  type_attr: "T"
+<a name="line-11904"></a>}
+<a name="line-11905"></a>output_arg {
+<a name="line-11906"></a>  description: "The list of tensors unpacked from `value`."
+<a name="line-11907"></a>  name: "output"
+<a name="line-11908"></a>  number_attr: "num"
+<a name="line-11909"></a>  type_attr: "T"
+<a name="line-11910"></a>}
+<a name="line-11911"></a>-}</span>
+<a name="line-11912"></a>
+<a name="line-11913"></a><a name="sparseSplit"></a><span class='hs-comment'>-- | Split a `SparseTensor` into `num_split` tensors along one dimension.</span>
+<a name="line-11914"></a><span class='hs-comment'>--</span>
+<a name="line-11915"></a><span class='hs-comment'>-- If the `shape[split_dim]` is not an integer multiple of `num_split`. Slices</span>
+<a name="line-11916"></a><span class='hs-comment'>-- `[0 : shape[split_dim] % num_split]` gets one extra dimension.</span>
+<a name="line-11917"></a><span class='hs-comment'>-- For example, if `split_dim = 1` and `num_split = 2` and the input is</span>
+<a name="line-11918"></a><span class='hs-comment'>-- </span>
+<a name="line-11919"></a><span class='hs-comment'>--     input_tensor = shape = [2, 7]</span>
+<a name="line-11920"></a><span class='hs-comment'>--     [    a   d e  ]</span>
+<a name="line-11921"></a><span class='hs-comment'>--     [b c          ]</span>
+<a name="line-11922"></a><span class='hs-comment'>-- </span>
+<a name="line-11923"></a><span class='hs-comment'>-- Graphically the output tensors are:</span>
+<a name="line-11924"></a><span class='hs-comment'>-- </span>
+<a name="line-11925"></a><span class='hs-comment'>--     output_tensor[0] = shape = [2, 4]</span>
+<a name="line-11926"></a><span class='hs-comment'>--     [    a  ]</span>
+<a name="line-11927"></a><span class='hs-comment'>--     [b c    ]</span>
+<a name="line-11928"></a><span class='hs-comment'>-- </span>
+<a name="line-11929"></a><span class='hs-comment'>--     output_tensor[1] = shape = [2, 3]</span>
+<a name="line-11930"></a><span class='hs-comment'>--     [ d e  ]</span>
+<a name="line-11931"></a><span class='hs-comment'>--     [      ]</span>
+<a name="line-11932"></a><span class='hs-definition'>sparseSplit</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-11933"></a>               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_split__: The number of ways to split.</span>
+<a name="line-11934"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __split_dim__: 0-D.  The dimension along which to split.  Must be in the range</span>
+<a name="line-11935"></a>                                           <span class='hs-comment'>-- `[0, rank(shape))`.</span>
+<a name="line-11936"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __indices__: 2-D tensor represents the indices of the sparse tensor.</span>
+<a name="line-11937"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __values__: 1-D tensor represents the values of the sparse tensor.</span>
+<a name="line-11938"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __shape__: 1-D. tensor represents the shape of the sparse tensor.</span>
+<a name="line-11939"></a>                                           <span class='hs-comment'>-- output indices: A list of 1-D tensors represents the indices of the output</span>
+<a name="line-11940"></a>                                           <span class='hs-comment'>-- sparse tensors.</span>
+<a name="line-11941"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-11942"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-11943"></a>               <span class='hs-comment'>-- ^ (__output_indices__, __output_values__, __output_shape__)</span>
+<a name="line-11944"></a>               <span class='hs-comment'>--</span>
+<a name="line-11945"></a>               <span class='hs-comment'>-- * __output_indices__</span>
+<a name="line-11946"></a>               <span class='hs-comment'>--</span>
+<a name="line-11947"></a>               <span class='hs-comment'>-- * __output_values__: A list of 1-D tensors represents the values of the output sparse</span>
+<a name="line-11948"></a>               <span class='hs-comment'>-- tensors.</span>
+<a name="line-11949"></a>               <span class='hs-comment'>--</span>
+<a name="line-11950"></a>               <span class='hs-comment'>-- * __output_shape__: A list of 1-D tensors represents the shape of the output sparse</span>
+<a name="line-11951"></a>               <span class='hs-comment'>-- tensors.</span>
+<a name="line-11952"></a><span class='hs-definition'>sparseSplit</span> <span class='hs-varid'>num_split</span> <span class='hs-varid'>split_dim</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>values</span> <span class='hs-varid'>shape</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-11953"></a>    <span class='hs-varid'>buildListOp</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>num_split</span><span class='hs-layout'>,</span> <span class='hs-varid'>num_split</span><span class='hs-layout'>,</span> <span class='hs-varid'>num_split</span><span class='hs-keyglyph'>]</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseSplit"</span>
+<a name="line-11954"></a>                                                   <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-11955"></a>                                                   <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_split"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_split</span><span class='hs-layout'>)</span>
+<a name="line-11956"></a>        <span class='hs-varid'>split_dim</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>values</span> <span class='hs-varid'>shape</span>
+<a name="line-11957"></a><span class='hs-comment'>{-
+<a name="line-11958"></a>attr {
+<a name="line-11959"></a>  description: "The number of ways to split."
+<a name="line-11960"></a>  has_minimum: true
+<a name="line-11961"></a>  minimum: 1
+<a name="line-11962"></a>  name: "num_split"
+<a name="line-11963"></a>  type: "int"
+<a name="line-11964"></a>}
+<a name="line-11965"></a>attr { name: "T" type: "type" }
+<a name="line-11966"></a>input_arg {
+<a name="line-11967"></a>  description: "0-D.  The dimension along which to split.  Must be in the range\n`[0, rank(shape))`."
+<a name="line-11968"></a>  name: "split_dim"
+<a name="line-11969"></a>  type: DT_INT64
+<a name="line-11970"></a>}
+<a name="line-11971"></a>input_arg {
+<a name="line-11972"></a>  description: "2-D tensor represents the indices of the sparse tensor."
+<a name="line-11973"></a>  name: "indices"
+<a name="line-11974"></a>  type: DT_INT64
+<a name="line-11975"></a>}
+<a name="line-11976"></a>input_arg {
+<a name="line-11977"></a>  description: "1-D tensor represents the values of the sparse tensor."
+<a name="line-11978"></a>  name: "values"
+<a name="line-11979"></a>  type_attr: "T"
+<a name="line-11980"></a>}
+<a name="line-11981"></a>input_arg {
+<a name="line-11982"></a>  description: "1-D. tensor represents the shape of the sparse tensor.\noutput indices: A list of 1-D tensors represents the indices of the output\nsparse tensors."
+<a name="line-11983"></a>  name: "shape"
+<a name="line-11984"></a>  type: DT_INT64
+<a name="line-11985"></a>}
+<a name="line-11986"></a>output_arg {
+<a name="line-11987"></a>  name: "output_indices" number_attr: "num_split" type: DT_INT64
+<a name="line-11988"></a>}
+<a name="line-11989"></a>output_arg {
+<a name="line-11990"></a>  description: "A list of 1-D tensors represents the values of the output sparse\ntensors."
+<a name="line-11991"></a>  name: "output_values"
+<a name="line-11992"></a>  number_attr: "num_split"
+<a name="line-11993"></a>  type_attr: "T"
+<a name="line-11994"></a>}
+<a name="line-11995"></a>output_arg {
+<a name="line-11996"></a>  description: "A list of 1-D tensors represents the shape of the output sparse\ntensors."
+<a name="line-11997"></a>  name: "output_shape"
+<a name="line-11998"></a>  number_attr: "num_split"
+<a name="line-11999"></a>  type: DT_INT64
+<a name="line-12000"></a>}
+<a name="line-12001"></a>-}</span>
+<a name="line-12002"></a>
+<a name="line-12003"></a><a name="mirrorPad"></a><span class='hs-comment'>-- | Pads a tensor with mirrored values.</span>
+<a name="line-12004"></a><span class='hs-comment'>--</span>
+<a name="line-12005"></a><span class='hs-comment'>-- This operation pads a `input` with mirrored values according to the `paddings`</span>
+<a name="line-12006"></a><span class='hs-comment'>-- you specify. `paddings` is an integer tensor with shape `[n, 2]`, where n is</span>
+<a name="line-12007"></a><span class='hs-comment'>-- the rank of `input`. For each dimension D of `input`, `paddings[D, 0]` indicates</span>
+<a name="line-12008"></a><span class='hs-comment'>-- how many values to add before the contents of `input` in that dimension, and</span>
+<a name="line-12009"></a><span class='hs-comment'>-- `paddings[D, 1]` indicates how many values to add after the contents of `input`</span>
+<a name="line-12010"></a><span class='hs-comment'>-- in that dimension. Both `paddings[D, 0]` and `paddings[D, 1]` must be no greater</span>
+<a name="line-12011"></a><span class='hs-comment'>-- than `input.dim_size(D)` (or `input.dim_size(D) - 1`) if `copy_border` is true</span>
+<a name="line-12012"></a><span class='hs-comment'>-- (if false, respectively).</span>
+<a name="line-12013"></a><span class='hs-comment'>-- </span>
+<a name="line-12014"></a><span class='hs-comment'>-- The padded size of each dimension D of the output is:</span>
+<a name="line-12015"></a><span class='hs-comment'>-- </span>
+<a name="line-12016"></a><span class='hs-comment'>-- `paddings(D, 0) + input.dim_size(D) + paddings(D, 1)`</span>
+<a name="line-12017"></a><span class='hs-comment'>-- </span>
+<a name="line-12018"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-12019"></a><span class='hs-comment'>-- </span>
+<a name="line-12020"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-12021"></a><span class='hs-comment'>-- # 't' is [[1, 2, 3], [4, 5, 6]].</span>
+<a name="line-12022"></a><span class='hs-comment'>-- # 'paddings' is [[1, 1]], [2, 2]].</span>
+<a name="line-12023"></a><span class='hs-comment'>-- # 'mode' is SYMMETRIC.</span>
+<a name="line-12024"></a><span class='hs-comment'>-- # rank of 't' is 2.</span>
+<a name="line-12025"></a><span class='hs-comment'>-- pad(t, paddings) ==&gt; [[2, 1, 1, 2, 3, 3, 2]</span>
+<a name="line-12026"></a><span class='hs-comment'>--                       [2, 1, 1, 2, 3, 3, 2]</span>
+<a name="line-12027"></a><span class='hs-comment'>--                       [5, 4, 4, 5, 6, 6, 5]</span>
+<a name="line-12028"></a><span class='hs-comment'>--                       [5, 4, 4, 5, 6, 6, 5]]</span>
+<a name="line-12029"></a><span class='hs-comment'>-- ```</span>
+<a name="line-12030"></a><span class='hs-definition'>mirrorPad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tpaddings</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tpaddings</span><span class='hs-layout'>,</span>
+<a name="line-12031"></a>                                         <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-12032"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tpaddings</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-12033"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: The input tensor to be padded.</span>
+<a name="line-12034"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tpaddings</span> <span class='hs-comment'>-- ^ __paddings__: A two-column matrix specifying the padding sizes. The number of</span>
+<a name="line-12035"></a>                                    <span class='hs-comment'>-- rows must be the same as the rank of `input`.</span>
+<a name="line-12036"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The padded tensor.</span>
+<a name="line-12037"></a><span class='hs-definition'>mirrorPad</span> <span class='hs-varid'>input</span> <span class='hs-varid'>paddings</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12038"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MirrorPad"</span>
+<a name="line-12039"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-12040"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tpaddings"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tpaddings</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-12041"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>paddings</span>
+<a name="line-12042"></a><span class='hs-comment'>{-
+<a name="line-12043"></a>attr { name: "T" type: "type" }
+<a name="line-12044"></a>attr {
+<a name="line-12045"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-12046"></a>  default_value { type: DT_INT32 }
+<a name="line-12047"></a>  name: "Tpaddings"
+<a name="line-12048"></a>  type: "type"
+<a name="line-12049"></a>}
+<a name="line-12050"></a>attr {
+<a name="line-12051"></a>  allowed_values { list { s: "REFLECT" s: "SYMMETRIC" } }
+<a name="line-12052"></a>  description: "Either `REFLECT` or `SYMMETRIC`. In reflect mode the padded regions\ndo not include the borders, while in symmetric mode the padded regions\ndo include the borders. For example, if `input` is `[1, 2, 3]` and `paddings`\nis `[0, 2]`, then the output is `[1, 2, 3, 2, 1]` in reflect mode, and\nit is `[1, 2, 3, 3, 2]` in symmetric mode."
+<a name="line-12053"></a>  name: "mode"
+<a name="line-12054"></a>  type: "string"
+<a name="line-12055"></a>}
+<a name="line-12056"></a>input_arg {
+<a name="line-12057"></a>  description: "The input tensor to be padded."
+<a name="line-12058"></a>  name: "input"
+<a name="line-12059"></a>  type_attr: "T"
+<a name="line-12060"></a>}
+<a name="line-12061"></a>input_arg {
+<a name="line-12062"></a>  description: "A two-column matrix specifying the padding sizes. The number of\nrows must be the same as the rank of `input`."
+<a name="line-12063"></a>  name: "paddings"
+<a name="line-12064"></a>  type_attr: "Tpaddings"
+<a name="line-12065"></a>}
+<a name="line-12066"></a>output_arg {
+<a name="line-12067"></a>  description: "The padded tensor." name: "output" type_attr: "T"
+<a name="line-12068"></a>}
+<a name="line-12069"></a>-}</span>
+<a name="line-12070"></a>
+<a name="line-12071"></a><span class='hs-comment'>-- | </span>
+<a name="line-12072"></a>
+<a name="line-12073"></a><a name="batchMatrixDiagPart"></a><span class='hs-definition'>batchMatrixDiagPart</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-12074"></a>                       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-12075"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __diagonal__</span>
+<a name="line-12076"></a><span class='hs-definition'>batchMatrixDiagPart</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12077"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchMatrixDiagPart"</span>
+<a name="line-12078"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-12079"></a>        <span class='hs-varid'>input</span>
+<a name="line-12080"></a><span class='hs-comment'>{-
+<a name="line-12081"></a>attr { name: "T" type: "type" }
+<a name="line-12082"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-12083"></a>output_arg { name: "diagonal" type_attr: "T" }
+<a name="line-12084"></a>-}</span>
+<a name="line-12085"></a>
+<a name="line-12086"></a><span class='hs-comment'>-- | Computes gradient of the FractionalMaxPool function.</span>
+<a name="line-12087"></a>
+<a name="line-12088"></a><a name="fractionalMaxPoolGrad"></a><span class='hs-definition'>fractionalMaxPoolGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-12089"></a>                                                    <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-12090"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-12091"></a>                                                            <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-12092"></a>                         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __orig_input__: Original input for `fractional_max_pool`</span>
+<a name="line-12093"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __orig_output__: Original output for `fractional_max_pool`</span>
+<a name="line-12094"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out_backprop__: 4-D with shape `[batch, height, width, channels]`.  Gradients</span>
+<a name="line-12095"></a>                                        <span class='hs-comment'>-- w.r.t. the output of `fractional_max_pool`.</span>
+<a name="line-12096"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __row_pooling_sequence__: row pooling sequence, form pooling region with</span>
+<a name="line-12097"></a>                                                     <span class='hs-comment'>-- col_pooling_sequence.</span>
+<a name="line-12098"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __col_pooling_sequence__: column pooling sequence, form pooling region with</span>
+<a name="line-12099"></a>                                                     <span class='hs-comment'>-- row_pooling sequence.</span>
+<a name="line-12100"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 4-D.  Gradients w.r.t. the input of `fractional_max_pool`.</span>
+<a name="line-12101"></a><span class='hs-definition'>fractionalMaxPoolGrad</span> <span class='hs-varid'>orig_input</span> <span class='hs-varid'>orig_output</span> <span class='hs-varid'>out_backprop</span> <span class='hs-varid'>row_pooling_sequence</span>
+<a name="line-12102"></a>                      <span class='hs-varid'>col_pooling_sequence</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12103"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"FractionalMaxPoolGrad"</span>
+<a name="line-12104"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-12105"></a>        <span class='hs-varid'>orig_input</span> <span class='hs-varid'>orig_output</span> <span class='hs-varid'>out_backprop</span> <span class='hs-varid'>row_pooling_sequence</span>
+<a name="line-12106"></a>        <span class='hs-varid'>col_pooling_sequence</span>
+<a name="line-12107"></a><span class='hs-comment'>{-
+<a name="line-12108"></a>attr {
+<a name="line-12109"></a>  default_value { b: false }
+<a name="line-12110"></a>  description: "When set to True, it means when pooling, the values at the boundary\nof adjacent pooling cells are used by both cells. For example:\n\n`index  0  1  2  3  4`\n\n`value  20 5  16 3  7`\n\nIf the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.\nThe result would be [20, 16] for fractional max pooling."
+<a name="line-12111"></a>  name: "overlapping"
+<a name="line-12112"></a>  type: "bool"
+<a name="line-12113"></a>}
+<a name="line-12114"></a>attr {
+<a name="line-12115"></a>  allowed_values {
+<a name="line-12116"></a>    list {
+<a name="line-12117"></a>      type: DT_FLOAT type: DT_DOUBLE type: DT_INT32 type: DT_INT64
+<a name="line-12118"></a>    }
+<a name="line-12119"></a>  }
+<a name="line-12120"></a>  name: "T"
+<a name="line-12121"></a>  type: "type"
+<a name="line-12122"></a>}
+<a name="line-12123"></a>input_arg {
+<a name="line-12124"></a>  description: "Original input for `fractional_max_pool`"
+<a name="line-12125"></a>  name: "orig_input"
+<a name="line-12126"></a>  type_attr: "T"
+<a name="line-12127"></a>}
+<a name="line-12128"></a>input_arg {
+<a name="line-12129"></a>  description: "Original output for `fractional_max_pool`"
+<a name="line-12130"></a>  name: "orig_output"
+<a name="line-12131"></a>  type_attr: "T"
+<a name="line-12132"></a>}
+<a name="line-12133"></a>input_arg {
+<a name="line-12134"></a>  description: "4-D with shape `[batch, height, width, channels]`.  Gradients\nw.r.t. the output of `fractional_max_pool`."
+<a name="line-12135"></a>  name: "out_backprop"
+<a name="line-12136"></a>  type_attr: "T"
+<a name="line-12137"></a>}
+<a name="line-12138"></a>input_arg {
+<a name="line-12139"></a>  description: "row pooling sequence, form pooling region with\ncol_pooling_sequence."
+<a name="line-12140"></a>  name: "row_pooling_sequence"
+<a name="line-12141"></a>  type: DT_INT64
+<a name="line-12142"></a>}
+<a name="line-12143"></a>input_arg {
+<a name="line-12144"></a>  description: "column pooling sequence, form pooling region with\nrow_pooling sequence."
+<a name="line-12145"></a>  name: "col_pooling_sequence"
+<a name="line-12146"></a>  type: DT_INT64
+<a name="line-12147"></a>}
+<a name="line-12148"></a>output_arg {
+<a name="line-12149"></a>  description: "4-D.  Gradients w.r.t. the input of `fractional_max_pool`."
+<a name="line-12150"></a>  name: "output"
+<a name="line-12151"></a>  type_attr: "T"
+<a name="line-12152"></a>}
+<a name="line-12153"></a>-}</span>
+<a name="line-12154"></a>
+<a name="line-12155"></a><a name="matchingFiles"></a><span class='hs-comment'>-- | Returns the set of files matching a pattern.</span>
+<a name="line-12156"></a><span class='hs-comment'>--</span>
+<a name="line-12157"></a><span class='hs-comment'>-- Note that this routine only supports wildcard characters in the</span>
+<a name="line-12158"></a><span class='hs-comment'>-- basename portion of the pattern, not in the directory portion.</span>
+<a name="line-12159"></a><span class='hs-definition'>matchingFiles</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __pattern__: A (scalar) shell wildcard pattern.</span>
+<a name="line-12160"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __filenames__: A vector of matching filenames.</span>
+<a name="line-12161"></a><span class='hs-definition'>matchingFiles</span> <span class='hs-varid'>pattern</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12162"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MatchingFiles"</span><span class='hs-layout'>)</span>
+<a name="line-12163"></a>        <span class='hs-varid'>pattern</span>
+<a name="line-12164"></a><span class='hs-comment'>{-
+<a name="line-12165"></a>input_arg {
+<a name="line-12166"></a>  description: "A (scalar) shell wildcard pattern."
+<a name="line-12167"></a>  name: "pattern"
+<a name="line-12168"></a>  type: DT_STRING
+<a name="line-12169"></a>}
+<a name="line-12170"></a>output_arg {
+<a name="line-12171"></a>  description: "A vector of matching filenames."
+<a name="line-12172"></a>  name: "filenames"
+<a name="line-12173"></a>  type: DT_STRING
+<a name="line-12174"></a>}
+<a name="line-12175"></a>-}</span>
+<a name="line-12176"></a>
+<a name="line-12177"></a><a name="tile"></a><span class='hs-comment'>-- | Constructs a tensor by tiling a given tensor.</span>
+<a name="line-12178"></a><span class='hs-comment'>--</span>
+<a name="line-12179"></a><span class='hs-comment'>-- This operation creates a new tensor by replicating `input` `multiples` times.</span>
+<a name="line-12180"></a><span class='hs-comment'>-- The output tensor's i'th dimension has `input.dims(i) * multiples[i]` elements,</span>
+<a name="line-12181"></a><span class='hs-comment'>-- and the values of `input` are replicated `multiples[i]` times along the 'i'th</span>
+<a name="line-12182"></a><span class='hs-comment'>-- dimension. For example, tiling `[a b c d]` by `[2]` produces</span>
+<a name="line-12183"></a><span class='hs-comment'>-- `[a b c d a b c d]`.</span>
+<a name="line-12184"></a><span class='hs-definition'>tile</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tmultiples</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tmultiples</span><span class='hs-layout'>,</span>
+<a name="line-12185"></a>                                     <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-12186"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tmultiples</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-12187"></a>        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: 1-D or higher.</span>
+<a name="line-12188"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tmultiples</span> <span class='hs-comment'>-- ^ __multiples__: 1-D. Length must be the same as the number of dimensions in `input`</span>
+<a name="line-12189"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-12190"></a><span class='hs-definition'>tile</span> <span class='hs-varid'>input</span> <span class='hs-varid'>multiples</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12191"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Tile"</span>
+<a name="line-12192"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-12193"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tmultiples"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tmultiples</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-12194"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>multiples</span>
+<a name="line-12195"></a><span class='hs-comment'>{-
+<a name="line-12196"></a>attr { name: "T" type: "type" }
+<a name="line-12197"></a>attr {
+<a name="line-12198"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-12199"></a>  default_value { type: DT_INT32 }
+<a name="line-12200"></a>  name: "Tmultiples"
+<a name="line-12201"></a>  type: "type"
+<a name="line-12202"></a>}
+<a name="line-12203"></a>input_arg {
+<a name="line-12204"></a>  description: "1-D or higher." name: "input" type_attr: "T"
+<a name="line-12205"></a>}
+<a name="line-12206"></a>input_arg {
+<a name="line-12207"></a>  description: "1-D. Length must be the same as the number of dimensions in `input`"
+<a name="line-12208"></a>  name: "multiples"
+<a name="line-12209"></a>  type_attr: "Tmultiples"
+<a name="line-12210"></a>}
+<a name="line-12211"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-12212"></a>-}</span>
+<a name="line-12213"></a>
+<a name="line-12214"></a><a name="sparseSparseMinimum"></a><span class='hs-comment'>-- | Returns the element-wise min of two SparseTensors.</span>
+<a name="line-12215"></a><span class='hs-comment'>--</span>
+<a name="line-12216"></a><span class='hs-comment'>-- Assumes the two SparseTensors have the same shape, i.e., no broadcasting.</span>
+<a name="line-12217"></a><span class='hs-definition'>sparseSparseMinimum</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-12218"></a>                                                     <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-12219"></a>                                                             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-12220"></a>                                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-12221"></a>                                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-12222"></a>                                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-12223"></a>                                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-12224"></a>                                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-12225"></a>                                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-12226"></a>                                                             <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-12227"></a>                                                             <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-12228"></a>                       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __a_indices__: 2-D.  `N x R` matrix with the indices of non-empty values in a</span>
+<a name="line-12229"></a>                                                <span class='hs-comment'>-- SparseTensor, in the canonical lexicographic ordering.</span>
+<a name="line-12230"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __a_values__: 1-D.  `N` non-empty values corresponding to `a_indices`.</span>
+<a name="line-12231"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __a_shape__: 1-D.  Shape of the input SparseTensor.</span>
+<a name="line-12232"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __b_indices__: counterpart to `a_indices` for the other operand.</span>
+<a name="line-12233"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __b_values__: counterpart to `a_values` for the other operand; must be of the same dtype.</span>
+<a name="line-12234"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __b_shape__: counterpart to `a_shape` for the other operand; the two shapes must be equal.</span>
+<a name="line-12235"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-12236"></a>                       <span class='hs-comment'>-- ^ (__output_indices__, __output_values__)</span>
+<a name="line-12237"></a>                       <span class='hs-comment'>--</span>
+<a name="line-12238"></a>                       <span class='hs-comment'>-- * __output_indices__: 2-D.  The indices of the output SparseTensor.</span>
+<a name="line-12239"></a>                       <span class='hs-comment'>--</span>
+<a name="line-12240"></a>                       <span class='hs-comment'>-- * __output_values__: 1-D.  The values of the output SparseTensor.</span>
+<a name="line-12241"></a><span class='hs-definition'>sparseSparseMinimum</span> <span class='hs-varid'>a_indices</span> <span class='hs-varid'>a_values</span> <span class='hs-varid'>a_shape</span> <span class='hs-varid'>b_indices</span> <span class='hs-varid'>b_values</span>
+<a name="line-12242"></a>                    <span class='hs-varid'>b_shape</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12243"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseSparseMinimum"</span>
+<a name="line-12244"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-12245"></a>        <span class='hs-varid'>a_indices</span> <span class='hs-varid'>a_values</span> <span class='hs-varid'>a_shape</span> <span class='hs-varid'>b_indices</span> <span class='hs-varid'>b_values</span> <span class='hs-varid'>b_shape</span>
+<a name="line-12246"></a><span class='hs-comment'>{-
+<a name="line-12247"></a>attr {
+<a name="line-12248"></a>  allowed_values {
+<a name="line-12249"></a>    list {
+<a name="line-12250"></a>      type: DT_FLOAT
+<a name="line-12251"></a>      type: DT_DOUBLE
+<a name="line-12252"></a>      type: DT_INT64
+<a name="line-12253"></a>      type: DT_INT32
+<a name="line-12254"></a>      type: DT_UINT8
+<a name="line-12255"></a>      type: DT_UINT16
+<a name="line-12256"></a>      type: DT_INT16
+<a name="line-12257"></a>      type: DT_INT8
+<a name="line-12258"></a>      type: DT_COMPLEX64
+<a name="line-12259"></a>      type: DT_COMPLEX128
+<a name="line-12260"></a>      type: DT_QINT8
+<a name="line-12261"></a>      type: DT_QUINT8
+<a name="line-12262"></a>      type: DT_QINT32
+<a name="line-12263"></a>      type: DT_HALF
+<a name="line-12264"></a>    }
+<a name="line-12265"></a>  }
+<a name="line-12266"></a>  name: "T"
+<a name="line-12267"></a>  type: "type"
+<a name="line-12268"></a>}
+<a name="line-12269"></a>input_arg {
+<a name="line-12270"></a>  description: "2-D.  `N x R` matrix with the indices of non-empty values in a\nSparseTensor, in the canonical lexicographic ordering."
+<a name="line-12271"></a>  name: "a_indices"
+<a name="line-12272"></a>  type: DT_INT64
+<a name="line-12273"></a>}
+<a name="line-12274"></a>input_arg {
+<a name="line-12275"></a>  description: "1-D.  `N` non-empty values corresponding to `a_indices`."
+<a name="line-12276"></a>  name: "a_values"
+<a name="line-12277"></a>  type_attr: "T"
+<a name="line-12278"></a>}
+<a name="line-12279"></a>input_arg {
+<a name="line-12280"></a>  description: "1-D.  Shape of the input SparseTensor."
+<a name="line-12281"></a>  name: "a_shape"
+<a name="line-12282"></a>  type: DT_INT64
+<a name="line-12283"></a>}
+<a name="line-12284"></a>input_arg {
+<a name="line-12285"></a>  description: "counterpart to `a_indices` for the other operand."
+<a name="line-12286"></a>  name: "b_indices"
+<a name="line-12287"></a>  type: DT_INT64
+<a name="line-12288"></a>}
+<a name="line-12289"></a>input_arg {
+<a name="line-12290"></a>  description: "counterpart to `a_values` for the other operand; must be of the same dtype."
+<a name="line-12291"></a>  name: "b_values"
+<a name="line-12292"></a>  type_attr: "T"
+<a name="line-12293"></a>}
+<a name="line-12294"></a>input_arg {
+<a name="line-12295"></a>  description: "counterpart to `a_shape` for the other operand; the two shapes must be equal."
+<a name="line-12296"></a>  name: "b_shape"
+<a name="line-12297"></a>  type: DT_INT64
+<a name="line-12298"></a>}
+<a name="line-12299"></a>output_arg {
+<a name="line-12300"></a>  description: "2-D.  The indices of the output SparseTensor."
+<a name="line-12301"></a>  name: "output_indices"
+<a name="line-12302"></a>  type: DT_INT64
+<a name="line-12303"></a>}
+<a name="line-12304"></a>output_arg {
+<a name="line-12305"></a>  description: "1-D.  The values of the output SparseTensor."
+<a name="line-12306"></a>  name: "output_values"
+<a name="line-12307"></a>  type_attr: "T"
+<a name="line-12308"></a>}
+<a name="line-12309"></a>-}</span>
+<a name="line-12310"></a>
+<a name="line-12311"></a><a name="allCandidateSampler"></a><span class='hs-comment'>-- | Generates labels for candidate sampling with a learned unigram distribution.</span>
+<a name="line-12312"></a><span class='hs-comment'>--</span>
+<a name="line-12313"></a><span class='hs-comment'>-- See explanations of candidate sampling and the data formats at</span>
+<a name="line-12314"></a><span class='hs-comment'>-- go/candidate-sampling.</span>
+<a name="line-12315"></a><span class='hs-comment'>-- </span>
+<a name="line-12316"></a><span class='hs-comment'>-- For each batch, this op picks a single set of sampled candidate labels.</span>
+<a name="line-12317"></a><span class='hs-comment'>-- </span>
+<a name="line-12318"></a><span class='hs-comment'>-- The advantages of sampling candidates per-batch are simplicity and the</span>
+<a name="line-12319"></a><span class='hs-comment'>-- possibility of efficient dense matrix multiplication. The disadvantage is that</span>
+<a name="line-12320"></a><span class='hs-comment'>-- the sampled candidates must be chosen independently of the context and of the</span>
+<a name="line-12321"></a><span class='hs-comment'>-- true labels.</span>
+<a name="line-12322"></a><span class='hs-definition'>allCandidateSampler</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_sampled__: Number of candidates to produce per batch.</span>
+<a name="line-12323"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_true__: Number of true labels per context.</span>
+<a name="line-12324"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __unique__: If unique is true, we sample with rejection, so that all sampled</span>
+<a name="line-12325"></a>                               <span class='hs-comment'>-- candidates in a batch are unique. This requires some approximation to</span>
+<a name="line-12326"></a>                               <span class='hs-comment'>-- estimate the post-rejection sampling probabilities.</span>
+<a name="line-12327"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __true_classes__: A batch_size * num_true matrix, in which each row contains the</span>
+<a name="line-12328"></a>                                                   <span class='hs-comment'>-- IDs of the num_true target_classes in the corresponding original label.</span>
+<a name="line-12329"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>,</span>
+<a name="line-12330"></a>                           <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span>
+<a name="line-12331"></a>                       <span class='hs-comment'>-- ^ (__sampled_candidates__, __true_expected_count__, __sampled_expected_count__)</span>
+<a name="line-12332"></a>                       <span class='hs-comment'>--</span>
+<a name="line-12333"></a>                       <span class='hs-comment'>-- * __sampled_candidates__: A vector of length num_sampled, in which each element is</span>
+<a name="line-12334"></a>                       <span class='hs-comment'>-- the ID of a sampled candidate.</span>
+<a name="line-12335"></a>                       <span class='hs-comment'>--</span>
+<a name="line-12336"></a>                       <span class='hs-comment'>-- * __true_expected_count__: A batch_size * num_true matrix, representing</span>
+<a name="line-12337"></a>                       <span class='hs-comment'>-- the number of times each candidate is expected to occur in a batch</span>
+<a name="line-12338"></a>                       <span class='hs-comment'>-- of sampled candidates. If unique=true, then this is a probability.</span>
+<a name="line-12339"></a>                       <span class='hs-comment'>--</span>
+<a name="line-12340"></a>                       <span class='hs-comment'>-- * __sampled_expected_count__: A vector of length num_sampled, for each sampled</span>
+<a name="line-12341"></a>                       <span class='hs-comment'>-- candidate representing the number of times the candidate is expected</span>
+<a name="line-12342"></a>                       <span class='hs-comment'>-- to occur in a batch of sampled candidates.  If unique=true, then this is a</span>
+<a name="line-12343"></a>                       <span class='hs-comment'>-- probability.</span>
+<a name="line-12344"></a><span class='hs-definition'>allCandidateSampler</span> <span class='hs-varid'>num_sampled</span> <span class='hs-varid'>num_true</span> <span class='hs-varid'>unique</span>
+<a name="line-12345"></a>                    <span class='hs-varid'>true_classes</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12346"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"AllCandidateSampler"</span>
+<a name="line-12347"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_sampled"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_sampled</span>
+<a name="line-12348"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_true"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_true</span>
+<a name="line-12349"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"unique"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>unique</span><span class='hs-layout'>)</span>
+<a name="line-12350"></a>        <span class='hs-varid'>true_classes</span>
+<a name="line-12351"></a><span class='hs-comment'>{-
+<a name="line-12352"></a>attr {
+<a name="line-12353"></a>  description: "Number of true labels per context."
+<a name="line-12354"></a>  has_minimum: true
+<a name="line-12355"></a>  minimum: 1
+<a name="line-12356"></a>  name: "num_true"
+<a name="line-12357"></a>  type: "int"
+<a name="line-12358"></a>}
+<a name="line-12359"></a>attr {
+<a name="line-12360"></a>  description: "Number of candidates to produce per batch."
+<a name="line-12361"></a>  has_minimum: true
+<a name="line-12362"></a>  minimum: 1
+<a name="line-12363"></a>  name: "num_sampled"
+<a name="line-12364"></a>  type: "int"
+<a name="line-12365"></a>}
+<a name="line-12366"></a>attr {
+<a name="line-12367"></a>  description: "If unique is true, we sample with rejection, so that all sampled\ncandidates in a batch are unique. This requires some approximation to\nestimate the post-rejection sampling probabilities."
+<a name="line-12368"></a>  name: "unique"
+<a name="line-12369"></a>  type: "bool"
+<a name="line-12370"></a>}
+<a name="line-12371"></a>attr {
+<a name="line-12372"></a>  default_value { i: 0 }
+<a name="line-12373"></a>  description: "If either seed or seed2 are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-12374"></a>  name: "seed"
+<a name="line-12375"></a>  type: "int"
+<a name="line-12376"></a>}
+<a name="line-12377"></a>attr {
+<a name="line-12378"></a>  default_value { i: 0 }
+<a name="line-12379"></a>  description: "An second seed to avoid seed collision."
+<a name="line-12380"></a>  name: "seed2"
+<a name="line-12381"></a>  type: "int"
+<a name="line-12382"></a>}
+<a name="line-12383"></a>input_arg {
+<a name="line-12384"></a>  description: "A batch_size * num_true matrix, in which each row contains the\nIDs of the num_true target_classes in the corresponding original label."
+<a name="line-12385"></a>  name: "true_classes"
+<a name="line-12386"></a>  type: DT_INT64
+<a name="line-12387"></a>}
+<a name="line-12388"></a>output_arg {
+<a name="line-12389"></a>  description: "A vector of length num_sampled, in which each element is\nthe ID of a sampled candidate."
+<a name="line-12390"></a>  name: "sampled_candidates"
+<a name="line-12391"></a>  type: DT_INT64
+<a name="line-12392"></a>}
+<a name="line-12393"></a>output_arg {
+<a name="line-12394"></a>  description: "A batch_size * num_true matrix, representing\nthe number of times each candidate is expected to occur in a batch\nof sampled candidates. If unique=true, then this is a probability."
+<a name="line-12395"></a>  name: "true_expected_count"
+<a name="line-12396"></a>  type: DT_FLOAT
+<a name="line-12397"></a>}
+<a name="line-12398"></a>output_arg {
+<a name="line-12399"></a>  description: "A vector of length num_sampled, for each sampled\ncandidate representing the number of times the candidate is expected\nto occur in a batch of sampled candidates.  If unique=true, then this is a\nprobability."
+<a name="line-12400"></a>  name: "sampled_expected_count"
+<a name="line-12401"></a>  type: DT_FLOAT
+<a name="line-12402"></a>}
+<a name="line-12403"></a>-}</span>
+<a name="line-12404"></a>
+<a name="line-12405"></a><a name="refSwitch"></a><span class='hs-comment'>-- | Forwards the ref tensor `data` to the output port determined by `pred`.</span>
+<a name="line-12406"></a><span class='hs-comment'>--</span>
+<a name="line-12407"></a><span class='hs-comment'>-- If `pred` is true, the `data` input is forwarded to `output_true`. Otherwise,</span>
+<a name="line-12408"></a><span class='hs-comment'>-- the data goes to `output_false`.</span>
+<a name="line-12409"></a><span class='hs-comment'>-- </span>
+<a name="line-12410"></a><span class='hs-comment'>-- See also `Switch` and `Merge`.</span>
+<a name="line-12411"></a><span class='hs-definition'>refSwitch</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-12412"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__: The ref tensor to be forwarded to the appropriate output.</span>
+<a name="line-12413"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __pred__: A scalar that specifies which output port will receive data.</span>
+<a name="line-12414"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-12415"></a>             <span class='hs-comment'>-- ^ (__output_false__, __output_true__)</span>
+<a name="line-12416"></a>             <span class='hs-comment'>--</span>
+<a name="line-12417"></a>             <span class='hs-comment'>-- * __output_false__: If `pred` is false, data will be forwarded to this output.</span>
+<a name="line-12418"></a>             <span class='hs-comment'>--</span>
+<a name="line-12419"></a>             <span class='hs-comment'>-- * __output_true__: If `pred` is true, data will be forwarded to this output.</span>
+<a name="line-12420"></a><span class='hs-definition'>refSwitch</span> <span class='hs-varid'>data'</span> <span class='hs-varid'>pred</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12421"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RefSwitch"</span>
+<a name="line-12422"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-12423"></a>        <span class='hs-varid'>data'</span> <span class='hs-varid'>pred</span>
+<a name="line-12424"></a><span class='hs-comment'>{-
+<a name="line-12425"></a>attr { name: "T" type: "type" }
+<a name="line-12426"></a>input_arg {
+<a name="line-12427"></a>  description: "The ref tensor to be forwarded to the appropriate output."
+<a name="line-12428"></a>  is_ref: true
+<a name="line-12429"></a>  name: "data"
+<a name="line-12430"></a>  type_attr: "T"
+<a name="line-12431"></a>}
+<a name="line-12432"></a>input_arg {
+<a name="line-12433"></a>  description: "A scalar that specifies which output port will receive data."
+<a name="line-12434"></a>  name: "pred"
+<a name="line-12435"></a>  type: DT_BOOL
+<a name="line-12436"></a>}
+<a name="line-12437"></a>output_arg {
+<a name="line-12438"></a>  description: "If `pred` is false, data will be forwarded to this output."
+<a name="line-12439"></a>  is_ref: true
+<a name="line-12440"></a>  name: "output_false"
+<a name="line-12441"></a>  type_attr: "T"
+<a name="line-12442"></a>}
+<a name="line-12443"></a>output_arg {
+<a name="line-12444"></a>  description: "If `pred` is true, data will be forwarded to this output."
+<a name="line-12445"></a>  is_ref: true
+<a name="line-12446"></a>  name: "output_true"
+<a name="line-12447"></a>  type_attr: "T"
+<a name="line-12448"></a>}
+<a name="line-12449"></a>-}</span>
+<a name="line-12450"></a>
+<a name="line-12451"></a><a name="mergeSummary"></a><span class='hs-comment'>-- | Merges summaries.</span>
+<a name="line-12452"></a><span class='hs-comment'>--</span>
+<a name="line-12453"></a><span class='hs-comment'>-- This op creates a</span>
+<a name="line-12454"></a><span class='hs-comment'>-- [`Summary`](https://www.tensorflow.org/code/tensorflow/core/framework/summary.proto)</span>
+<a name="line-12455"></a><span class='hs-comment'>-- protocol buffer that contains the union of all the values in the input</span>
+<a name="line-12456"></a><span class='hs-comment'>-- summaries.</span>
+<a name="line-12457"></a><span class='hs-comment'>-- </span>
+<a name="line-12458"></a><span class='hs-comment'>-- When the Op is run, it reports an `InvalidArgument` error if multiple values</span>
+<a name="line-12459"></a><span class='hs-comment'>-- in the summaries to merge use the same tag.</span>
+<a name="line-12460"></a><span class='hs-definition'>mergeSummary</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __inputs__: Can be of any shape.  Each must contain serialized `Summary` protocol</span>
+<a name="line-12461"></a>                                                       <span class='hs-comment'>-- buffers.</span>
+<a name="line-12462"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __summary__: Scalar. Serialized `Summary` protocol buffer.</span>
+<a name="line-12463"></a><span class='hs-definition'>mergeSummary</span> <span class='hs-varid'>inputs</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"N"</span><span class='hs-layout'>,</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"inputs"</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>inputs</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12464"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MergeSummary"</span>
+<a name="line-12465"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"N"</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>inputs</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-12466"></a>        <span class='hs-varid'>inputs</span>
+<a name="line-12467"></a><span class='hs-comment'>{-
+<a name="line-12468"></a>attr { has_minimum: true minimum: 1 name: "N" type: "int" }
+<a name="line-12469"></a>input_arg {
+<a name="line-12470"></a>  description: "Can be of any shape.  Each must contain serialized `Summary` protocol\nbuffers."
+<a name="line-12471"></a>  name: "inputs"
+<a name="line-12472"></a>  number_attr: "N"
+<a name="line-12473"></a>  type: DT_STRING
+<a name="line-12474"></a>}
+<a name="line-12475"></a>output_arg {
+<a name="line-12476"></a>  description: "Scalar. Serialized `Summary` protocol buffer."
+<a name="line-12477"></a>  name: "summary"
+<a name="line-12478"></a>  type: DT_STRING
+<a name="line-12479"></a>}
+<a name="line-12480"></a>-}</span>
+<a name="line-12481"></a>
+<a name="line-12482"></a><span class='hs-comment'>-- | Returns the truth value of NOT x element-wise.</span>
+<a name="line-12483"></a>
+<a name="line-12484"></a><a name="logicalNot"></a><span class='hs-definition'>logicalNot</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-12485"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-12486"></a><span class='hs-definition'>logicalNot</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12487"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"LogicalNot"</span><span class='hs-layout'>)</span>
+<a name="line-12488"></a>        <span class='hs-varid'>x</span>
+<a name="line-12489"></a><span class='hs-comment'>{-
+<a name="line-12490"></a>input_arg { name: "x" type: DT_BOOL }
+<a name="line-12491"></a>output_arg { name: "y" type: DT_BOOL }
+<a name="line-12492"></a>-}</span>
+<a name="line-12493"></a>
+<a name="line-12494"></a><span class='hs-comment'>-- | Gradients for Local Response Normalization.</span>
+<a name="line-12495"></a>
+<a name="line-12496"></a><a name="lRNGrad"></a><span class='hs-definition'>lRNGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-12497"></a>                                                      <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-12498"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input_grads__: 4-D with shape `[batch, height, width, channels]`.</span>
+<a name="line-12499"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input_image__: 4-D with shape `[batch, height, width, channels]`.</span>
+<a name="line-12500"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output_image__: 4-D with shape `[batch, height, width, channels]`.</span>
+<a name="line-12501"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The gradients for LRN.</span>
+<a name="line-12502"></a><span class='hs-definition'>lRNGrad</span> <span class='hs-varid'>input_grads</span> <span class='hs-varid'>input_image</span> <span class='hs-varid'>output_image</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12503"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"LRNGrad"</span>
+<a name="line-12504"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-12505"></a>        <span class='hs-varid'>input_grads</span> <span class='hs-varid'>input_image</span> <span class='hs-varid'>output_image</span>
+<a name="line-12506"></a><span class='hs-comment'>{-
+<a name="line-12507"></a>attr {
+<a name="line-12508"></a>  default_value { i: 5 }
+<a name="line-12509"></a>  description: "A depth radius."
+<a name="line-12510"></a>  name: "depth_radius"
+<a name="line-12511"></a>  type: "int"
+<a name="line-12512"></a>}
+<a name="line-12513"></a>attr {
+<a name="line-12514"></a>  default_value { f: 1.0 }
+<a name="line-12515"></a>  description: "An offset (usually &gt; 0 to avoid dividing by 0)."
+<a name="line-12516"></a>  name: "bias"
+<a name="line-12517"></a>  type: "float"
+<a name="line-12518"></a>}
+<a name="line-12519"></a>attr {
+<a name="line-12520"></a>  default_value { f: 1.0 }
+<a name="line-12521"></a>  description: "A scale factor, usually positive."
+<a name="line-12522"></a>  name: "alpha"
+<a name="line-12523"></a>  type: "float"
+<a name="line-12524"></a>}
+<a name="line-12525"></a>attr {
+<a name="line-12526"></a>  default_value { f: 0.5 }
+<a name="line-12527"></a>  description: "An exponent."
+<a name="line-12528"></a>  name: "beta"
+<a name="line-12529"></a>  type: "float"
+<a name="line-12530"></a>}
+<a name="line-12531"></a>attr {
+<a name="line-12532"></a>  allowed_values { list { type: DT_FLOAT type: DT_HALF } }
+<a name="line-12533"></a>  default_value { type: DT_FLOAT }
+<a name="line-12534"></a>  name: "T"
+<a name="line-12535"></a>  type: "type"
+<a name="line-12536"></a>}
+<a name="line-12537"></a>input_arg {
+<a name="line-12538"></a>  description: "4-D with shape `[batch, height, width, channels]`."
+<a name="line-12539"></a>  name: "input_grads"
+<a name="line-12540"></a>  type_attr: "T"
+<a name="line-12541"></a>}
+<a name="line-12542"></a>input_arg {
+<a name="line-12543"></a>  description: "4-D with shape `[batch, height, width, channels]`."
+<a name="line-12544"></a>  name: "input_image"
+<a name="line-12545"></a>  type_attr: "T"
+<a name="line-12546"></a>}
+<a name="line-12547"></a>input_arg {
+<a name="line-12548"></a>  description: "4-D with shape `[batch, height, width, channels]`."
+<a name="line-12549"></a>  name: "output_image"
+<a name="line-12550"></a>  type_attr: "T"
+<a name="line-12551"></a>}
+<a name="line-12552"></a>output_arg {
+<a name="line-12553"></a>  description: "The gradients for LRN." name: "output" type_attr: "T"
+<a name="line-12554"></a>}
+<a name="line-12555"></a>-}</span>
+<a name="line-12556"></a>
+<a name="line-12557"></a><a name="stringToNumber"></a><span class='hs-comment'>-- | Converts each string in the input Tensor to the specified numeric type.</span>
+<a name="line-12558"></a><span class='hs-comment'>--</span>
+<a name="line-12559"></a><span class='hs-comment'>-- (Note that int32 overflow results in an error while float overflow</span>
+<a name="line-12560"></a><span class='hs-comment'>-- results in a rounded value.)</span>
+<a name="line-12561"></a><span class='hs-definition'>stringToNumber</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>out_type</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>,</span>
+<a name="line-12562"></a>                                        <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-12563"></a>                                                <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-12564"></a>                  <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __string_tensor__</span>
+<a name="line-12565"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>out_type</span> <span class='hs-comment'>-- ^ __output__: A Tensor of the same shape as the input `string_tensor`.</span>
+<a name="line-12566"></a><span class='hs-definition'>stringToNumber</span> <span class='hs-varid'>string_tensor</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12567"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"StringToNumber"</span>
+<a name="line-12568"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"out_type"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-12569"></a>        <span class='hs-varid'>string_tensor</span>
+<a name="line-12570"></a><span class='hs-comment'>{-
+<a name="line-12571"></a>attr {
+<a name="line-12572"></a>  allowed_values { list { type: DT_FLOAT type: DT_INT32 } }
+<a name="line-12573"></a>  default_value { type: DT_FLOAT }
+<a name="line-12574"></a>  description: "The numeric type to interpret each string in string_tensor as."
+<a name="line-12575"></a>  name: "out_type"
+<a name="line-12576"></a>  type: "type"
+<a name="line-12577"></a>}
+<a name="line-12578"></a>input_arg { name: "string_tensor" type: DT_STRING }
+<a name="line-12579"></a>output_arg {
+<a name="line-12580"></a>  description: "A Tensor of the same shape as the input `string_tensor`."
+<a name="line-12581"></a>  name: "output"
+<a name="line-12582"></a>  type_attr: "out_type"
+<a name="line-12583"></a>}
+<a name="line-12584"></a>-}</span>
+<a name="line-12585"></a>
+<a name="line-12586"></a><a name="sparseMatMul"></a><span class='hs-comment'>-- | Multiply matrix "a" by matrix "b".</span>
+<a name="line-12587"></a><span class='hs-comment'>--</span>
+<a name="line-12588"></a><span class='hs-comment'>-- The inputs must be two-dimensional matrices and the inner dimension of "a" must</span>
+<a name="line-12589"></a><span class='hs-comment'>-- match the outer dimension of "b". This op is optimized for the case where at</span>
+<a name="line-12590"></a><span class='hs-comment'>-- least one of "a" or "b" is sparse. The breakeven for using this versus a dense</span>
+<a name="line-12591"></a><span class='hs-comment'>-- matrix multiply on one platform was 30% zero values in the sparse matrix.</span>
+<a name="line-12592"></a><span class='hs-definition'>sparseMatMul</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>ta</span> <span class='hs-varid'>tb</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>ta</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-12593"></a>                                                             <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>ta</span><span class='hs-layout'>,</span>
+<a name="line-12594"></a>                                      <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tb</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-12595"></a>                                                             <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tb</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-12596"></a>                <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>ta</span> <span class='hs-comment'>-- ^ __a__</span>
+<a name="line-12597"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tb</span> <span class='hs-comment'>-- ^ __b__</span>
+<a name="line-12598"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __product__</span>
+<a name="line-12599"></a><span class='hs-definition'>sparseMatMul</span> <span class='hs-varid'>a</span> <span class='hs-varid'>b</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12600"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseMatMul"</span>
+<a name="line-12601"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Ta"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>ta</span><span class='hs-layout'>)</span>
+<a name="line-12602"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tb"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tb</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-12603"></a>        <span class='hs-varid'>a</span> <span class='hs-varid'>b</span>
+<a name="line-12604"></a><span class='hs-comment'>{-
+<a name="line-12605"></a>attr {
+<a name="line-12606"></a>  default_value { b: false } name: "transpose_a" type: "bool"
+<a name="line-12607"></a>}
+<a name="line-12608"></a>attr {
+<a name="line-12609"></a>  default_value { b: false } name: "transpose_b" type: "bool"
+<a name="line-12610"></a>}
+<a name="line-12611"></a>attr {
+<a name="line-12612"></a>  default_value { b: false } name: "a_is_sparse" type: "bool"
+<a name="line-12613"></a>}
+<a name="line-12614"></a>attr {
+<a name="line-12615"></a>  default_value { b: false } name: "b_is_sparse" type: "bool"
+<a name="line-12616"></a>}
+<a name="line-12617"></a>attr {
+<a name="line-12618"></a>  allowed_values { list { type: DT_FLOAT type: DT_BFLOAT16 } }
+<a name="line-12619"></a>  default_value { type: DT_FLOAT }
+<a name="line-12620"></a>  name: "Ta"
+<a name="line-12621"></a>  type: "type"
+<a name="line-12622"></a>}
+<a name="line-12623"></a>attr {
+<a name="line-12624"></a>  allowed_values { list { type: DT_FLOAT type: DT_BFLOAT16 } }
+<a name="line-12625"></a>  default_value { type: DT_FLOAT }
+<a name="line-12626"></a>  name: "Tb"
+<a name="line-12627"></a>  type: "type"
+<a name="line-12628"></a>}
+<a name="line-12629"></a>input_arg { name: "a" type_attr: "Ta" }
+<a name="line-12630"></a>input_arg { name: "b" type_attr: "Tb" }
+<a name="line-12631"></a>output_arg { name: "product" type: DT_FLOAT }
+<a name="line-12632"></a>-}</span>
+<a name="line-12633"></a>
+<a name="line-12634"></a><a name="merge"></a><span class='hs-comment'>-- | Forwards the value of an available tensor from `inputs` to `output`.</span>
+<a name="line-12635"></a><span class='hs-comment'>--</span>
+<a name="line-12636"></a><span class='hs-comment'>-- `Merge` waits for at least one of the tensors in `inputs` to become available.</span>
+<a name="line-12637"></a><span class='hs-comment'>-- It is usually combined with `Switch` to implement branching.</span>
+<a name="line-12638"></a><span class='hs-comment'>-- </span>
+<a name="line-12639"></a><span class='hs-comment'>-- `Merge` forwards the first tensor for become available to `output`, and sets</span>
+<a name="line-12640"></a><span class='hs-comment'>-- `value_index` to its index in `inputs`.</span>
+<a name="line-12641"></a><span class='hs-definition'>merge</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-12642"></a>         <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __inputs__: The input tensors, exactly one of which will become available.</span>
+<a name="line-12643"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-12644"></a>         <span class='hs-comment'>-- ^ (__output__, __value_index__)</span>
+<a name="line-12645"></a>         <span class='hs-comment'>--</span>
+<a name="line-12646"></a>         <span class='hs-comment'>-- * __output__: Will be set to the available input tensor.</span>
+<a name="line-12647"></a>         <span class='hs-comment'>--</span>
+<a name="line-12648"></a>         <span class='hs-comment'>-- * __value_index__: The index of the chosen input tensor in `inputs`.</span>
+<a name="line-12649"></a><span class='hs-definition'>merge</span> <span class='hs-varid'>inputs</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"N"</span><span class='hs-layout'>,</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"inputs"</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>inputs</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12650"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Merge"</span>
+<a name="line-12651"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-12652"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"N"</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>inputs</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-12653"></a>        <span class='hs-varid'>inputs</span>
+<a name="line-12654"></a><span class='hs-comment'>{-
+<a name="line-12655"></a>attr { name: "T" type: "type" }
+<a name="line-12656"></a>attr { has_minimum: true minimum: 1 name: "N" type: "int" }
+<a name="line-12657"></a>input_arg {
+<a name="line-12658"></a>  description: "The input tensors, exactly one of which will become available."
+<a name="line-12659"></a>  name: "inputs"
+<a name="line-12660"></a>  number_attr: "N"
+<a name="line-12661"></a>  type_attr: "T"
+<a name="line-12662"></a>}
+<a name="line-12663"></a>output_arg {
+<a name="line-12664"></a>  description: "Will be set to the available input tensor."
+<a name="line-12665"></a>  name: "output"
+<a name="line-12666"></a>  type_attr: "T"
+<a name="line-12667"></a>}
+<a name="line-12668"></a>output_arg {
+<a name="line-12669"></a>  description: "The index of the chosen input tensor in `inputs`."
+<a name="line-12670"></a>  name: "value_index"
+<a name="line-12671"></a>  type: DT_INT32
+<a name="line-12672"></a>}
+<a name="line-12673"></a>-}</span>
+<a name="line-12674"></a>
+<a name="line-12675"></a><a name="choleskyGrad"></a><span class='hs-comment'>-- | Computes the reverse mode backpropagated gradient of the Cholesky algorithm.</span>
+<a name="line-12676"></a><span class='hs-comment'>--</span>
+<a name="line-12677"></a><span class='hs-comment'>-- For an explanation see "Differentiation of the Cholesky algorithm" by</span>
+<a name="line-12678"></a><span class='hs-comment'>-- Iain Murray <a href="http://arxiv.org/abs/1602.07527.">http://arxiv.org/abs/1602.07527.</a></span>
+<a name="line-12679"></a><span class='hs-definition'>choleskyGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-12680"></a>                <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l__: Output of batch Cholesky algorithm l = cholesky(A). Shape is `[..., M, M]`.</span>
+<a name="line-12681"></a>                            <span class='hs-comment'>-- Algorithm depends only on lower triangular part of the innermost matrices of</span>
+<a name="line-12682"></a>                            <span class='hs-comment'>-- this tensor.</span>
+<a name="line-12683"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: df/dl where f is some scalar function. Shape is `[..., M, M]`.</span>
+<a name="line-12684"></a>                               <span class='hs-comment'>-- Algorithm depends only on lower triangular part of the innermost matrices of</span>
+<a name="line-12685"></a>                               <span class='hs-comment'>-- this tensor.</span>
+<a name="line-12686"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Symmetrized version of df/dA . Shape is `[..., M, M]`</span>
+<a name="line-12687"></a><span class='hs-definition'>choleskyGrad</span> <span class='hs-varid'>l</span> <span class='hs-varid'>grad</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12688"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"CholeskyGrad"</span>
+<a name="line-12689"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-12690"></a>        <span class='hs-varid'>l</span> <span class='hs-varid'>grad</span>
+<a name="line-12691"></a><span class='hs-comment'>{-
+<a name="line-12692"></a>attr {
+<a name="line-12693"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-12694"></a>  name: "T"
+<a name="line-12695"></a>  type: "type"
+<a name="line-12696"></a>}
+<a name="line-12697"></a>input_arg {
+<a name="line-12698"></a>  description: "Output of batch Cholesky algorithm l = cholesky(A). Shape is `[..., M, M]`.\nAlgorithm depends only on lower triangular part of the innermost matrices of\nthis tensor."
+<a name="line-12699"></a>  name: "l"
+<a name="line-12700"></a>  type_attr: "T"
+<a name="line-12701"></a>}
+<a name="line-12702"></a>input_arg {
+<a name="line-12703"></a>  description: "df/dl where f is some scalar function. Shape is `[..., M, M]`.\nAlgorithm depends only on lower triangular part of the innermost matrices of\nthis tensor."
+<a name="line-12704"></a>  name: "grad"
+<a name="line-12705"></a>  type_attr: "T"
+<a name="line-12706"></a>}
+<a name="line-12707"></a>output_arg {
+<a name="line-12708"></a>  description: "Symmetrized version of df/dA . Shape is `[..., M, M]`"
+<a name="line-12709"></a>  name: "output"
+<a name="line-12710"></a>  type_attr: "T"
+<a name="line-12711"></a>}
+<a name="line-12712"></a>-}</span>
+<a name="line-12713"></a>
+<a name="line-12714"></a><span class='hs-comment'>-- | </span>
+<a name="line-12715"></a>
+<a name="line-12716"></a><a name="batchCholeskyGrad"></a><span class='hs-definition'>batchCholeskyGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-12717"></a>                                                             <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-12718"></a>                     <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l__</span>
+<a name="line-12719"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__</span>
+<a name="line-12720"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-12721"></a><span class='hs-definition'>batchCholeskyGrad</span> <span class='hs-varid'>l</span> <span class='hs-varid'>grad</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12722"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchCholeskyGrad"</span>
+<a name="line-12723"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-12724"></a>        <span class='hs-varid'>l</span> <span class='hs-varid'>grad</span>
+<a name="line-12725"></a><span class='hs-comment'>{-
+<a name="line-12726"></a>attr {
+<a name="line-12727"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-12728"></a>  name: "T"
+<a name="line-12729"></a>  type: "type"
+<a name="line-12730"></a>}
+<a name="line-12731"></a>input_arg { name: "l" type_attr: "T" }
+<a name="line-12732"></a>input_arg { name: "grad" type_attr: "T" }
+<a name="line-12733"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-12734"></a>-}</span>
+<a name="line-12735"></a>
+<a name="line-12736"></a><a name="tensorArrayGather"></a><span class='hs-comment'>-- | Gather specific elements from the TensorArray into output `value`.</span>
+<a name="line-12737"></a><span class='hs-comment'>--</span>
+<a name="line-12738"></a><span class='hs-comment'>-- All elements selected by `indices` must have the same shape.</span>
+<a name="line-12739"></a><span class='hs-definition'>tensorArrayGather</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-12740"></a>                     <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a TensorArray.</span>
+<a name="line-12741"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __indices__: The locations in the TensorArray from which to read tensor elements.</span>
+<a name="line-12742"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __flow_in__: A float scalar that enforces proper chaining of operations.</span>
+<a name="line-12743"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __value__: All of the elements in the TensorArray, concatenated along a new</span>
+<a name="line-12744"></a>                     <span class='hs-comment'>-- axis (the new dimension 0).</span>
+<a name="line-12745"></a><span class='hs-definition'>tensorArrayGather</span> <span class='hs-varid'>handle</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>flow_in</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12746"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TensorArrayGather"</span>
+<a name="line-12747"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-12748"></a>        <span class='hs-varid'>handle</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>flow_in</span>
+<a name="line-12749"></a><span class='hs-comment'>{-
+<a name="line-12750"></a>attr {
+<a name="line-12751"></a>  description: "The type of the elem that is returned."
+<a name="line-12752"></a>  name: "dtype"
+<a name="line-12753"></a>  type: "type"
+<a name="line-12754"></a>}
+<a name="line-12755"></a>attr {
+<a name="line-12756"></a>  default_value { shape { unknown_rank: true } }
+<a name="line-12757"></a>  description: "The expected shape of an element, if known. Used to\nvalidate the shapes of TensorArray elements. If this shape is not\nfully specified, gathering zero-size TensorArrays is an error."
+<a name="line-12758"></a>  name: "element_shape"
+<a name="line-12759"></a>  type: "shape"
+<a name="line-12760"></a>}
+<a name="line-12761"></a>input_arg {
+<a name="line-12762"></a>  description: "The handle to a TensorArray."
+<a name="line-12763"></a>  is_ref: true
+<a name="line-12764"></a>  name: "handle"
+<a name="line-12765"></a>  type: DT_STRING
+<a name="line-12766"></a>}
+<a name="line-12767"></a>input_arg {
+<a name="line-12768"></a>  description: "The locations in the TensorArray from which to read tensor elements."
+<a name="line-12769"></a>  name: "indices"
+<a name="line-12770"></a>  type: DT_INT32
+<a name="line-12771"></a>}
+<a name="line-12772"></a>input_arg {
+<a name="line-12773"></a>  description: "A float scalar that enforces proper chaining of operations."
+<a name="line-12774"></a>  name: "flow_in"
+<a name="line-12775"></a>  type: DT_FLOAT
+<a name="line-12776"></a>}
+<a name="line-12777"></a>output_arg {
+<a name="line-12778"></a>  description: "All of the elements in the TensorArray, concatenated along a new\naxis (the new dimension 0)."
+<a name="line-12779"></a>  name: "value"
+<a name="line-12780"></a>  type_attr: "dtype"
+<a name="line-12781"></a>}
+<a name="line-12782"></a>-}</span>
+<a name="line-12783"></a>
+<a name="line-12784"></a><span class='hs-comment'>-- | Resize `images` to `size` using nearest neighbor interpolation.</span>
+<a name="line-12785"></a>
+<a name="line-12786"></a><a name="resizeNearestNeighbor"></a><span class='hs-definition'>resizeNearestNeighbor</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-12787"></a>                                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-12788"></a>                                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-12789"></a>                                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-12790"></a>                                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-12791"></a>                                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-12792"></a>                                                                 <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-12793"></a>                                                                 <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-12794"></a>                         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __images__: 4-D with shape `[batch, height, width, channels]`.</span>
+<a name="line-12795"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __size__: = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The</span>
+<a name="line-12796"></a>                                                     <span class='hs-comment'>-- new size for the images.</span>
+<a name="line-12797"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __resized_images__: 4-D with shape</span>
+<a name="line-12798"></a>                         <span class='hs-comment'>-- `[batch, new_height, new_width, channels]`.</span>
+<a name="line-12799"></a><span class='hs-definition'>resizeNearestNeighbor</span> <span class='hs-varid'>images</span> <span class='hs-varid'>size</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12800"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ResizeNearestNeighbor"</span>
+<a name="line-12801"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-12802"></a>        <span class='hs-varid'>images</span> <span class='hs-varid'>size</span>
+<a name="line-12803"></a><span class='hs-comment'>{-
+<a name="line-12804"></a>attr {
+<a name="line-12805"></a>  allowed_values {
+<a name="line-12806"></a>    list {
+<a name="line-12807"></a>      type: DT_UINT8
+<a name="line-12808"></a>      type: DT_INT8
+<a name="line-12809"></a>      type: DT_INT16
+<a name="line-12810"></a>      type: DT_INT32
+<a name="line-12811"></a>      type: DT_INT64
+<a name="line-12812"></a>      type: DT_HALF
+<a name="line-12813"></a>      type: DT_FLOAT
+<a name="line-12814"></a>      type: DT_DOUBLE
+<a name="line-12815"></a>    }
+<a name="line-12816"></a>  }
+<a name="line-12817"></a>  name: "T"
+<a name="line-12818"></a>  type: "type"
+<a name="line-12819"></a>}
+<a name="line-12820"></a>attr {
+<a name="line-12821"></a>  default_value { b: false }
+<a name="line-12822"></a>  description: "If true, rescale input by (new_height - 1) / (height - 1), which\nexactly aligns the 4 corners of images and resized images. If false, rescale\nby new_height / height. Treat similarly the width dimension."
+<a name="line-12823"></a>  name: "align_corners"
+<a name="line-12824"></a>  type: "bool"
+<a name="line-12825"></a>}
+<a name="line-12826"></a>input_arg {
+<a name="line-12827"></a>  description: "4-D with shape `[batch, height, width, channels]`."
+<a name="line-12828"></a>  name: "images"
+<a name="line-12829"></a>  type_attr: "T"
+<a name="line-12830"></a>}
+<a name="line-12831"></a>input_arg {
+<a name="line-12832"></a>  description: "= A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The\nnew size for the images."
+<a name="line-12833"></a>  name: "size"
+<a name="line-12834"></a>  type: DT_INT32
+<a name="line-12835"></a>}
+<a name="line-12836"></a>output_arg {
+<a name="line-12837"></a>  description: "4-D with shape\n`[batch, new_height, new_width, channels]`."
+<a name="line-12838"></a>  name: "resized_images"
+<a name="line-12839"></a>  type_attr: "T"
+<a name="line-12840"></a>}
+<a name="line-12841"></a>-}</span>
+<a name="line-12842"></a>
+<a name="line-12843"></a><span class='hs-comment'>-- | Training via negative sampling.</span>
+<a name="line-12844"></a>
+<a name="line-12845"></a><a name="negTrain"></a><span class='hs-definition'>negTrain</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_negative_samples__: Number of negative samples per example.</span>
+<a name="line-12846"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __w_in__: input word embedding.</span>
+<a name="line-12847"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __w_out__: output word embedding.</span>
+<a name="line-12848"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __examples__: A vector of word ids.</span>
+<a name="line-12849"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __labels__: A vector of word ids.</span>
+<a name="line-12850"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __lr__</span>
+<a name="line-12851"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-12852"></a><span class='hs-definition'>negTrain</span> <span class='hs-varid'>num_negative_samples</span> <span class='hs-varid'>w_in</span> <span class='hs-varid'>w_out</span> <span class='hs-varid'>examples</span> <span class='hs-varid'>labels</span> <span class='hs-varid'>lr</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12853"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"NegTrain"</span>
+<a name="line-12854"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_negative_samples"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_negative_samples</span><span class='hs-layout'>)</span>
+<a name="line-12855"></a>        <span class='hs-varid'>w_in</span> <span class='hs-varid'>w_out</span> <span class='hs-varid'>examples</span> <span class='hs-varid'>labels</span> <span class='hs-varid'>lr</span>
+<a name="line-12856"></a><span class='hs-comment'>{-
+<a name="line-12857"></a>attr {
+<a name="line-12858"></a>  description: "Count of words in the vocabulary."
+<a name="line-12859"></a>  name: "vocab_count"
+<a name="line-12860"></a>  type: "list(int)"
+<a name="line-12861"></a>}
+<a name="line-12862"></a>attr {
+<a name="line-12863"></a>  description: "Number of negative samples per example."
+<a name="line-12864"></a>  name: "num_negative_samples"
+<a name="line-12865"></a>  type: "int"
+<a name="line-12866"></a>}
+<a name="line-12867"></a>input_arg {
+<a name="line-12868"></a>  description: "input word embedding."
+<a name="line-12869"></a>  is_ref: true
+<a name="line-12870"></a>  name: "w_in"
+<a name="line-12871"></a>  type: DT_FLOAT
+<a name="line-12872"></a>}
+<a name="line-12873"></a>input_arg {
+<a name="line-12874"></a>  description: "output word embedding."
+<a name="line-12875"></a>  is_ref: true
+<a name="line-12876"></a>  name: "w_out"
+<a name="line-12877"></a>  type: DT_FLOAT
+<a name="line-12878"></a>}
+<a name="line-12879"></a>input_arg {
+<a name="line-12880"></a>  description: "A vector of word ids."
+<a name="line-12881"></a>  name: "examples"
+<a name="line-12882"></a>  type: DT_INT32
+<a name="line-12883"></a>}
+<a name="line-12884"></a>input_arg {
+<a name="line-12885"></a>  description: "A vector of word ids." name: "labels" type: DT_INT32
+<a name="line-12886"></a>}
+<a name="line-12887"></a>input_arg { name: "lr" type: DT_FLOAT }
+<a name="line-12888"></a>-}</span>
+<a name="line-12889"></a>
+<a name="line-12890"></a><a name="tensorArrayGrad"></a><span class='hs-comment'>-- | Creates a TensorArray for storing the gradients of values in the given handle.</span>
+<a name="line-12891"></a><span class='hs-comment'>--</span>
+<a name="line-12892"></a><span class='hs-comment'>-- If the given TensorArray gradient already exists, returns a reference to it.</span>
+<a name="line-12893"></a><span class='hs-comment'>-- </span>
+<a name="line-12894"></a><span class='hs-comment'>-- Locks the size of the original TensorArray by disabling its dynamic size flag.</span>
+<a name="line-12895"></a><span class='hs-comment'>-- </span>
+<a name="line-12896"></a><span class='hs-comment'>-- **A note about the input flow_in:**</span>
+<a name="line-12897"></a><span class='hs-comment'>-- </span>
+<a name="line-12898"></a><span class='hs-comment'>-- The handle flow_in forces the execution of the gradient lookup to occur</span>
+<a name="line-12899"></a><span class='hs-comment'>-- only after certain other operations have occurred.  For example, when</span>
+<a name="line-12900"></a><span class='hs-comment'>-- the forward TensorArray is dynamically sized, writes to this TensorArray</span>
+<a name="line-12901"></a><span class='hs-comment'>-- may resize the object.  The gradient TensorArray is statically sized based</span>
+<a name="line-12902"></a><span class='hs-comment'>-- on the size of the forward TensorArray when this operation executes.</span>
+<a name="line-12903"></a><span class='hs-comment'>-- Furthermore, the size of the forward TensorArray is frozen by this call.</span>
+<a name="line-12904"></a><span class='hs-comment'>-- As a result, the flow is used to ensure that the call to generate the gradient</span>
+<a name="line-12905"></a><span class='hs-comment'>-- TensorArray only happens after all writes are executed.</span>
+<a name="line-12906"></a><span class='hs-comment'>-- </span>
+<a name="line-12907"></a><span class='hs-comment'>-- In the case of dynamically sized TensorArrays, gradient computation should</span>
+<a name="line-12908"></a><span class='hs-comment'>-- only be performed on read operations that have themselves been chained via</span>
+<a name="line-12909"></a><span class='hs-comment'>-- flow to occur only after all writes have executed. That way the final size</span>
+<a name="line-12910"></a><span class='hs-comment'>-- of the forward TensorArray is known when this operation is called.</span>
+<a name="line-12911"></a><span class='hs-comment'>-- </span>
+<a name="line-12912"></a><span class='hs-comment'>-- **A note about the source attribute:**</span>
+<a name="line-12913"></a><span class='hs-comment'>-- </span>
+<a name="line-12914"></a><span class='hs-comment'>-- TensorArray gradient calls use an accumulator TensorArray object.  If</span>
+<a name="line-12915"></a><span class='hs-comment'>-- multiple gradients are calculated and run in the same session, the multiple</span>
+<a name="line-12916"></a><span class='hs-comment'>-- gradient nodes may accidentally flow throuth the same accumulator TensorArray.</span>
+<a name="line-12917"></a><span class='hs-comment'>-- This double counts and generally breaks the TensorArray gradient flow.</span>
+<a name="line-12918"></a><span class='hs-comment'>-- </span>
+<a name="line-12919"></a><span class='hs-comment'>-- The solution is to identify which gradient call this particular</span>
+<a name="line-12920"></a><span class='hs-comment'>-- TensorArray gradient is being called in.  This is performed by identifying</span>
+<a name="line-12921"></a><span class='hs-comment'>-- a unique string (e.g. "gradients", "gradients_1", ...) from the input</span>
+<a name="line-12922"></a><span class='hs-comment'>-- gradient Tensor's name.  This string is used as a suffix when creating</span>
+<a name="line-12923"></a><span class='hs-comment'>-- the TensorArray gradient object here (the attribute `source`).</span>
+<a name="line-12924"></a><span class='hs-comment'>-- </span>
+<a name="line-12925"></a><span class='hs-comment'>-- The attribute `source` is added as a suffix to the forward TensorArray's</span>
+<a name="line-12926"></a><span class='hs-comment'>-- name when performing the creation / lookup, so that each separate gradient</span>
+<a name="line-12927"></a><span class='hs-comment'>-- calculation gets its own TensorArray accumulator.</span>
+<a name="line-12928"></a><span class='hs-definition'>tensorArrayGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to the forward TensorArray.</span>
+<a name="line-12929"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __flow_in__: A float scalar that enforces proper chaining of operations.</span>
+<a name="line-12930"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __grad_handle__</span>
+<a name="line-12931"></a><span class='hs-definition'>tensorArrayGrad</span> <span class='hs-varid'>handle</span> <span class='hs-varid'>flow_in</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12932"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TensorArrayGrad"</span><span class='hs-layout'>)</span>
+<a name="line-12933"></a>        <span class='hs-varid'>handle</span> <span class='hs-varid'>flow_in</span>
+<a name="line-12934"></a><span class='hs-comment'>{-
+<a name="line-12935"></a>attr {
+<a name="line-12936"></a>  description: "The gradient source string, used to decide which gradient TensorArray\nto return."
+<a name="line-12937"></a>  name: "source"
+<a name="line-12938"></a>  type: "string"
+<a name="line-12939"></a>}
+<a name="line-12940"></a>input_arg {
+<a name="line-12941"></a>  description: "The handle to the forward TensorArray."
+<a name="line-12942"></a>  name: "handle"
+<a name="line-12943"></a>  type: DT_STRING
+<a name="line-12944"></a>}
+<a name="line-12945"></a>input_arg {
+<a name="line-12946"></a>  description: "A float scalar that enforces proper chaining of operations."
+<a name="line-12947"></a>  name: "flow_in"
+<a name="line-12948"></a>  type: DT_FLOAT
+<a name="line-12949"></a>}
+<a name="line-12950"></a>output_arg { is_ref: true name: "grad_handle" type: DT_STRING }
+<a name="line-12951"></a>-}</span>
+<a name="line-12952"></a>
+<a name="line-12953"></a><a name="audioSummary"></a><span class='hs-comment'>-- | Outputs a `Summary` protocol buffer with audio.</span>
+<a name="line-12954"></a><span class='hs-comment'>--</span>
+<a name="line-12955"></a><span class='hs-comment'>-- The summary has up to `max_outputs` summary values containing audio. The</span>
+<a name="line-12956"></a><span class='hs-comment'>-- audio is built from `tensor` which must be 3-D with shape `[batch_size,</span>
+<a name="line-12957"></a><span class='hs-comment'>-- frames, channels]` or 2-D with shape `[batch_size, frames]`. The values are</span>
+<a name="line-12958"></a><span class='hs-comment'>-- assumed to be in the range of `[-1.0, 1.0]` with a sample rate of `sample_rate`.</span>
+<a name="line-12959"></a><span class='hs-comment'>-- </span>
+<a name="line-12960"></a><span class='hs-comment'>-- The `tag` argument is a scalar `Tensor` of type `string`.  It is used to</span>
+<a name="line-12961"></a><span class='hs-comment'>-- build the `tag` of the summary values:</span>
+<a name="line-12962"></a><span class='hs-comment'>-- </span>
+<a name="line-12963"></a><span class='hs-comment'>-- *  If `max_outputs` is 1, the summary value tag is '*tag*/audio'.</span>
+<a name="line-12964"></a><span class='hs-comment'>-- *  If `max_outputs` is greater than 1, the summary value tags are</span>
+<a name="line-12965"></a><span class='hs-comment'>--    generated sequentially as '*tag*/audio/0', '*tag*/audio/1', etc.</span>
+<a name="line-12966"></a><span class='hs-definition'>audioSummary</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __sample_rate__: The sample rate of the signal in hertz.</span>
+<a name="line-12967"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __tag__: Scalar. Used to build the `tag` attribute of the summary values.</span>
+<a name="line-12968"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __tensor__: 2-D of shape `[batch_size, frames]`.</span>
+<a name="line-12969"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __summary__: Scalar. Serialized `Summary` protocol buffer.</span>
+<a name="line-12970"></a><span class='hs-definition'>audioSummary</span> <span class='hs-varid'>sample_rate</span> <span class='hs-varid'>tag</span> <span class='hs-varid'>tensor</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-12971"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"AudioSummary"</span>
+<a name="line-12972"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"sample_rate"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>sample_rate</span><span class='hs-layout'>)</span>
+<a name="line-12973"></a>        <span class='hs-varid'>tag</span> <span class='hs-varid'>tensor</span>
+<a name="line-12974"></a><span class='hs-comment'>{-
+<a name="line-12975"></a>attr {
+<a name="line-12976"></a>  description: "The sample rate of the signal in hertz."
+<a name="line-12977"></a>  name: "sample_rate"
+<a name="line-12978"></a>  type: "float"
+<a name="line-12979"></a>}
+<a name="line-12980"></a>attr {
+<a name="line-12981"></a>  default_value { i: 3 }
+<a name="line-12982"></a>  description: "Max number of batch elements to generate audio for."
+<a name="line-12983"></a>  has_minimum: true
+<a name="line-12984"></a>  minimum: 1
+<a name="line-12985"></a>  name: "max_outputs"
+<a name="line-12986"></a>  type: "int"
+<a name="line-12987"></a>}
+<a name="line-12988"></a>input_arg {
+<a name="line-12989"></a>  description: "Scalar. Used to build the `tag` attribute of the summary values."
+<a name="line-12990"></a>  name: "tag"
+<a name="line-12991"></a>  type: DT_STRING
+<a name="line-12992"></a>}
+<a name="line-12993"></a>input_arg {
+<a name="line-12994"></a>  description: "2-D of shape `[batch_size, frames]`."
+<a name="line-12995"></a>  name: "tensor"
+<a name="line-12996"></a>  type: DT_FLOAT
+<a name="line-12997"></a>}
+<a name="line-12998"></a>output_arg {
+<a name="line-12999"></a>  description: "Scalar. Serialized `Summary` protocol buffer."
+<a name="line-13000"></a>  name: "summary"
+<a name="line-13001"></a>  type: DT_STRING
+<a name="line-13002"></a>}
+<a name="line-13003"></a>-}</span>
+<a name="line-13004"></a>
+<a name="line-13005"></a><span class='hs-comment'>-- | Does nothing. Only useful as a placeholder for control edges.</span>
+<a name="line-13006"></a>
+<a name="line-13007"></a><a name="noOp"></a><span class='hs-definition'>noOp</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-13008"></a><span class='hs-definition'>noOp</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13009"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"NoOp"</span><span class='hs-layout'>)</span>
+<a name="line-13010"></a>        
+<a name="line-13011"></a><span class='hs-comment'>{-
+<a name="line-13012"></a>
+<a name="line-13013"></a>-}</span>
+<a name="line-13014"></a>
+<a name="line-13015"></a><span class='hs-comment'>-- | Makes its input available to the next iteration.</span>
+<a name="line-13016"></a>
+<a name="line-13017"></a><a name="nextIteration"></a><span class='hs-definition'>nextIteration</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13018"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__: The tensor to be made available to the next iteration.</span>
+<a name="line-13019"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The same tensor as `data`.</span>
+<a name="line-13020"></a><span class='hs-definition'>nextIteration</span> <span class='hs-varid'>data'</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13021"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"NextIteration"</span>
+<a name="line-13022"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13023"></a>        <span class='hs-varid'>data'</span>
+<a name="line-13024"></a><span class='hs-comment'>{-
+<a name="line-13025"></a>attr { name: "T" type: "type" }
+<a name="line-13026"></a>input_arg {
+<a name="line-13027"></a>  description: "The tensor to be made available to the next iteration."
+<a name="line-13028"></a>  name: "data"
+<a name="line-13029"></a>  type_attr: "T"
+<a name="line-13030"></a>}
+<a name="line-13031"></a>output_arg {
+<a name="line-13032"></a>  description: "The same tensor as `data`."
+<a name="line-13033"></a>  name: "output"
+<a name="line-13034"></a>  type_attr: "T"
+<a name="line-13035"></a>}
+<a name="line-13036"></a>-}</span>
+<a name="line-13037"></a>
+<a name="line-13038"></a><span class='hs-comment'>-- | Computes softplus gradients for a softplus operation.</span>
+<a name="line-13039"></a>
+<a name="line-13040"></a><a name="softplusGrad"></a><span class='hs-definition'>softplusGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-13041"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-13042"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-13043"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-13044"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-13045"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-13046"></a>                                                        <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13047"></a>                <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __gradients__: The backpropagated gradients to the corresponding softplus operation.</span>
+<a name="line-13048"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __features__: The features passed as input to the corresponding softplus operation.</span>
+<a name="line-13049"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __backprops__: The gradients: `gradients / (1 + exp(-features))`.</span>
+<a name="line-13050"></a><span class='hs-definition'>softplusGrad</span> <span class='hs-varid'>gradients</span> <span class='hs-varid'>features</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13051"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SoftplusGrad"</span>
+<a name="line-13052"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13053"></a>        <span class='hs-varid'>gradients</span> <span class='hs-varid'>features</span>
+<a name="line-13054"></a><span class='hs-comment'>{-
+<a name="line-13055"></a>attr {
+<a name="line-13056"></a>  allowed_values {
+<a name="line-13057"></a>    list {
+<a name="line-13058"></a>      type: DT_FLOAT
+<a name="line-13059"></a>      type: DT_DOUBLE
+<a name="line-13060"></a>      type: DT_INT32
+<a name="line-13061"></a>      type: DT_INT64
+<a name="line-13062"></a>      type: DT_UINT8
+<a name="line-13063"></a>      type: DT_INT16
+<a name="line-13064"></a>      type: DT_INT8
+<a name="line-13065"></a>      type: DT_UINT16
+<a name="line-13066"></a>      type: DT_HALF
+<a name="line-13067"></a>    }
+<a name="line-13068"></a>  }
+<a name="line-13069"></a>  name: "T"
+<a name="line-13070"></a>  type: "type"
+<a name="line-13071"></a>}
+<a name="line-13072"></a>input_arg {
+<a name="line-13073"></a>  description: "The backpropagated gradients to the corresponding softplus operation."
+<a name="line-13074"></a>  name: "gradients"
+<a name="line-13075"></a>  type_attr: "T"
+<a name="line-13076"></a>}
+<a name="line-13077"></a>input_arg {
+<a name="line-13078"></a>  description: "The features passed as input to the corresponding softplus operation."
+<a name="line-13079"></a>  name: "features"
+<a name="line-13080"></a>  type_attr: "T"
+<a name="line-13081"></a>}
+<a name="line-13082"></a>output_arg {
+<a name="line-13083"></a>  description: "The gradients: `gradients / (1 + exp(-features))`."
+<a name="line-13084"></a>  name: "backprops"
+<a name="line-13085"></a>  type_attr: "T"
+<a name="line-13086"></a>}
+<a name="line-13087"></a>-}</span>
+<a name="line-13088"></a>
+<a name="line-13089"></a><a name="svd"></a><span class='hs-comment'>-- | Computes the singular value decompositions of one or more matrices.</span>
+<a name="line-13090"></a><span class='hs-comment'>--</span>
+<a name="line-13091"></a><span class='hs-comment'>-- Computes the SVD of each inner matrix in `input` such that</span>
+<a name="line-13092"></a><span class='hs-comment'>-- `input[..., :, :] = u[..., :, :] * diag(s[..., :, :]) * transpose(v[..., :, :])`</span>
+<a name="line-13093"></a><span class='hs-comment'>-- </span>
+<a name="line-13094"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-13095"></a><span class='hs-comment'>-- # a is a tensor containing a batch of matrices.</span>
+<a name="line-13096"></a><span class='hs-comment'>-- # s is a tensor of singular values for each matrix.</span>
+<a name="line-13097"></a><span class='hs-comment'>-- # u is the tensor containing of left singular vectors for each matrix.</span>
+<a name="line-13098"></a><span class='hs-comment'>-- # v is the tensor containing of right singular vectors for each matrix.</span>
+<a name="line-13099"></a><span class='hs-comment'>-- s, u, v = svd(a)</span>
+<a name="line-13100"></a><span class='hs-comment'>-- s, _, _ = svd(a, compute_uv=False)</span>
+<a name="line-13101"></a><span class='hs-comment'>-- ```</span>
+<a name="line-13102"></a><span class='hs-definition'>svd</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-13103"></a>                                            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-13104"></a>                                            <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13105"></a>       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: A tensor of shape `[..., M, N]` whose inner-most 2 dimensions</span>
+<a name="line-13106"></a>                   <span class='hs-comment'>-- form matrices of size `[M, N]`. Let `P` be the minimum of `M` and `N`.</span>
+<a name="line-13107"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-13108"></a>       <span class='hs-comment'>-- ^ (__s__, __u__, __v__)</span>
+<a name="line-13109"></a>       <span class='hs-comment'>--</span>
+<a name="line-13110"></a>       <span class='hs-comment'>-- * __s__: Singular values. Shape is `[..., P]`.</span>
+<a name="line-13111"></a>       <span class='hs-comment'>--</span>
+<a name="line-13112"></a>       <span class='hs-comment'>-- * __u__: Left singular vectors. If `full_matrices` is `False` then shape is</span>
+<a name="line-13113"></a>       <span class='hs-comment'>-- `[..., M, M]`; if `full_matrices` is `True` then shape is</span>
+<a name="line-13114"></a>       <span class='hs-comment'>-- `[..., M, P]`. Undefined if `compute_uv` is `False`.</span>
+<a name="line-13115"></a>       <span class='hs-comment'>--</span>
+<a name="line-13116"></a>       <span class='hs-comment'>-- * __v__: Left singular vectors. If `full_matrices` is `False` then shape is</span>
+<a name="line-13117"></a>       <span class='hs-comment'>-- `[..., N, N]`. If `full_matrices` is `True` then shape is `[..., N, P]`.</span>
+<a name="line-13118"></a>       <span class='hs-comment'>-- Undefined if `compute_uv` is false.</span>
+<a name="line-13119"></a><span class='hs-definition'>svd</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13120"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Svd"</span>
+<a name="line-13121"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13122"></a>        <span class='hs-varid'>input</span>
+<a name="line-13123"></a><span class='hs-comment'>{-
+<a name="line-13124"></a>attr {
+<a name="line-13125"></a>  default_value { b: true }
+<a name="line-13126"></a>  description: "If true, left and right singular vectors will be\ncomputed and returned in `u` and `v`, respectively.\nIf false, `u` and `v` are not set and should never referenced."
+<a name="line-13127"></a>  name: "compute_uv"
+<a name="line-13128"></a>  type: "bool"
+<a name="line-13129"></a>}
+<a name="line-13130"></a>attr {
+<a name="line-13131"></a>  default_value { b: false }
+<a name="line-13132"></a>  description: "If true, compute full-sized `u` and `v`. If false\n(the default), compute only the leading `P` singular vectors.\nIgnored if `compute_uv` is `False`."
+<a name="line-13133"></a>  name: "full_matrices"
+<a name="line-13134"></a>  type: "bool"
+<a name="line-13135"></a>}
+<a name="line-13136"></a>attr {
+<a name="line-13137"></a>  allowed_values {
+<a name="line-13138"></a>    list {
+<a name="line-13139"></a>      type: DT_DOUBLE
+<a name="line-13140"></a>      type: DT_FLOAT
+<a name="line-13141"></a>      type: DT_COMPLEX64
+<a name="line-13142"></a>      type: DT_COMPLEX128
+<a name="line-13143"></a>    }
+<a name="line-13144"></a>  }
+<a name="line-13145"></a>  name: "T"
+<a name="line-13146"></a>  type: "type"
+<a name="line-13147"></a>}
+<a name="line-13148"></a>input_arg {
+<a name="line-13149"></a>  description: "A tensor of shape `[..., M, N]` whose inner-most 2 dimensions\nform matrices of size `[M, N]`. Let `P` be the minimum of `M` and `N`."
+<a name="line-13150"></a>  name: "input"
+<a name="line-13151"></a>  type_attr: "T"
+<a name="line-13152"></a>}
+<a name="line-13153"></a>output_arg {
+<a name="line-13154"></a>  description: "Singular values. Shape is `[..., P]`."
+<a name="line-13155"></a>  name: "s"
+<a name="line-13156"></a>  type_attr: "T"
+<a name="line-13157"></a>}
+<a name="line-13158"></a>output_arg {
+<a name="line-13159"></a>  description: "Left singular vectors. If `full_matrices` is `False` then shape is\n`[..., M, M]`; if `full_matrices` is `True` then shape is\n`[..., M, P]`. Undefined if `compute_uv` is `False`."
+<a name="line-13160"></a>  name: "u"
+<a name="line-13161"></a>  type_attr: "T"
+<a name="line-13162"></a>}
+<a name="line-13163"></a>output_arg {
+<a name="line-13164"></a>  description: "Left singular vectors. If `full_matrices` is `False` then shape is\n`[..., N, N]`. If `full_matrices` is `True` then shape is `[..., N, P]`.\nUndefined if `compute_uv` is false."
+<a name="line-13165"></a>  name: "v"
+<a name="line-13166"></a>  type_attr: "T"
+<a name="line-13167"></a>}
+<a name="line-13168"></a>-}</span>
+<a name="line-13169"></a>
+<a name="line-13170"></a><a name="hSVToRGB"></a><span class='hs-comment'>-- | Convert one or more images from HSV to RGB.</span>
+<a name="line-13171"></a><span class='hs-comment'>--</span>
+<a name="line-13172"></a><span class='hs-comment'>-- Outputs a tensor of the same shape as the `images` tensor, containing the RGB</span>
+<a name="line-13173"></a><span class='hs-comment'>-- value of the pixels. The output is only well defined if the value in `images`</span>
+<a name="line-13174"></a><span class='hs-comment'>-- are in `[0,1]`.</span>
+<a name="line-13175"></a><span class='hs-comment'>-- </span>
+<a name="line-13176"></a><span class='hs-comment'>-- See `rgb_to_hsv` for a description of the HSV encoding.</span>
+<a name="line-13177"></a><span class='hs-definition'>hSVToRGB</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13178"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __images__: 1-D or higher rank. HSV data to convert. Last dimension must be size 3.</span>
+<a name="line-13179"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: `images` converted to RGB.</span>
+<a name="line-13180"></a><span class='hs-definition'>hSVToRGB</span> <span class='hs-varid'>images</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13181"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"HSVToRGB"</span>
+<a name="line-13182"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13183"></a>        <span class='hs-varid'>images</span>
+<a name="line-13184"></a><span class='hs-comment'>{-
+<a name="line-13185"></a>attr {
+<a name="line-13186"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-13187"></a>  default_value { type: DT_FLOAT }
+<a name="line-13188"></a>  name: "T"
+<a name="line-13189"></a>  type: "type"
+<a name="line-13190"></a>}
+<a name="line-13191"></a>input_arg {
+<a name="line-13192"></a>  description: "1-D or higher rank. HSV data to convert. Last dimension must be size 3."
+<a name="line-13193"></a>  name: "images"
+<a name="line-13194"></a>  type_attr: "T"
+<a name="line-13195"></a>}
+<a name="line-13196"></a>output_arg {
+<a name="line-13197"></a>  description: "`images` converted to RGB."
+<a name="line-13198"></a>  name: "output"
+<a name="line-13199"></a>  type_attr: "T"
+<a name="line-13200"></a>}
+<a name="line-13201"></a>-}</span>
+<a name="line-13202"></a>
+<a name="line-13203"></a><a name="parameterizedTruncatedNormal"></a><span class='hs-comment'>-- | Outputs random values from a normal distribution. The parameters may each be a</span>
+<a name="line-13204"></a><span class='hs-comment'>--</span>
+<a name="line-13205"></a><span class='hs-comment'>-- scalar which applies to the entire output, or a vector of length shape[0] which</span>
+<a name="line-13206"></a><span class='hs-comment'>-- stores the parameters for each batch.</span>
+<a name="line-13207"></a><span class='hs-definition'>parameterizedTruncatedNormal</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-13208"></a>                                                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-13209"></a>                                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-13210"></a>                                                                 <span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>,</span>
+<a name="line-13211"></a>                                                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-13212"></a>                                                                         <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-13213"></a>                                                                         <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13214"></a>                                <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __shape__: The shape of the output tensor. Batches are indexed by the 0th dimension.</span>
+<a name="line-13215"></a>                                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __means__: The mean parameter of each batch.</span>
+<a name="line-13216"></a>                                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __stdevs__: The standard deviation parameter of each batch. Must be greater than 0.</span>
+<a name="line-13217"></a>                                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __minvals__: The minimum cutoff. May be -infinity.</span>
+<a name="line-13218"></a>                                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __maxvals__: The maximum cutoff. May be +infinity, and must be more than the minval</span>
+<a name="line-13219"></a>                                                   <span class='hs-comment'>-- for each batch.</span>
+<a name="line-13220"></a>                                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __output__: A matrix of shape num_batches x samples_per_batch, filled with random</span>
+<a name="line-13221"></a>                                <span class='hs-comment'>-- truncated normal values using the parameters for each row.</span>
+<a name="line-13222"></a><span class='hs-definition'>parameterizedTruncatedNormal</span> <span class='hs-varid'>shape</span> <span class='hs-varid'>means</span> <span class='hs-varid'>stdevs</span> <span class='hs-varid'>minvals</span>
+<a name="line-13223"></a>                             <span class='hs-varid'>maxvals</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13224"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ParameterizedTruncatedNormal"</span>
+<a name="line-13225"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-13226"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13227"></a>        <span class='hs-varid'>shape</span> <span class='hs-varid'>means</span> <span class='hs-varid'>stdevs</span> <span class='hs-varid'>minvals</span> <span class='hs-varid'>maxvals</span>
+<a name="line-13228"></a><span class='hs-comment'>{-
+<a name="line-13229"></a>attr {
+<a name="line-13230"></a>  default_value { i: 0 }
+<a name="line-13231"></a>  description: "If either `seed` or `seed2` are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-13232"></a>  name: "seed"
+<a name="line-13233"></a>  type: "int"
+<a name="line-13234"></a>}
+<a name="line-13235"></a>attr {
+<a name="line-13236"></a>  default_value { i: 0 }
+<a name="line-13237"></a>  description: "A second seed to avoid seed collision."
+<a name="line-13238"></a>  name: "seed2"
+<a name="line-13239"></a>  type: "int"
+<a name="line-13240"></a>}
+<a name="line-13241"></a>attr {
+<a name="line-13242"></a>  allowed_values {
+<a name="line-13243"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-13244"></a>  }
+<a name="line-13245"></a>  description: "The type of the output."
+<a name="line-13246"></a>  name: "dtype"
+<a name="line-13247"></a>  type: "type"
+<a name="line-13248"></a>}
+<a name="line-13249"></a>attr {
+<a name="line-13250"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-13251"></a>  name: "T"
+<a name="line-13252"></a>  type: "type"
+<a name="line-13253"></a>}
+<a name="line-13254"></a>input_arg {
+<a name="line-13255"></a>  description: "The shape of the output tensor. Batches are indexed by the 0th dimension."
+<a name="line-13256"></a>  name: "shape"
+<a name="line-13257"></a>  type_attr: "T"
+<a name="line-13258"></a>}
+<a name="line-13259"></a>input_arg {
+<a name="line-13260"></a>  description: "The mean parameter of each batch."
+<a name="line-13261"></a>  name: "means"
+<a name="line-13262"></a>  type_attr: "dtype"
+<a name="line-13263"></a>}
+<a name="line-13264"></a>input_arg {
+<a name="line-13265"></a>  description: "The standard deviation parameter of each batch. Must be greater than 0."
+<a name="line-13266"></a>  name: "stdevs"
+<a name="line-13267"></a>  type_attr: "dtype"
+<a name="line-13268"></a>}
+<a name="line-13269"></a>input_arg {
+<a name="line-13270"></a>  description: "The minimum cutoff. May be -infinity."
+<a name="line-13271"></a>  name: "minvals"
+<a name="line-13272"></a>  type_attr: "dtype"
+<a name="line-13273"></a>}
+<a name="line-13274"></a>input_arg {
+<a name="line-13275"></a>  description: "The maximum cutoff. May be +infinity, and must be more than the minval\nfor each batch."
+<a name="line-13276"></a>  name: "maxvals"
+<a name="line-13277"></a>  type_attr: "dtype"
+<a name="line-13278"></a>}
+<a name="line-13279"></a>output_arg {
+<a name="line-13280"></a>  description: "A matrix of shape num_batches x samples_per_batch, filled with random\ntruncated normal values using the parameters for each row."
+<a name="line-13281"></a>  name: "output"
+<a name="line-13282"></a>  type_attr: "dtype"
+<a name="line-13283"></a>}
+<a name="line-13284"></a>-}</span>
+<a name="line-13285"></a>
+<a name="line-13286"></a><a name="square"></a><span class='hs-comment'>-- | Computes square of x element-wise.</span>
+<a name="line-13287"></a><span class='hs-comment'>--</span>
+<a name="line-13288"></a><span class='hs-comment'>-- I.e., \\(y = x * x = x^2\\).</span>
+<a name="line-13289"></a><span class='hs-definition'>square</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-13290"></a>                                               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-13291"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-13292"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-13293"></a>                                               <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13294"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-13295"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-13296"></a><span class='hs-definition'>square</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13297"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Square"</span>
+<a name="line-13298"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13299"></a>        <span class='hs-varid'>x</span>
+<a name="line-13300"></a><span class='hs-comment'>{-
+<a name="line-13301"></a>attr {
+<a name="line-13302"></a>  allowed_values {
+<a name="line-13303"></a>    list {
+<a name="line-13304"></a>      type: DT_HALF
+<a name="line-13305"></a>      type: DT_FLOAT
+<a name="line-13306"></a>      type: DT_DOUBLE
+<a name="line-13307"></a>      type: DT_INT32
+<a name="line-13308"></a>      type: DT_INT64
+<a name="line-13309"></a>      type: DT_COMPLEX64
+<a name="line-13310"></a>      type: DT_COMPLEX128
+<a name="line-13311"></a>    }
+<a name="line-13312"></a>  }
+<a name="line-13313"></a>  name: "T"
+<a name="line-13314"></a>  type: "type"
+<a name="line-13315"></a>}
+<a name="line-13316"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-13317"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-13318"></a>-}</span>
+<a name="line-13319"></a>
+<a name="line-13320"></a><a name="elu"></a><span class='hs-comment'>-- | Computes exponential linear: `exp(features) - 1` if &lt; 0, `features` otherwise.</span>
+<a name="line-13321"></a><span class='hs-comment'>--</span>
+<a name="line-13322"></a><span class='hs-comment'>-- See [Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)</span>
+<a name="line-13323"></a><span class='hs-comment'>-- ](<a href="http://arxiv.org/abs/1511.07289)">http://arxiv.org/abs/1511.07289)</a></span>
+<a name="line-13324"></a><span class='hs-definition'>elu</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-13325"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-13326"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-13327"></a>                                            <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13328"></a>       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __features__</span>
+<a name="line-13329"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __activations__</span>
+<a name="line-13330"></a><span class='hs-definition'>elu</span> <span class='hs-varid'>features</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13331"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Elu"</span>
+<a name="line-13332"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13333"></a>        <span class='hs-varid'>features</span>
+<a name="line-13334"></a><span class='hs-comment'>{-
+<a name="line-13335"></a>attr {
+<a name="line-13336"></a>  allowed_values {
+<a name="line-13337"></a>    list {
+<a name="line-13338"></a>      type: DT_FLOAT
+<a name="line-13339"></a>      type: DT_DOUBLE
+<a name="line-13340"></a>      type: DT_INT32
+<a name="line-13341"></a>      type: DT_INT64
+<a name="line-13342"></a>      type: DT_UINT8
+<a name="line-13343"></a>      type: DT_INT16
+<a name="line-13344"></a>      type: DT_INT8
+<a name="line-13345"></a>      type: DT_UINT16
+<a name="line-13346"></a>      type: DT_HALF
+<a name="line-13347"></a>    }
+<a name="line-13348"></a>  }
+<a name="line-13349"></a>  name: "T"
+<a name="line-13350"></a>  type: "type"
+<a name="line-13351"></a>}
+<a name="line-13352"></a>input_arg { name: "features" type_attr: "T" }
+<a name="line-13353"></a>output_arg { name: "activations" type_attr: "T" }
+<a name="line-13354"></a>-}</span>
+<a name="line-13355"></a>
+<a name="line-13356"></a><span class='hs-comment'>-- | Outputs all keys and values in the table.</span>
+<a name="line-13357"></a>
+<a name="line-13358"></a><a name="lookupTableExport"></a><span class='hs-definition'>lookupTableExport</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>tkeys</span> <span class='hs-varid'>tvalues</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>tkeys</span><span class='hs-layout'>,</span>
+<a name="line-13359"></a>                                                <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tvalues</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13360"></a>                     <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __table_handle__: Handle to the table.</span>
+<a name="line-13361"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>tkeys</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>tvalues</span><span class='hs-layout'>)</span>
+<a name="line-13362"></a>                     <span class='hs-comment'>-- ^ (__keys__, __values__)</span>
+<a name="line-13363"></a>                     <span class='hs-comment'>--</span>
+<a name="line-13364"></a>                     <span class='hs-comment'>-- * __keys__: Vector of all keys present in the table.</span>
+<a name="line-13365"></a>                     <span class='hs-comment'>--</span>
+<a name="line-13366"></a>                     <span class='hs-comment'>-- * __values__: Tensor of all values in the table. Indexed in parallel with `keys`.</span>
+<a name="line-13367"></a><span class='hs-definition'>lookupTableExport</span> <span class='hs-varid'>table_handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13368"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"LookupTableExport"</span>
+<a name="line-13369"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tkeys"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tkeys</span><span class='hs-layout'>)</span>
+<a name="line-13370"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tvalues"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tvalues</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13371"></a>        <span class='hs-varid'>table_handle</span>
+<a name="line-13372"></a><span class='hs-comment'>{-
+<a name="line-13373"></a>attr { name: "Tkeys" type: "type" }
+<a name="line-13374"></a>attr { name: "Tvalues" type: "type" }
+<a name="line-13375"></a>input_arg {
+<a name="line-13376"></a>  description: "Handle to the table."
+<a name="line-13377"></a>  is_ref: true
+<a name="line-13378"></a>  name: "table_handle"
+<a name="line-13379"></a>  type: DT_STRING
+<a name="line-13380"></a>}
+<a name="line-13381"></a>output_arg {
+<a name="line-13382"></a>  description: "Vector of all keys present in the table."
+<a name="line-13383"></a>  name: "keys"
+<a name="line-13384"></a>  type_attr: "Tkeys"
+<a name="line-13385"></a>}
+<a name="line-13386"></a>output_arg {
+<a name="line-13387"></a>  description: "Tensor of all values in the table. Indexed in parallel with `keys`."
+<a name="line-13388"></a>  name: "values"
+<a name="line-13389"></a>  type_attr: "Tvalues"
+<a name="line-13390"></a>}
+<a name="line-13391"></a>-}</span>
+<a name="line-13392"></a>
+<a name="line-13393"></a><span class='hs-comment'>-- | Computes the number of elements in the given table.</span>
+<a name="line-13394"></a>
+<a name="line-13395"></a><a name="lookupTableSize"></a><span class='hs-definition'>lookupTableSize</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __table_handle__: Handle to the table.</span>
+<a name="line-13396"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __size__: Scalar that contains number of elements in the table.</span>
+<a name="line-13397"></a><span class='hs-definition'>lookupTableSize</span> <span class='hs-varid'>table_handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13398"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"LookupTableSize"</span><span class='hs-layout'>)</span>
+<a name="line-13399"></a>        <span class='hs-varid'>table_handle</span>
+<a name="line-13400"></a><span class='hs-comment'>{-
+<a name="line-13401"></a>input_arg {
+<a name="line-13402"></a>  description: "Handle to the table."
+<a name="line-13403"></a>  is_ref: true
+<a name="line-13404"></a>  name: "table_handle"
+<a name="line-13405"></a>  type: DT_STRING
+<a name="line-13406"></a>}
+<a name="line-13407"></a>output_arg {
+<a name="line-13408"></a>  description: "Scalar that contains number of elements in the table."
+<a name="line-13409"></a>  name: "size"
+<a name="line-13410"></a>  type: DT_INT64
+<a name="line-13411"></a>}
+<a name="line-13412"></a>-}</span>
+<a name="line-13413"></a>
+<a name="line-13414"></a><span class='hs-comment'>-- | Computes gradients of the average pooling function.</span>
+<a name="line-13415"></a>
+<a name="line-13416"></a><a name="avgPoolGrad"></a><span class='hs-definition'>avgPoolGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-13417"></a>                                                       <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13418"></a>               <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __orig_input_shape__: 1-D.  Shape of the original input to `avg_pool`.</span>
+<a name="line-13419"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: 4-D with shape `[batch, height, width, channels]`.  Gradients w.r.t.</span>
+<a name="line-13420"></a>                              <span class='hs-comment'>-- the output of `avg_pool`.</span>
+<a name="line-13421"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 4-D.  Gradients w.r.t. the input of `avg_pool`.</span>
+<a name="line-13422"></a><span class='hs-definition'>avgPoolGrad</span> <span class='hs-varid'>orig_input_shape</span> <span class='hs-varid'>grad</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13423"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"AvgPoolGrad"</span>
+<a name="line-13424"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13425"></a>        <span class='hs-varid'>orig_input_shape</span> <span class='hs-varid'>grad</span>
+<a name="line-13426"></a><span class='hs-comment'>{-
+<a name="line-13427"></a>attr {
+<a name="line-13428"></a>  description: "The size of the sliding window for each dimension of the input."
+<a name="line-13429"></a>  has_minimum: true
+<a name="line-13430"></a>  minimum: 4
+<a name="line-13431"></a>  name: "ksize"
+<a name="line-13432"></a>  type: "list(int)"
+<a name="line-13433"></a>}
+<a name="line-13434"></a>attr {
+<a name="line-13435"></a>  description: "The stride of the sliding window for each dimension of the input."
+<a name="line-13436"></a>  has_minimum: true
+<a name="line-13437"></a>  minimum: 4
+<a name="line-13438"></a>  name: "strides"
+<a name="line-13439"></a>  type: "list(int)"
+<a name="line-13440"></a>}
+<a name="line-13441"></a>attr {
+<a name="line-13442"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-13443"></a>  description: "The type of padding algorithm to use."
+<a name="line-13444"></a>  name: "padding"
+<a name="line-13445"></a>  type: "string"
+<a name="line-13446"></a>}
+<a name="line-13447"></a>attr {
+<a name="line-13448"></a>  allowed_values { list { s: "NHWC" s: "NCHW" } }
+<a name="line-13449"></a>  default_value { s: "NHWC" }
+<a name="line-13450"></a>  description: "Specify the data format of the input and output data. With the\ndefault format \"NHWC\", the data is stored in the order of:\n    [batch, in_height, in_width, in_channels].\nAlternatively, the format could be \"NCHW\", the data storage order of:\n    [batch, in_channels, in_height, in_width]."
+<a name="line-13451"></a>  name: "data_format"
+<a name="line-13452"></a>  type: "string"
+<a name="line-13453"></a>}
+<a name="line-13454"></a>attr {
+<a name="line-13455"></a>  allowed_values {
+<a name="line-13456"></a>    list { type: DT_FLOAT type: DT_HALF type: DT_DOUBLE }
+<a name="line-13457"></a>  }
+<a name="line-13458"></a>  name: "T"
+<a name="line-13459"></a>  type: "type"
+<a name="line-13460"></a>}
+<a name="line-13461"></a>input_arg {
+<a name="line-13462"></a>  description: "1-D.  Shape of the original input to `avg_pool`."
+<a name="line-13463"></a>  name: "orig_input_shape"
+<a name="line-13464"></a>  type: DT_INT32
+<a name="line-13465"></a>}
+<a name="line-13466"></a>input_arg {
+<a name="line-13467"></a>  description: "4-D with shape `[batch, height, width, channels]`.  Gradients w.r.t.\nthe output of `avg_pool`."
+<a name="line-13468"></a>  name: "grad"
+<a name="line-13469"></a>  type_attr: "T"
+<a name="line-13470"></a>}
+<a name="line-13471"></a>output_arg {
+<a name="line-13472"></a>  description: "4-D.  Gradients w.r.t. the input of `avg_pool`."
+<a name="line-13473"></a>  name: "output"
+<a name="line-13474"></a>  type_attr: "T"
+<a name="line-13475"></a>}
+<a name="line-13476"></a>-}</span>
+<a name="line-13477"></a>
+<a name="line-13478"></a><a name="computeAccidentalHits"></a><span class='hs-comment'>-- | Computes the ids of the positions in sampled_candidates that match true_labels.</span>
+<a name="line-13479"></a><span class='hs-comment'>--</span>
+<a name="line-13480"></a><span class='hs-comment'>-- When doing log-odds NCE, the result of this op should be passed through a</span>
+<a name="line-13481"></a><span class='hs-comment'>-- SparseToDense op, then added to the logits of the sampled candidates. This has</span>
+<a name="line-13482"></a><span class='hs-comment'>-- the effect of 'removing' the sampled labels that match the true labels by</span>
+<a name="line-13483"></a><span class='hs-comment'>-- making the classifier sure that they are sampled labels.</span>
+<a name="line-13484"></a><span class='hs-definition'>computeAccidentalHits</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_true__: Number of true labels per context.</span>
+<a name="line-13485"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __true_classes__: The true_classes output of UnpackSparseLabels.</span>
+<a name="line-13486"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __sampled_candidates__: The sampled_candidates output of CandidateSampler.</span>
+<a name="line-13487"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-13488"></a>                             <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span>
+<a name="line-13489"></a>                         <span class='hs-comment'>-- ^ (__indices__, __ids__, __weights__)</span>
+<a name="line-13490"></a>                         <span class='hs-comment'>--</span>
+<a name="line-13491"></a>                         <span class='hs-comment'>-- * __indices__: A vector of indices corresponding to rows of true_candidates.</span>
+<a name="line-13492"></a>                         <span class='hs-comment'>--</span>
+<a name="line-13493"></a>                         <span class='hs-comment'>-- * __ids__: A vector of IDs of positions in sampled_candidates that match a true_label</span>
+<a name="line-13494"></a>                         <span class='hs-comment'>-- for the row with the corresponding index in indices.</span>
+<a name="line-13495"></a>                         <span class='hs-comment'>--</span>
+<a name="line-13496"></a>                         <span class='hs-comment'>-- * __weights__: A vector of the same length as indices and ids, in which each element</span>
+<a name="line-13497"></a>                         <span class='hs-comment'>-- is -FLOAT_MAX.</span>
+<a name="line-13498"></a><span class='hs-definition'>computeAccidentalHits</span> <span class='hs-varid'>num_true</span> <span class='hs-varid'>true_classes</span>
+<a name="line-13499"></a>                      <span class='hs-varid'>sampled_candidates</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13500"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ComputeAccidentalHits"</span>
+<a name="line-13501"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_true"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_true</span><span class='hs-layout'>)</span>
+<a name="line-13502"></a>        <span class='hs-varid'>true_classes</span> <span class='hs-varid'>sampled_candidates</span>
+<a name="line-13503"></a><span class='hs-comment'>{-
+<a name="line-13504"></a>attr {
+<a name="line-13505"></a>  description: "Number of true labels per context."
+<a name="line-13506"></a>  name: "num_true"
+<a name="line-13507"></a>  type: "int"
+<a name="line-13508"></a>}
+<a name="line-13509"></a>attr {
+<a name="line-13510"></a>  default_value { i: 0 }
+<a name="line-13511"></a>  description: "If either seed or seed2 are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-13512"></a>  name: "seed"
+<a name="line-13513"></a>  type: "int"
+<a name="line-13514"></a>}
+<a name="line-13515"></a>attr {
+<a name="line-13516"></a>  default_value { i: 0 }
+<a name="line-13517"></a>  description: "An second seed to avoid seed collision."
+<a name="line-13518"></a>  name: "seed2"
+<a name="line-13519"></a>  type: "int"
+<a name="line-13520"></a>}
+<a name="line-13521"></a>input_arg {
+<a name="line-13522"></a>  description: "The true_classes output of UnpackSparseLabels."
+<a name="line-13523"></a>  name: "true_classes"
+<a name="line-13524"></a>  type: DT_INT64
+<a name="line-13525"></a>}
+<a name="line-13526"></a>input_arg {
+<a name="line-13527"></a>  description: "The sampled_candidates output of CandidateSampler."
+<a name="line-13528"></a>  name: "sampled_candidates"
+<a name="line-13529"></a>  type: DT_INT64
+<a name="line-13530"></a>}
+<a name="line-13531"></a>output_arg {
+<a name="line-13532"></a>  description: "A vector of indices corresponding to rows of true_candidates."
+<a name="line-13533"></a>  name: "indices"
+<a name="line-13534"></a>  type: DT_INT32
+<a name="line-13535"></a>}
+<a name="line-13536"></a>output_arg {
+<a name="line-13537"></a>  description: "A vector of IDs of positions in sampled_candidates that match a true_label\nfor the row with the corresponding index in indices."
+<a name="line-13538"></a>  name: "ids"
+<a name="line-13539"></a>  type: DT_INT64
+<a name="line-13540"></a>}
+<a name="line-13541"></a>output_arg {
+<a name="line-13542"></a>  description: "A vector of the same length as indices and ids, in which each element\nis -FLOAT_MAX."
+<a name="line-13543"></a>  name: "weights"
+<a name="line-13544"></a>  type: DT_FLOAT
+<a name="line-13545"></a>}
+<a name="line-13546"></a>-}</span>
+<a name="line-13547"></a>
+<a name="line-13548"></a><a name="cTCLoss"></a><span class='hs-comment'>-- | Calculates the CTC Loss (log probability) for each batch entry.  Also calculates</span>
+<a name="line-13549"></a><span class='hs-comment'>--</span>
+<a name="line-13550"></a><span class='hs-comment'>-- the gradient.  This class performs the softmax operation for you, so inputs</span>
+<a name="line-13551"></a><span class='hs-comment'>-- should be e.g. linear projections of outputs by an LSTM.</span>
+<a name="line-13552"></a><span class='hs-definition'>cTCLoss</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __inputs__: 3-D, shape: `(max_time x batch_size x num_classes)`, the logits.</span>
+<a name="line-13553"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __labels_indices__: The indices of a `SparseTensor&lt;int32, 2&gt;`.</span>
+<a name="line-13554"></a>                                       <span class='hs-comment'>-- `labels_indices(i, :) == [b, t]` means `labels_values(i)` stores the id for</span>
+<a name="line-13555"></a>                                       <span class='hs-comment'>-- `(batch b, time t)`.</span>
+<a name="line-13556"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __labels_values__: The values (labels) associated with the given batch and time.</span>
+<a name="line-13557"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __sequence_length__: A vector containing sequence lengths (batch).</span>
+<a name="line-13558"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span>
+<a name="line-13559"></a>           <span class='hs-comment'>-- ^ (__loss__, __gradient__)</span>
+<a name="line-13560"></a>           <span class='hs-comment'>--</span>
+<a name="line-13561"></a>           <span class='hs-comment'>-- * __loss__: A vector (batch) containing log-probabilities.</span>
+<a name="line-13562"></a>           <span class='hs-comment'>--</span>
+<a name="line-13563"></a>           <span class='hs-comment'>-- * __gradient__: The gradient of `loss`.  3-D, shape:</span>
+<a name="line-13564"></a>           <span class='hs-comment'>-- `(max_time x batch_size x num_classes)`.</span>
+<a name="line-13565"></a><span class='hs-definition'>cTCLoss</span> <span class='hs-varid'>inputs</span> <span class='hs-varid'>labels_indices</span> <span class='hs-varid'>labels_values</span> <span class='hs-varid'>sequence_length</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13566"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"CTCLoss"</span><span class='hs-layout'>)</span>
+<a name="line-13567"></a>        <span class='hs-varid'>inputs</span> <span class='hs-varid'>labels_indices</span> <span class='hs-varid'>labels_values</span> <span class='hs-varid'>sequence_length</span>
+<a name="line-13568"></a><span class='hs-comment'>{-
+<a name="line-13569"></a>attr {
+<a name="line-13570"></a>  default_value { b: false }
+<a name="line-13571"></a>  description: "Scalar, if true then repeated labels are\ncollapsed prior to the CTC calculation."
+<a name="line-13572"></a>  name: "preprocess_collapse_repeated"
+<a name="line-13573"></a>  type: "bool"
+<a name="line-13574"></a>}
+<a name="line-13575"></a>attr {
+<a name="line-13576"></a>  default_value { b: true }
+<a name="line-13577"></a>  description: "Scalar.  If set to false, *during* CTC calculation\nrepeated non-blank labels will not be merged and are interpreted as\nindividual labels.  This is a simplified version of CTC."
+<a name="line-13578"></a>  name: "ctc_merge_repeated"
+<a name="line-13579"></a>  type: "bool"
+<a name="line-13580"></a>}
+<a name="line-13581"></a>input_arg {
+<a name="line-13582"></a>  description: "3-D, shape: `(max_time x batch_size x num_classes)`, the logits."
+<a name="line-13583"></a>  name: "inputs"
+<a name="line-13584"></a>  type: DT_FLOAT
+<a name="line-13585"></a>}
+<a name="line-13586"></a>input_arg {
+<a name="line-13587"></a>  description: "The indices of a `SparseTensor&lt;int32, 2&gt;`.\n`labels_indices(i, :) == [b, t]` means `labels_values(i)` stores the id for\n`(batch b, time t)`."
+<a name="line-13588"></a>  name: "labels_indices"
+<a name="line-13589"></a>  type: DT_INT64
+<a name="line-13590"></a>}
+<a name="line-13591"></a>input_arg {
+<a name="line-13592"></a>  description: "The values (labels) associated with the given batch and time."
+<a name="line-13593"></a>  name: "labels_values"
+<a name="line-13594"></a>  type: DT_INT32
+<a name="line-13595"></a>}
+<a name="line-13596"></a>input_arg {
+<a name="line-13597"></a>  description: "A vector containing sequence lengths (batch)."
+<a name="line-13598"></a>  name: "sequence_length"
+<a name="line-13599"></a>  type: DT_INT32
+<a name="line-13600"></a>}
+<a name="line-13601"></a>output_arg {
+<a name="line-13602"></a>  description: "A vector (batch) containing log-probabilities."
+<a name="line-13603"></a>  name: "loss"
+<a name="line-13604"></a>  type: DT_FLOAT
+<a name="line-13605"></a>}
+<a name="line-13606"></a>output_arg {
+<a name="line-13607"></a>  description: "The gradient of `loss`.  3-D, shape:\n`(max_time x batch_size x num_classes)`."
+<a name="line-13608"></a>  name: "gradient"
+<a name="line-13609"></a>  type: DT_FLOAT
+<a name="line-13610"></a>}
+<a name="line-13611"></a>-}</span>
+<a name="line-13612"></a>
+<a name="line-13613"></a><span class='hs-comment'>-- | Performs 3D average pooling on the input.</span>
+<a name="line-13614"></a>
+<a name="line-13615"></a><a name="avgPool3D"></a><span class='hs-definition'>avgPool3D</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-13616"></a>                                                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-13617"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-13618"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-13619"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-13620"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-13621"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-13622"></a>                                                  <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13623"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Shape `[batch, depth, rows, cols, channels]` tensor to pool over.</span>
+<a name="line-13624"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The average pooled output tensor.</span>
+<a name="line-13625"></a><span class='hs-definition'>avgPool3D</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13626"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"AvgPool3D"</span>
+<a name="line-13627"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13628"></a>        <span class='hs-varid'>input</span>
+<a name="line-13629"></a><span class='hs-comment'>{-
+<a name="line-13630"></a>attr {
+<a name="line-13631"></a>  description: "1-D tensor of length 5. The size of the window for each dimension of\nthe input tensor. Must have `ksize[0] = ksize[4] = 1`."
+<a name="line-13632"></a>  has_minimum: true
+<a name="line-13633"></a>  minimum: 5
+<a name="line-13634"></a>  name: "ksize"
+<a name="line-13635"></a>  type: "list(int)"
+<a name="line-13636"></a>}
+<a name="line-13637"></a>attr {
+<a name="line-13638"></a>  description: "1-D tensor of length 5. The stride of the sliding window for each\ndimension of `input`. Must have `strides[0] = strides[4] = 1`."
+<a name="line-13639"></a>  has_minimum: true
+<a name="line-13640"></a>  minimum: 5
+<a name="line-13641"></a>  name: "strides"
+<a name="line-13642"></a>  type: "list(int)"
+<a name="line-13643"></a>}
+<a name="line-13644"></a>attr {
+<a name="line-13645"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-13646"></a>  description: "The type of padding algorithm to use."
+<a name="line-13647"></a>  name: "padding"
+<a name="line-13648"></a>  type: "string"
+<a name="line-13649"></a>}
+<a name="line-13650"></a>attr {
+<a name="line-13651"></a>  allowed_values {
+<a name="line-13652"></a>    list {
+<a name="line-13653"></a>      type: DT_FLOAT
+<a name="line-13654"></a>      type: DT_DOUBLE
+<a name="line-13655"></a>      type: DT_INT64
+<a name="line-13656"></a>      type: DT_INT32
+<a name="line-13657"></a>      type: DT_UINT8
+<a name="line-13658"></a>      type: DT_UINT16
+<a name="line-13659"></a>      type: DT_INT16
+<a name="line-13660"></a>      type: DT_INT8
+<a name="line-13661"></a>      type: DT_COMPLEX64
+<a name="line-13662"></a>      type: DT_COMPLEX128
+<a name="line-13663"></a>      type: DT_QINT8
+<a name="line-13664"></a>      type: DT_QUINT8
+<a name="line-13665"></a>      type: DT_QINT32
+<a name="line-13666"></a>      type: DT_HALF
+<a name="line-13667"></a>    }
+<a name="line-13668"></a>  }
+<a name="line-13669"></a>  name: "T"
+<a name="line-13670"></a>  type: "type"
+<a name="line-13671"></a>}
+<a name="line-13672"></a>input_arg {
+<a name="line-13673"></a>  description: "Shape `[batch, depth, rows, cols, channels]` tensor to pool over."
+<a name="line-13674"></a>  name: "input"
+<a name="line-13675"></a>  type_attr: "T"
+<a name="line-13676"></a>}
+<a name="line-13677"></a>output_arg {
+<a name="line-13678"></a>  description: "The average pooled output tensor."
+<a name="line-13679"></a>  name: "output"
+<a name="line-13680"></a>  type_attr: "T"
+<a name="line-13681"></a>}
+<a name="line-13682"></a>-}</span>
+<a name="line-13683"></a>
+<a name="line-13684"></a><a name="inv"></a><span class='hs-comment'>-- | Computes the reciprocal of x element-wise.</span>
+<a name="line-13685"></a><span class='hs-comment'>--</span>
+<a name="line-13686"></a><span class='hs-comment'>-- I.e., \\(y = 1 / x\\).</span>
+<a name="line-13687"></a><span class='hs-definition'>inv</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-13688"></a>                                            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-13689"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-13690"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-13691"></a>                                            <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-13692"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-13693"></a><span class='hs-definition'>inv</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13694"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Inv"</span>
+<a name="line-13695"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13696"></a>        <span class='hs-varid'>x</span>
+<a name="line-13697"></a><span class='hs-comment'>{-
+<a name="line-13698"></a>attr {
+<a name="line-13699"></a>  allowed_values {
+<a name="line-13700"></a>    list {
+<a name="line-13701"></a>      type: DT_HALF
+<a name="line-13702"></a>      type: DT_FLOAT
+<a name="line-13703"></a>      type: DT_DOUBLE
+<a name="line-13704"></a>      type: DT_INT32
+<a name="line-13705"></a>      type: DT_INT64
+<a name="line-13706"></a>      type: DT_COMPLEX64
+<a name="line-13707"></a>      type: DT_COMPLEX128
+<a name="line-13708"></a>    }
+<a name="line-13709"></a>  }
+<a name="line-13710"></a>  name: "T"
+<a name="line-13711"></a>  type: "type"
+<a name="line-13712"></a>}
+<a name="line-13713"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-13714"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-13715"></a>-}</span>
+<a name="line-13716"></a>
+<a name="line-13717"></a><span class='hs-comment'>-- | Pop the element at the top of the stack.</span>
+<a name="line-13718"></a>
+<a name="line-13719"></a><a name="stackPop"></a><span class='hs-definition'>stackPop</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>elem_type</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>elem_type</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13720"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a stack.</span>
+<a name="line-13721"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>elem_type</span> <span class='hs-comment'>-- ^ __elem__: The tensor that is popped from the top of the stack.</span>
+<a name="line-13722"></a><span class='hs-definition'>stackPop</span> <span class='hs-varid'>handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13723"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"StackPop"</span>
+<a name="line-13724"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"elem_type"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>elem_type</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13725"></a>        <span class='hs-varid'>handle</span>
+<a name="line-13726"></a><span class='hs-comment'>{-
+<a name="line-13727"></a>attr {
+<a name="line-13728"></a>  description: "The type of the elem that is popped."
+<a name="line-13729"></a>  name: "elem_type"
+<a name="line-13730"></a>  type: "type"
+<a name="line-13731"></a>}
+<a name="line-13732"></a>input_arg {
+<a name="line-13733"></a>  description: "The handle to a stack."
+<a name="line-13734"></a>  is_ref: true
+<a name="line-13735"></a>  name: "handle"
+<a name="line-13736"></a>  type: DT_STRING
+<a name="line-13737"></a>}
+<a name="line-13738"></a>output_arg {
+<a name="line-13739"></a>  description: "The tensor that is popped from the top of the stack."
+<a name="line-13740"></a>  name: "elem"
+<a name="line-13741"></a>  type_attr: "elem_type"
+<a name="line-13742"></a>}
+<a name="line-13743"></a>-}</span>
+<a name="line-13744"></a>
+<a name="line-13745"></a><a name="paddingFIFOQueue"></a><span class='hs-comment'>-- | A queue that produces elements in first-in first-out order.</span>
+<a name="line-13746"></a><span class='hs-comment'>--</span>
+<a name="line-13747"></a><span class='hs-comment'>-- Variable-size shapes are allowed by setting the corresponding shape dimensions</span>
+<a name="line-13748"></a><span class='hs-comment'>-- to 0 in the shape attr.  In this case DequeueMany will pad up to the maximum</span>
+<a name="line-13749"></a><span class='hs-comment'>-- size of any given element in the minibatch.  See below for details.</span>
+<a name="line-13750"></a><span class='hs-definition'>paddingFIFOQueue</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to the queue.</span>
+<a name="line-13751"></a><span class='hs-definition'>paddingFIFOQueue</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13752"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"PaddingFIFOQueue"</span><span class='hs-layout'>)</span>
+<a name="line-13753"></a>        
+<a name="line-13754"></a><span class='hs-comment'>{-
+<a name="line-13755"></a>attr {
+<a name="line-13756"></a>  description: "The type of each component in a value."
+<a name="line-13757"></a>  has_minimum: true
+<a name="line-13758"></a>  minimum: 1
+<a name="line-13759"></a>  name: "component_types"
+<a name="line-13760"></a>  type: "list(type)"
+<a name="line-13761"></a>}
+<a name="line-13762"></a>attr {
+<a name="line-13763"></a>  default_value { list { } }
+<a name="line-13764"></a>  description: "The shape of each component in a value. The length of this attr must\nbe either 0 or the same as the length of component_types.\nShapes of fixed rank but variable size are allowed by setting\nany shape dimension to -1.  In this case, the inputs\' shape may vary along\nthe given dimension, and DequeueMany will pad the given dimension with\nzeros up to the maximum shape of all elements in the given batch.\nIf the length of this attr is 0, different queue elements may have\ndifferent ranks and shapes, but only one element may be dequeued at a time."
+<a name="line-13765"></a>  has_minimum: true
+<a name="line-13766"></a>  name: "shapes"
+<a name="line-13767"></a>  type: "list(shape)"
+<a name="line-13768"></a>}
+<a name="line-13769"></a>attr {
+<a name="line-13770"></a>  default_value { i: -1 }
+<a name="line-13771"></a>  description: "The upper bound on the number of elements in this queue.\nNegative numbers mean no limit."
+<a name="line-13772"></a>  name: "capacity"
+<a name="line-13773"></a>  type: "int"
+<a name="line-13774"></a>}
+<a name="line-13775"></a>attr {
+<a name="line-13776"></a>  default_value { s: "" }
+<a name="line-13777"></a>  description: "If non-empty, this queue is placed in the given container.\nOtherwise, a default container is used."
+<a name="line-13778"></a>  name: "container"
+<a name="line-13779"></a>  type: "string"
+<a name="line-13780"></a>}
+<a name="line-13781"></a>attr {
+<a name="line-13782"></a>  default_value { s: "" }
+<a name="line-13783"></a>  description: "If non-empty, this queue will be shared under the given name\nacross multiple sessions."
+<a name="line-13784"></a>  name: "shared_name"
+<a name="line-13785"></a>  type: "string"
+<a name="line-13786"></a>}
+<a name="line-13787"></a>output_arg {
+<a name="line-13788"></a>  description: "The handle to the queue."
+<a name="line-13789"></a>  is_ref: true
+<a name="line-13790"></a>  name: "handle"
+<a name="line-13791"></a>  type: DT_STRING
+<a name="line-13792"></a>}
+<a name="line-13793"></a>-}</span>
+<a name="line-13794"></a>
+<a name="line-13795"></a><span class='hs-comment'>-- | </span>
+<a name="line-13796"></a>
+<a name="line-13797"></a><a name="batchSelfAdjointEigV2"></a><span class='hs-definition'>batchSelfAdjointEigV2</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-13798"></a>                                                              <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13799"></a>                         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-13800"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-13801"></a>                         <span class='hs-comment'>-- ^ (__e__, __v__)</span>
+<a name="line-13802"></a>                         <span class='hs-comment'>--</span>
+<a name="line-13803"></a>                         <span class='hs-comment'>-- * __e__</span>
+<a name="line-13804"></a>                         <span class='hs-comment'>--</span>
+<a name="line-13805"></a>                         <span class='hs-comment'>-- * __v__</span>
+<a name="line-13806"></a><span class='hs-definition'>batchSelfAdjointEigV2</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13807"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchSelfAdjointEigV2"</span>
+<a name="line-13808"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13809"></a>        <span class='hs-varid'>input</span>
+<a name="line-13810"></a><span class='hs-comment'>{-
+<a name="line-13811"></a>attr { default_value { b: true } name: "compute_v" type: "bool" }
+<a name="line-13812"></a>attr {
+<a name="line-13813"></a>  allowed_values { list { type: DT_DOUBLE type: DT_FLOAT } }
+<a name="line-13814"></a>  name: "T"
+<a name="line-13815"></a>  type: "type"
+<a name="line-13816"></a>}
+<a name="line-13817"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-13818"></a>output_arg { name: "e" type_attr: "T" }
+<a name="line-13819"></a>output_arg { name: "v" type_attr: "T" }
+<a name="line-13820"></a>-}</span>
+<a name="line-13821"></a>
+<a name="line-13822"></a><span class='hs-comment'>-- | </span>
+<a name="line-13823"></a>
+<a name="line-13824"></a><a name="batchMatrixTriangularSolve"></a><span class='hs-definition'>batchMatrixTriangularSolve</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-13825"></a>                                                                      <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13826"></a>                              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __matrix__</span>
+<a name="line-13827"></a>                              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __rhs__</span>
+<a name="line-13828"></a>                              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-13829"></a><span class='hs-definition'>batchMatrixTriangularSolve</span> <span class='hs-varid'>matrix</span> <span class='hs-varid'>rhs</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13830"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchMatrixTriangularSolve"</span>
+<a name="line-13831"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13832"></a>        <span class='hs-varid'>matrix</span> <span class='hs-varid'>rhs</span>
+<a name="line-13833"></a><span class='hs-comment'>{-
+<a name="line-13834"></a>attr { default_value { b: true } name: "lower" type: "bool" }
+<a name="line-13835"></a>attr { default_value { b: false } name: "adjoint" type: "bool" }
+<a name="line-13836"></a>attr {
+<a name="line-13837"></a>  allowed_values { list { type: DT_DOUBLE type: DT_FLOAT } }
+<a name="line-13838"></a>  name: "T"
+<a name="line-13839"></a>  type: "type"
+<a name="line-13840"></a>}
+<a name="line-13841"></a>input_arg { name: "matrix" type_attr: "T" }
+<a name="line-13842"></a>input_arg { name: "rhs" type_attr: "T" }
+<a name="line-13843"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-13844"></a>-}</span>
+<a name="line-13845"></a>
+<a name="line-13846"></a><span class='hs-comment'>-- | </span>
+<a name="line-13847"></a>
+<a name="line-13848"></a><a name="batchMatrixSolveLs"></a><span class='hs-definition'>batchMatrixSolveLs</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-13849"></a>                                                                 <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13850"></a>                      <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __matrix__</span>
+<a name="line-13851"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __rhs__</span>
+<a name="line-13852"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Double</span> <span class='hs-comment'>-- ^ __l2_regularizer__</span>
+<a name="line-13853"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-13854"></a><span class='hs-definition'>batchMatrixSolveLs</span> <span class='hs-varid'>matrix</span> <span class='hs-varid'>rhs</span> <span class='hs-varid'>l2_regularizer</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13855"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchMatrixSolveLs"</span>
+<a name="line-13856"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13857"></a>        <span class='hs-varid'>matrix</span> <span class='hs-varid'>rhs</span> <span class='hs-varid'>l2_regularizer</span>
+<a name="line-13858"></a><span class='hs-comment'>{-
+<a name="line-13859"></a>attr {
+<a name="line-13860"></a>  allowed_values { list { type: DT_DOUBLE type: DT_FLOAT } }
+<a name="line-13861"></a>  name: "T"
+<a name="line-13862"></a>  type: "type"
+<a name="line-13863"></a>}
+<a name="line-13864"></a>attr { default_value { b: true } name: "fast" type: "bool" }
+<a name="line-13865"></a>input_arg { name: "matrix" type_attr: "T" }
+<a name="line-13866"></a>input_arg { name: "rhs" type_attr: "T" }
+<a name="line-13867"></a>input_arg { name: "l2_regularizer" type: DT_DOUBLE }
+<a name="line-13868"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-13869"></a>-}</span>
+<a name="line-13870"></a>
+<a name="line-13871"></a><span class='hs-comment'>-- | </span>
+<a name="line-13872"></a>
+<a name="line-13873"></a><a name="batchSvd"></a><span class='hs-definition'>batchSvd</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-13874"></a>                                                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-13875"></a>                                                 <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13876"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-13877"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-13878"></a>            <span class='hs-comment'>-- ^ (__s__, __u__, __v__)</span>
+<a name="line-13879"></a>            <span class='hs-comment'>--</span>
+<a name="line-13880"></a>            <span class='hs-comment'>-- * __s__</span>
+<a name="line-13881"></a>            <span class='hs-comment'>--</span>
+<a name="line-13882"></a>            <span class='hs-comment'>-- * __u__</span>
+<a name="line-13883"></a>            <span class='hs-comment'>--</span>
+<a name="line-13884"></a>            <span class='hs-comment'>-- * __v__</span>
+<a name="line-13885"></a><span class='hs-definition'>batchSvd</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13886"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchSvd"</span>
+<a name="line-13887"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13888"></a>        <span class='hs-varid'>input</span>
+<a name="line-13889"></a><span class='hs-comment'>{-
+<a name="line-13890"></a>attr { default_value { b: true } name: "compute_uv" type: "bool" }
+<a name="line-13891"></a>attr {
+<a name="line-13892"></a>  default_value { b: false } name: "full_matrices" type: "bool"
+<a name="line-13893"></a>}
+<a name="line-13894"></a>attr {
+<a name="line-13895"></a>  allowed_values {
+<a name="line-13896"></a>    list {
+<a name="line-13897"></a>      type: DT_DOUBLE
+<a name="line-13898"></a>      type: DT_FLOAT
+<a name="line-13899"></a>      type: DT_COMPLEX64
+<a name="line-13900"></a>      type: DT_COMPLEX128
+<a name="line-13901"></a>    }
+<a name="line-13902"></a>  }
+<a name="line-13903"></a>  name: "T"
+<a name="line-13904"></a>  type: "type"
+<a name="line-13905"></a>}
+<a name="line-13906"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-13907"></a>output_arg { name: "s" type_attr: "T" }
+<a name="line-13908"></a>output_arg { name: "u" type_attr: "T" }
+<a name="line-13909"></a>output_arg { name: "v" type_attr: "T" }
+<a name="line-13910"></a>-}</span>
+<a name="line-13911"></a>
+<a name="line-13912"></a><span class='hs-comment'>-- | Outputs a `Summary` protocol buffer with a tensor.</span>
+<a name="line-13913"></a>
+<a name="line-13914"></a><a name="tensorSummary"></a><span class='hs-definition'>tensorSummary</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13915"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __tensor__: A tensor to serialize.</span>
+<a name="line-13916"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __summary__</span>
+<a name="line-13917"></a><span class='hs-definition'>tensorSummary</span> <span class='hs-varid'>tensor</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13918"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TensorSummary"</span>
+<a name="line-13919"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13920"></a>        <span class='hs-varid'>tensor</span>
+<a name="line-13921"></a><span class='hs-comment'>{-
+<a name="line-13922"></a>attr { name: "T" type: "type" }
+<a name="line-13923"></a>attr {
+<a name="line-13924"></a>  description: "A name to associate with the data series."
+<a name="line-13925"></a>  name: "display_name"
+<a name="line-13926"></a>  type: "string"
+<a name="line-13927"></a>}
+<a name="line-13928"></a>attr {
+<a name="line-13929"></a>  default_value { s: "" }
+<a name="line-13930"></a>  description: "An optional long description of the data being output."
+<a name="line-13931"></a>  name: "description"
+<a name="line-13932"></a>  type: "string"
+<a name="line-13933"></a>}
+<a name="line-13934"></a>attr {
+<a name="line-13935"></a>  default_value { list { } }
+<a name="line-13936"></a>  description: "a list of strings used to specify how the data can be interpreted, e.g.\na string tensor containing jpg images should have \'encoding:image/jpg\'; a\nstring tensor with foo protos should have \'encoding:proto/X/Y/foo.proto\';\na numeric tensor containing bounding boxes may have\n\'bounding_box:x1,y1,x2,y2,\'. If the tensor is a part of a group of related\noutputs, that can be encoded through a \'group:$groupName/$roleInGroup\' label.\nLabels may be formatted as \'prefix:value\'. The prefix may be re-used."
+<a name="line-13937"></a>  name: "labels"
+<a name="line-13938"></a>  type: "list(string)"
+<a name="line-13939"></a>}
+<a name="line-13940"></a>input_arg {
+<a name="line-13941"></a>  description: "A tensor to serialize." name: "tensor" type_attr: "T"
+<a name="line-13942"></a>}
+<a name="line-13943"></a>output_arg { name: "summary" type: DT_STRING }
+<a name="line-13944"></a>-}</span>
+<a name="line-13945"></a>
+<a name="line-13946"></a><a name="sparseSoftmaxCrossEntropyWithLogits"></a><span class='hs-comment'>-- | Computes softmax cross entropy cost and gradients to backpropagate.</span>
+<a name="line-13947"></a><span class='hs-comment'>--</span>
+<a name="line-13948"></a><span class='hs-comment'>-- Unlike `SoftmaxCrossEntropyWithLogits`, this operation does not accept</span>
+<a name="line-13949"></a><span class='hs-comment'>-- a matrix of label probabilities, but rather a single label per row</span>
+<a name="line-13950"></a><span class='hs-comment'>-- of features.  This label is considered to have probability 1.0 for the</span>
+<a name="line-13951"></a><span class='hs-comment'>-- given row.</span>
+<a name="line-13952"></a><span class='hs-comment'>-- </span>
+<a name="line-13953"></a><span class='hs-comment'>-- Inputs are the logits, not probabilities.</span>
+<a name="line-13954"></a><span class='hs-definition'>sparseSoftmaxCrossEntropyWithLogits</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tlabels</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-13955"></a>                                                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-13956"></a>                                                                         <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-13957"></a>                                                                         <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-13958"></a>                                                                 <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tlabels</span><span class='hs-layout'>,</span>
+<a name="line-13959"></a>                                                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-13960"></a>                                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tlabels</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-13961"></a>                                       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __features__: batch_size x num_classes matrix</span>
+<a name="line-13962"></a>                                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tlabels</span> <span class='hs-comment'>-- ^ __labels__: batch_size vector with values in [0, num_classes).</span>
+<a name="line-13963"></a>                                                            <span class='hs-comment'>-- This is the label for the given minibatch entry.</span>
+<a name="line-13964"></a>                                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-13965"></a>                                       <span class='hs-comment'>-- ^ (__loss__, __backprop__)</span>
+<a name="line-13966"></a>                                       <span class='hs-comment'>--</span>
+<a name="line-13967"></a>                                       <span class='hs-comment'>-- * __loss__: Per example loss (batch_size vector).</span>
+<a name="line-13968"></a>                                       <span class='hs-comment'>--</span>
+<a name="line-13969"></a>                                       <span class='hs-comment'>-- * __backprop__: backpropagated gradients (batch_size x num_classes matrix).</span>
+<a name="line-13970"></a><span class='hs-definition'>sparseSoftmaxCrossEntropyWithLogits</span> <span class='hs-varid'>features</span> <span class='hs-varid'>labels</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-13971"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseSoftmaxCrossEntropyWithLogits"</span>
+<a name="line-13972"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-13973"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tlabels"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tlabels</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-13974"></a>        <span class='hs-varid'>features</span> <span class='hs-varid'>labels</span>
+<a name="line-13975"></a><span class='hs-comment'>{-
+<a name="line-13976"></a>attr {
+<a name="line-13977"></a>  allowed_values {
+<a name="line-13978"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-13979"></a>  }
+<a name="line-13980"></a>  name: "T"
+<a name="line-13981"></a>  type: "type"
+<a name="line-13982"></a>}
+<a name="line-13983"></a>attr {
+<a name="line-13984"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-13985"></a>  default_value { type: DT_INT64 }
+<a name="line-13986"></a>  name: "Tlabels"
+<a name="line-13987"></a>  type: "type"
+<a name="line-13988"></a>}
+<a name="line-13989"></a>input_arg {
+<a name="line-13990"></a>  description: "batch_size x num_classes matrix"
+<a name="line-13991"></a>  name: "features"
+<a name="line-13992"></a>  type_attr: "T"
+<a name="line-13993"></a>}
+<a name="line-13994"></a>input_arg {
+<a name="line-13995"></a>  description: "batch_size vector with values in [0, num_classes).\nThis is the label for the given minibatch entry."
+<a name="line-13996"></a>  name: "labels"
+<a name="line-13997"></a>  type_attr: "Tlabels"
+<a name="line-13998"></a>}
+<a name="line-13999"></a>output_arg {
+<a name="line-14000"></a>  description: "Per example loss (batch_size vector)."
+<a name="line-14001"></a>  name: "loss"
+<a name="line-14002"></a>  type_attr: "T"
+<a name="line-14003"></a>}
+<a name="line-14004"></a>output_arg {
+<a name="line-14005"></a>  description: "backpropagated gradients (batch_size x num_classes matrix)."
+<a name="line-14006"></a>  name: "backprop"
+<a name="line-14007"></a>  type_attr: "T"
+<a name="line-14008"></a>}
+<a name="line-14009"></a>-}</span>
+<a name="line-14010"></a>
+<a name="line-14011"></a><a name="maxPoolWithArgmax"></a><span class='hs-comment'>-- | Performs max pooling on the input and outputs both max values and indices.</span>
+<a name="line-14012"></a><span class='hs-comment'>--</span>
+<a name="line-14013"></a><span class='hs-comment'>-- The indices in `argmax` are flattened, so that a maximum value at position</span>
+<a name="line-14014"></a><span class='hs-comment'>-- `[b, y, x, c]` becomes flattened index</span>
+<a name="line-14015"></a><span class='hs-comment'>-- `((b * height + y) * width + x) * channels + c`.</span>
+<a name="line-14016"></a><span class='hs-definition'>maxPoolWithArgmax</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varid'>targmax</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-14017"></a>                                            <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-14018"></a>                                            <span class='hs-conid'>TensorType</span> <span class='hs-varid'>targmax</span><span class='hs-layout'>,</span>
+<a name="line-14019"></a>                                            <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-14020"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>targmax</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-14021"></a>                     <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: 4-D with shape `[batch, height, width, channels]`.  Input to pool over.</span>
+<a name="line-14022"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>targmax</span><span class='hs-layout'>)</span>
+<a name="line-14023"></a>                     <span class='hs-comment'>-- ^ (__output__, __argmax__)</span>
+<a name="line-14024"></a>                     <span class='hs-comment'>--</span>
+<a name="line-14025"></a>                     <span class='hs-comment'>-- * __output__: The max pooled output tensor.</span>
+<a name="line-14026"></a>                     <span class='hs-comment'>--</span>
+<a name="line-14027"></a>                     <span class='hs-comment'>-- * __argmax__: 4-D.  The flattened indices of the max values chosen for each output.</span>
+<a name="line-14028"></a><span class='hs-definition'>maxPoolWithArgmax</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-14029"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MaxPoolWithArgmax"</span>
+<a name="line-14030"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-14031"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Targmax"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>targmax</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-14032"></a>        <span class='hs-varid'>input</span>
+<a name="line-14033"></a><span class='hs-comment'>{-
+<a name="line-14034"></a>attr {
+<a name="line-14035"></a>  description: "The size of the window for each dimension of the input tensor."
+<a name="line-14036"></a>  has_minimum: true
+<a name="line-14037"></a>  minimum: 4
+<a name="line-14038"></a>  name: "ksize"
+<a name="line-14039"></a>  type: "list(int)"
+<a name="line-14040"></a>}
+<a name="line-14041"></a>attr {
+<a name="line-14042"></a>  description: "The stride of the sliding window for each dimension of the\ninput tensor."
+<a name="line-14043"></a>  has_minimum: true
+<a name="line-14044"></a>  minimum: 4
+<a name="line-14045"></a>  name: "strides"
+<a name="line-14046"></a>  type: "list(int)"
+<a name="line-14047"></a>}
+<a name="line-14048"></a>attr {
+<a name="line-14049"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-14050"></a>  default_value { type: DT_INT64 }
+<a name="line-14051"></a>  name: "Targmax"
+<a name="line-14052"></a>  type: "type"
+<a name="line-14053"></a>}
+<a name="line-14054"></a>attr {
+<a name="line-14055"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-14056"></a>  description: "The type of padding algorithm to use."
+<a name="line-14057"></a>  name: "padding"
+<a name="line-14058"></a>  type: "string"
+<a name="line-14059"></a>}
+<a name="line-14060"></a>attr {
+<a name="line-14061"></a>  allowed_values { list { type: DT_FLOAT type: DT_HALF } }
+<a name="line-14062"></a>  default_value { type: DT_FLOAT }
+<a name="line-14063"></a>  name: "T"
+<a name="line-14064"></a>  type: "type"
+<a name="line-14065"></a>}
+<a name="line-14066"></a>input_arg {
+<a name="line-14067"></a>  description: "4-D with shape `[batch, height, width, channels]`.  Input to pool over."
+<a name="line-14068"></a>  name: "input"
+<a name="line-14069"></a>  type_attr: "T"
+<a name="line-14070"></a>}
+<a name="line-14071"></a>output_arg {
+<a name="line-14072"></a>  description: "The max pooled output tensor."
+<a name="line-14073"></a>  name: "output"
+<a name="line-14074"></a>  type_attr: "T"
+<a name="line-14075"></a>}
+<a name="line-14076"></a>output_arg {
+<a name="line-14077"></a>  description: "4-D.  The flattened indices of the max values chosen for each output."
+<a name="line-14078"></a>  name: "argmax"
+<a name="line-14079"></a>  type_attr: "Targmax"
+<a name="line-14080"></a>}
+<a name="line-14081"></a>-}</span>
+<a name="line-14082"></a>
+<a name="line-14083"></a><a name="fFT"></a><span class='hs-comment'>-- | Compute the 1-dimensional discrete Fourier Transform over the inner-most</span>
+<a name="line-14084"></a><span class='hs-comment'>--</span>
+<a name="line-14085"></a><span class='hs-comment'>-- dimension of `input`.</span>
+<a name="line-14086"></a><span class='hs-definition'>fFT</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __input__: A complex64 tensor.</span>
+<a name="line-14087"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __output__: A complex64 tensor of the same shape as `input`. The inner-most</span>
+<a name="line-14088"></a>       <span class='hs-comment'>-- dimension of `input` is replaced with its 1D Fourier Transform.</span>
+<a name="line-14089"></a><span class='hs-definition'>fFT</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-14090"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"FFT"</span><span class='hs-layout'>)</span>
+<a name="line-14091"></a>        <span class='hs-varid'>input</span>
+<a name="line-14092"></a><span class='hs-comment'>{-
+<a name="line-14093"></a>input_arg {
+<a name="line-14094"></a>  description: "A complex64 tensor." name: "input" type: DT_COMPLEX64
+<a name="line-14095"></a>}
+<a name="line-14096"></a>output_arg {
+<a name="line-14097"></a>  description: "A complex64 tensor of the same shape as `input`. The inner-most\ndimension of `input` is replaced with its 1D Fourier Transform."
+<a name="line-14098"></a>  name: "output"
+<a name="line-14099"></a>  type: DT_COMPLEX64
+<a name="line-14100"></a>}
+<a name="line-14101"></a>-}</span>
+<a name="line-14102"></a>
+<a name="line-14103"></a><a name="histogramSummary"></a><span class='hs-comment'>-- | Outputs a `Summary` protocol buffer with a histogram.</span>
+<a name="line-14104"></a><span class='hs-comment'>--</span>
+<a name="line-14105"></a><span class='hs-comment'>-- The generated</span>
+<a name="line-14106"></a><span class='hs-comment'>-- [`Summary`](https://www.tensorflow.org/code/tensorflow/core/framework/summary.proto)</span>
+<a name="line-14107"></a><span class='hs-comment'>-- has one summary value containing a histogram for `values`.</span>
+<a name="line-14108"></a><span class='hs-comment'>-- </span>
+<a name="line-14109"></a><span class='hs-comment'>-- This op reports an `InvalidArgument` error if any value is not finite.</span>
+<a name="line-14110"></a><span class='hs-definition'>histogramSummary</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-14111"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-14112"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-14113"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-14114"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-14115"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-14116"></a>                                                            <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-14117"></a>                    <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __tag__: Scalar.  Tag to use for the `Summary.Value`.</span>
+<a name="line-14118"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __values__: Any shape. Values to use to build the histogram.</span>
+<a name="line-14119"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __summary__: Scalar. Serialized `Summary` protocol buffer.</span>
+<a name="line-14120"></a><span class='hs-definition'>histogramSummary</span> <span class='hs-varid'>tag</span> <span class='hs-varid'>values</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-14121"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"HistogramSummary"</span>
+<a name="line-14122"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-14123"></a>        <span class='hs-varid'>tag</span> <span class='hs-varid'>values</span>
+<a name="line-14124"></a><span class='hs-comment'>{-
+<a name="line-14125"></a>attr {
+<a name="line-14126"></a>  allowed_values {
+<a name="line-14127"></a>    list {
+<a name="line-14128"></a>      type: DT_FLOAT
+<a name="line-14129"></a>      type: DT_DOUBLE
+<a name="line-14130"></a>      type: DT_INT32
+<a name="line-14131"></a>      type: DT_INT64
+<a name="line-14132"></a>      type: DT_UINT8
+<a name="line-14133"></a>      type: DT_INT16
+<a name="line-14134"></a>      type: DT_INT8
+<a name="line-14135"></a>      type: DT_UINT16
+<a name="line-14136"></a>      type: DT_HALF
+<a name="line-14137"></a>    }
+<a name="line-14138"></a>  }
+<a name="line-14139"></a>  default_value { type: DT_FLOAT }
+<a name="line-14140"></a>  name: "T"
+<a name="line-14141"></a>  type: "type"
+<a name="line-14142"></a>}
+<a name="line-14143"></a>input_arg {
+<a name="line-14144"></a>  description: "Scalar.  Tag to use for the `Summary.Value`."
+<a name="line-14145"></a>  name: "tag"
+<a name="line-14146"></a>  type: DT_STRING
+<a name="line-14147"></a>}
+<a name="line-14148"></a>input_arg {
+<a name="line-14149"></a>  description: "Any shape. Values to use to build the histogram."
+<a name="line-14150"></a>  name: "values"
+<a name="line-14151"></a>  type_attr: "T"
+<a name="line-14152"></a>}
+<a name="line-14153"></a>output_arg {
+<a name="line-14154"></a>  description: "Scalar. Serialized `Summary` protocol buffer."
+<a name="line-14155"></a>  name: "summary"
+<a name="line-14156"></a>  type: DT_STRING
+<a name="line-14157"></a>}
+<a name="line-14158"></a>-}</span>
+<a name="line-14159"></a>
+<a name="line-14160"></a><a name="pad"></a><span class='hs-comment'>-- | Pads a tensor with zeros.</span>
+<a name="line-14161"></a><span class='hs-comment'>--</span>
+<a name="line-14162"></a><span class='hs-comment'>-- This operation pads a `input` with zeros according to the `paddings` you</span>
+<a name="line-14163"></a><span class='hs-comment'>-- specify. `paddings` is an integer tensor with shape `[Dn, 2]`, where n is the</span>
+<a name="line-14164"></a><span class='hs-comment'>-- rank of `input`. For each dimension D of `input`, `paddings[D, 0]` indicates</span>
+<a name="line-14165"></a><span class='hs-comment'>-- how many zeros to add before the contents of `input` in that dimension, and</span>
+<a name="line-14166"></a><span class='hs-comment'>-- `paddings[D, 1]` indicates how many zeros to add after the contents of `input`</span>
+<a name="line-14167"></a><span class='hs-comment'>-- in that dimension.</span>
+<a name="line-14168"></a><span class='hs-comment'>-- </span>
+<a name="line-14169"></a><span class='hs-comment'>-- The padded size of each dimension D of the output is:</span>
+<a name="line-14170"></a><span class='hs-comment'>-- </span>
+<a name="line-14171"></a><span class='hs-comment'>-- `paddings(D, 0) + input.dim_size(D) + paddings(D, 1)`</span>
+<a name="line-14172"></a><span class='hs-comment'>-- </span>
+<a name="line-14173"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-14174"></a><span class='hs-comment'>-- </span>
+<a name="line-14175"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-14176"></a><span class='hs-comment'>-- # 't' is [[1, 1], [2, 2]]</span>
+<a name="line-14177"></a><span class='hs-comment'>-- # 'paddings' is [[1, 1], [2, 2]]</span>
+<a name="line-14178"></a><span class='hs-comment'>-- # rank of 't' is 2</span>
+<a name="line-14179"></a><span class='hs-comment'>-- pad(t, paddings) ==&gt; [[0, 0, 0, 0, 0, 0]</span>
+<a name="line-14180"></a><span class='hs-comment'>--                       [0, 0, 1, 1, 0, 0]</span>
+<a name="line-14181"></a><span class='hs-comment'>--                       [0, 0, 2, 2, 0, 0]</span>
+<a name="line-14182"></a><span class='hs-comment'>--                       [0, 0, 0, 0, 0, 0]]</span>
+<a name="line-14183"></a><span class='hs-comment'>-- ```</span>
+<a name="line-14184"></a><span class='hs-definition'>pad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tpaddings</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tpaddings</span><span class='hs-layout'>,</span>
+<a name="line-14185"></a>                                   <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-14186"></a>                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tpaddings</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-14187"></a>       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-14188"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tpaddings</span> <span class='hs-comment'>-- ^ __paddings__</span>
+<a name="line-14189"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-14190"></a><span class='hs-definition'>pad</span> <span class='hs-varid'>input</span> <span class='hs-varid'>paddings</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-14191"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Pad"</span>
+<a name="line-14192"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-14193"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tpaddings"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tpaddings</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-14194"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>paddings</span>
+<a name="line-14195"></a><span class='hs-comment'>{-
+<a name="line-14196"></a>attr { name: "T" type: "type" }
+<a name="line-14197"></a>attr {
+<a name="line-14198"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-14199"></a>  default_value { type: DT_INT32 }
+<a name="line-14200"></a>  name: "Tpaddings"
+<a name="line-14201"></a>  type: "type"
+<a name="line-14202"></a>}
+<a name="line-14203"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-14204"></a>input_arg { name: "paddings" type_attr: "Tpaddings" }
+<a name="line-14205"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-14206"></a>-}</span>
+<a name="line-14207"></a>
+<a name="line-14208"></a><span class='hs-comment'>-- | </span>
+<a name="line-14209"></a>
+<a name="line-14210"></a><a name="batchIFFT3D"></a><span class='hs-definition'>batchIFFT3D</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-14211"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-14212"></a><span class='hs-definition'>batchIFFT3D</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-14213"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchIFFT3D"</span><span class='hs-layout'>)</span>
+<a name="line-14214"></a>        <span class='hs-varid'>input</span>
+<a name="line-14215"></a><span class='hs-comment'>{-
+<a name="line-14216"></a>input_arg { name: "input" type: DT_COMPLEX64 }
+<a name="line-14217"></a>output_arg { name: "output" type: DT_COMPLEX64 }
+<a name="line-14218"></a>-}</span>
+<a name="line-14219"></a>
+<a name="line-14220"></a><a name="imageSummary"></a><span class='hs-comment'>-- | Outputs a `Summary` protocol buffer with images.</span>
+<a name="line-14221"></a><span class='hs-comment'>--</span>
+<a name="line-14222"></a><span class='hs-comment'>-- The summary has up to `max_images` summary values containing images. The</span>
+<a name="line-14223"></a><span class='hs-comment'>-- images are built from `tensor` which must be 4-D with shape `[batch_size,</span>
+<a name="line-14224"></a><span class='hs-comment'>-- height, width, channels]` and where `channels` can be:</span>
+<a name="line-14225"></a><span class='hs-comment'>-- </span>
+<a name="line-14226"></a><span class='hs-comment'>-- *  1: `tensor` is interpreted as Grayscale.</span>
+<a name="line-14227"></a><span class='hs-comment'>-- *  3: `tensor` is interpreted as RGB.</span>
+<a name="line-14228"></a><span class='hs-comment'>-- *  4: `tensor` is interpreted as RGBA.</span>
+<a name="line-14229"></a><span class='hs-comment'>-- </span>
+<a name="line-14230"></a><span class='hs-comment'>-- The images have the same number of channels as the input tensor. For float</span>
+<a name="line-14231"></a><span class='hs-comment'>-- input, the values are normalized one image at a time to fit in the range</span>
+<a name="line-14232"></a><span class='hs-comment'>-- `[0, 255]`.  `uint8` values are unchanged.  The op uses two different</span>
+<a name="line-14233"></a><span class='hs-comment'>-- normalization algorithms:</span>
+<a name="line-14234"></a><span class='hs-comment'>-- </span>
+<a name="line-14235"></a><span class='hs-comment'>-- *  If the input values are all positive, they are rescaled so the largest one</span>
+<a name="line-14236"></a><span class='hs-comment'>--    is 255.</span>
+<a name="line-14237"></a><span class='hs-comment'>-- </span>
+<a name="line-14238"></a><span class='hs-comment'>-- *  If any input value is negative, the values are shifted so input value 0.0</span>
+<a name="line-14239"></a><span class='hs-comment'>--    is at 127.  They are then rescaled so that either the smallest value is 0,</span>
+<a name="line-14240"></a><span class='hs-comment'>--    or the largest one is 255.</span>
+<a name="line-14241"></a><span class='hs-comment'>-- </span>
+<a name="line-14242"></a><span class='hs-comment'>-- The `tag` argument is a scalar `Tensor` of type `string`.  It is used to</span>
+<a name="line-14243"></a><span class='hs-comment'>-- build the `tag` of the summary values:</span>
+<a name="line-14244"></a><span class='hs-comment'>-- </span>
+<a name="line-14245"></a><span class='hs-comment'>-- *  If `max_images` is 1, the summary value tag is '*tag*/image'.</span>
+<a name="line-14246"></a><span class='hs-comment'>-- *  If `max_images` is greater than 1, the summary value tags are</span>
+<a name="line-14247"></a><span class='hs-comment'>--    generated sequentially as '*tag*/image/0', '*tag*/image/1', etc.</span>
+<a name="line-14248"></a><span class='hs-comment'>-- </span>
+<a name="line-14249"></a><span class='hs-comment'>-- The `bad_color` argument is the color to use in the generated images for</span>
+<a name="line-14250"></a><span class='hs-comment'>-- non-finite input values.  It is a `unit8` 1-D tensor of length `channels`.</span>
+<a name="line-14251"></a><span class='hs-comment'>-- Each element must be in the range `[0, 255]` (It represents the value of a</span>
+<a name="line-14252"></a><span class='hs-comment'>-- pixel in the output image).  Non-finite values in the input tensor are</span>
+<a name="line-14253"></a><span class='hs-comment'>-- replaced by this tensor in the output image.  The default value is the color</span>
+<a name="line-14254"></a><span class='hs-comment'>-- red.</span>
+<a name="line-14255"></a><span class='hs-definition'>imageSummary</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-14256"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-14257"></a>                                                        <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-14258"></a>                <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __tag__: Scalar. Used to build the `tag` attribute of the summary values.</span>
+<a name="line-14259"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __tensor__: 4-D of shape `[batch_size, height, width, channels]` where</span>
+<a name="line-14260"></a>                               <span class='hs-comment'>-- `channels` is 1, 3, or 4.</span>
+<a name="line-14261"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __summary__: Scalar. Serialized `Summary` protocol buffer.</span>
+<a name="line-14262"></a><span class='hs-definition'>imageSummary</span> <span class='hs-varid'>tag</span> <span class='hs-varid'>tensor</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-14263"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ImageSummary"</span>
+<a name="line-14264"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-14265"></a>        <span class='hs-varid'>tag</span> <span class='hs-varid'>tensor</span>
+<a name="line-14266"></a><span class='hs-comment'>{-
+<a name="line-14267"></a>attr {
+<a name="line-14268"></a>  default_value { i: 3 }
+<a name="line-14269"></a>  description: "Max number of batch elements to generate images for."
+<a name="line-14270"></a>  has_minimum: true
+<a name="line-14271"></a>  minimum: 1
+<a name="line-14272"></a>  name: "max_images"
+<a name="line-14273"></a>  type: "int"
+<a name="line-14274"></a>}
+<a name="line-14275"></a>attr {
+<a name="line-14276"></a>  allowed_values {
+<a name="line-14277"></a>    list { type: DT_UINT8 type: DT_FLOAT type: DT_HALF }
+<a name="line-14278"></a>  }
+<a name="line-14279"></a>  default_value { type: DT_FLOAT }
+<a name="line-14280"></a>  name: "T"
+<a name="line-14281"></a>  type: "type"
+<a name="line-14282"></a>}
+<a name="line-14283"></a>attr {
+<a name="line-14284"></a>  default_value {
+<a name="line-14285"></a>    tensor {
+<a name="line-14286"></a>      dtype: DT_UINT8
+<a name="line-14287"></a>      int_val: 255
+<a name="line-14288"></a>      int_val: 0
+<a name="line-14289"></a>      int_val: 0
+<a name="line-14290"></a>      int_val: 255
+<a name="line-14291"></a>      tensor_shape { dim { size: 4 } }
+<a name="line-14292"></a>    }
+<a name="line-14293"></a>  }
+<a name="line-14294"></a>  description: "Color to use for pixels with non-finite values."
+<a name="line-14295"></a>  name: "bad_color"
+<a name="line-14296"></a>  type: "tensor"
+<a name="line-14297"></a>}
+<a name="line-14298"></a>input_arg {
+<a name="line-14299"></a>  description: "Scalar. Used to build the `tag` attribute of the summary values."
+<a name="line-14300"></a>  name: "tag"
+<a name="line-14301"></a>  type: DT_STRING
+<a name="line-14302"></a>}
+<a name="line-14303"></a>input_arg {
+<a name="line-14304"></a>  description: "4-D of shape `[batch_size, height, width, channels]` where\n`channels` is 1, 3, or 4."
+<a name="line-14305"></a>  name: "tensor"
+<a name="line-14306"></a>  type_attr: "T"
+<a name="line-14307"></a>}
+<a name="line-14308"></a>output_arg {
+<a name="line-14309"></a>  description: "Scalar. Serialized `Summary` protocol buffer."
+<a name="line-14310"></a>  name: "summary"
+<a name="line-14311"></a>  type: DT_STRING
+<a name="line-14312"></a>}
+<a name="line-14313"></a>-}</span>
+<a name="line-14314"></a>
+<a name="line-14315"></a><a name="segmentSum"></a><span class='hs-comment'>-- | Computes the sum along segments of a tensor.</span>
+<a name="line-14316"></a><span class='hs-comment'>--</span>
+<a name="line-14317"></a><span class='hs-comment'>-- Read [the section on Segmentation](../../api_docs/python/math_ops.md#segmentation)</span>
+<a name="line-14318"></a><span class='hs-comment'>-- for an explanation of segments.</span>
+<a name="line-14319"></a><span class='hs-comment'>-- </span>
+<a name="line-14320"></a><span class='hs-comment'>-- Computes a tensor such that</span>
+<a name="line-14321"></a><span class='hs-comment'>-- \\(output_i = \sum_j data_j\\) where sum is over `j` such</span>
+<a name="line-14322"></a><span class='hs-comment'>-- that `segment_ids[j] == i`.</span>
+<a name="line-14323"></a><span class='hs-comment'>-- </span>
+<a name="line-14324"></a><span class='hs-comment'>-- &lt;div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;"&gt;</span>
+<a name="line-14325"></a><span class='hs-comment'>-- &lt;img style="width:100%" src="../../images/SegmentSum.png" alt&gt;</span>
+<a name="line-14326"></a><span class='hs-comment'>-- &lt;/div&gt;</span>
+<a name="line-14327"></a><span class='hs-definition'>segmentSum</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-14328"></a>                                         <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-14329"></a>                                                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-14330"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-14331"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-14332"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-14333"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-14334"></a>                                                 <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-14335"></a>                                         <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-14336"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-14337"></a>              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__</span>
+<a name="line-14338"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __segment_ids__: A 1-D tensor whose rank is equal to the rank of `data`'s</span>
+<a name="line-14339"></a>                                    <span class='hs-comment'>-- first dimension.  Values should be sorted and can be repeated.</span>
+<a name="line-14340"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Has same shape as data, except for dimension 0 which</span>
+<a name="line-14341"></a>              <span class='hs-comment'>-- has size `k`, the number of segments.</span>
+<a name="line-14342"></a><span class='hs-definition'>segmentSum</span> <span class='hs-varid'>data'</span> <span class='hs-varid'>segment_ids</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-14343"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SegmentSum"</span>
+<a name="line-14344"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-14345"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-14346"></a>        <span class='hs-varid'>data'</span> <span class='hs-varid'>segment_ids</span>
+<a name="line-14347"></a><span class='hs-comment'>{-
+<a name="line-14348"></a>attr {
+<a name="line-14349"></a>  allowed_values {
+<a name="line-14350"></a>    list {
+<a name="line-14351"></a>      type: DT_FLOAT
+<a name="line-14352"></a>      type: DT_DOUBLE
+<a name="line-14353"></a>      type: DT_INT64
+<a name="line-14354"></a>      type: DT_INT32
+<a name="line-14355"></a>      type: DT_UINT8
+<a name="line-14356"></a>      type: DT_UINT16
+<a name="line-14357"></a>      type: DT_INT16
+<a name="line-14358"></a>      type: DT_INT8
+<a name="line-14359"></a>      type: DT_COMPLEX64
+<a name="line-14360"></a>      type: DT_COMPLEX128
+<a name="line-14361"></a>      type: DT_QINT8
+<a name="line-14362"></a>      type: DT_QUINT8
+<a name="line-14363"></a>      type: DT_QINT32
+<a name="line-14364"></a>      type: DT_HALF
+<a name="line-14365"></a>    }
+<a name="line-14366"></a>  }
+<a name="line-14367"></a>  name: "T"
+<a name="line-14368"></a>  type: "type"
+<a name="line-14369"></a>}
+<a name="line-14370"></a>attr {
+<a name="line-14371"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-14372"></a>  name: "Tindices"
+<a name="line-14373"></a>  type: "type"
+<a name="line-14374"></a>}
+<a name="line-14375"></a>input_arg { name: "data" type_attr: "T" }
+<a name="line-14376"></a>input_arg {
+<a name="line-14377"></a>  description: "A 1-D tensor whose rank is equal to the rank of `data`\'s\nfirst dimension.  Values should be sorted and can be repeated."
+<a name="line-14378"></a>  name: "segment_ids"
+<a name="line-14379"></a>  type_attr: "Tindices"
+<a name="line-14380"></a>}
+<a name="line-14381"></a>output_arg {
+<a name="line-14382"></a>  description: "Has same shape as data, except for dimension 0 which\nhas size `k`, the number of segments."
+<a name="line-14383"></a>  name: "output"
+<a name="line-14384"></a>  type_attr: "T"
+<a name="line-14385"></a>}
+<a name="line-14386"></a>-}</span>
+<a name="line-14387"></a>
+<a name="line-14388"></a><a name="encodeJpeg"></a><span class='hs-comment'>-- | JPEG-encode an image.</span>
+<a name="line-14389"></a><span class='hs-comment'>--</span>
+<a name="line-14390"></a><span class='hs-comment'>-- `image` is a 3-D uint8 Tensor of shape `[height, width, channels]`.</span>
+<a name="line-14391"></a><span class='hs-comment'>-- </span>
+<a name="line-14392"></a><span class='hs-comment'>-- The attr `format` can be used to override the color format of the encoded</span>
+<a name="line-14393"></a><span class='hs-comment'>-- output.  Values can be:</span>
+<a name="line-14394"></a><span class='hs-comment'>-- </span>
+<a name="line-14395"></a><span class='hs-comment'>-- *   `''`: Use a default format based on the number of channels in the image.</span>
+<a name="line-14396"></a><span class='hs-comment'>-- *   `grayscale`: Output a grayscale JPEG image.  The `channels` dimension</span>
+<a name="line-14397"></a><span class='hs-comment'>--     of `image` must be 1.</span>
+<a name="line-14398"></a><span class='hs-comment'>-- *   `rgb`: Output an RGB JPEG image. The `channels` dimension</span>
+<a name="line-14399"></a><span class='hs-comment'>--     of `image` must be 3.</span>
+<a name="line-14400"></a><span class='hs-comment'>-- </span>
+<a name="line-14401"></a><span class='hs-comment'>-- If `format` is not specified or is the empty string, a default format is picked</span>
+<a name="line-14402"></a><span class='hs-comment'>-- in function of the number of channels in `image`:</span>
+<a name="line-14403"></a><span class='hs-comment'>-- </span>
+<a name="line-14404"></a><span class='hs-comment'>-- *   1: Output a grayscale image.</span>
+<a name="line-14405"></a><span class='hs-comment'>-- *   3: Output an RGB image.</span>
+<a name="line-14406"></a><span class='hs-definition'>encodeJpeg</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span> <span class='hs-comment'>-- ^ __image__: 3-D with shape `[height, width, channels]`.</span>
+<a name="line-14407"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __contents__: 0-D. JPEG-encoded image.</span>
+<a name="line-14408"></a><span class='hs-definition'>encodeJpeg</span> <span class='hs-varid'>image</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-14409"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"EncodeJpeg"</span><span class='hs-layout'>)</span>
+<a name="line-14410"></a>        <span class='hs-varid'>image</span>
+<a name="line-14411"></a><span class='hs-comment'>{-
+<a name="line-14412"></a>attr {
+<a name="line-14413"></a>  allowed_values { list { s: "" s: "grayscale" s: "rgb" } }
+<a name="line-14414"></a>  default_value { s: "" }
+<a name="line-14415"></a>  description: "Per pixel image format."
+<a name="line-14416"></a>  name: "format"
+<a name="line-14417"></a>  type: "string"
+<a name="line-14418"></a>}
+<a name="line-14419"></a>attr {
+<a name="line-14420"></a>  default_value { i: 95 }
+<a name="line-14421"></a>  description: "Quality of the compression from 0 to 100 (higher is better and slower)."
+<a name="line-14422"></a>  name: "quality"
+<a name="line-14423"></a>  type: "int"
+<a name="line-14424"></a>}
+<a name="line-14425"></a>attr {
+<a name="line-14426"></a>  default_value { b: false }
+<a name="line-14427"></a>  description: "If True, create a JPEG that loads progressively (coarse to fine)."
+<a name="line-14428"></a>  name: "progressive"
+<a name="line-14429"></a>  type: "bool"
+<a name="line-14430"></a>}
+<a name="line-14431"></a>attr {
+<a name="line-14432"></a>  default_value { b: false }
+<a name="line-14433"></a>  description: "If True, spend CPU/RAM to reduce size with no quality change."
+<a name="line-14434"></a>  name: "optimize_size"
+<a name="line-14435"></a>  type: "bool"
+<a name="line-14436"></a>}
+<a name="line-14437"></a>attr {
+<a name="line-14438"></a>  default_value { b: true }
+<a name="line-14439"></a>  description: "See <a href="http://en.wikipedia.org/wiki/Chroma_subsampling.">http://en.wikipedia.org/wiki/Chroma_subsampling.</a>"
+<a name="line-14440"></a>  name: "chroma_downsampling"
+<a name="line-14441"></a>  type: "bool"
+<a name="line-14442"></a>}
+<a name="line-14443"></a>attr {
+<a name="line-14444"></a>  allowed_values { list { s: "in" s: "cm" } }
+<a name="line-14445"></a>  default_value { s: "in" }
+<a name="line-14446"></a>  description: "Unit used to specify `x_density` and `y_density`:\npixels per inch (`\'in\'`) or centimeter (`\'cm\'`)."
+<a name="line-14447"></a>  name: "density_unit"
+<a name="line-14448"></a>  type: "string"
+<a name="line-14449"></a>}
+<a name="line-14450"></a>attr {
+<a name="line-14451"></a>  default_value { i: 300 }
+<a name="line-14452"></a>  description: "Horizontal pixels per density unit."
+<a name="line-14453"></a>  name: "x_density"
+<a name="line-14454"></a>  type: "int"
+<a name="line-14455"></a>}
+<a name="line-14456"></a>attr {
+<a name="line-14457"></a>  default_value { i: 300 }
+<a name="line-14458"></a>  description: "Vertical pixels per density unit."
+<a name="line-14459"></a>  name: "y_density"
+<a name="line-14460"></a>  type: "int"
+<a name="line-14461"></a>}
+<a name="line-14462"></a>attr {
+<a name="line-14463"></a>  default_value { s: "" }
+<a name="line-14464"></a>  description: "If not empty, embed this XMP metadata in the image header."
+<a name="line-14465"></a>  name: "xmp_metadata"
+<a name="line-14466"></a>  type: "string"
+<a name="line-14467"></a>}
+<a name="line-14468"></a>input_arg {
+<a name="line-14469"></a>  description: "3-D with shape `[height, width, channels]`."
+<a name="line-14470"></a>  name: "image"
+<a name="line-14471"></a>  type: DT_UINT8
+<a name="line-14472"></a>}
+<a name="line-14473"></a>output_arg {
+<a name="line-14474"></a>  description: "0-D. JPEG-encoded image."
+<a name="line-14475"></a>  name: "contents"
+<a name="line-14476"></a>  type: DT_STRING
+<a name="line-14477"></a>}
+<a name="line-14478"></a>-}</span>
+<a name="line-14479"></a>
+<a name="line-14480"></a><a name="batchNormWithGlobalNormalizationGrad"></a><span class='hs-comment'>-- | Gradients for batch normalization.</span>
+<a name="line-14481"></a><span class='hs-comment'>--</span>
+<a name="line-14482"></a><span class='hs-comment'>-- This op is deprecated. See `tf.nn.batch_normalization`.</span>
+<a name="line-14483"></a><span class='hs-definition'>batchNormWithGlobalNormalizationGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-14484"></a>                                                                   <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-14485"></a>                                                                           <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-14486"></a>                                                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-14487"></a>                                                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-14488"></a>                                                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-14489"></a>                                                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-14490"></a>                                                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-14491"></a>                                                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-14492"></a>                                                                           <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-14493"></a>                                                                           <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-14494"></a>                                        <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __scale_after_normalization__: A bool indicating whether the resulted tensor</span>
+<a name="line-14495"></a>                                             <span class='hs-comment'>-- needs to be multiplied with gamma.</span>
+<a name="line-14496"></a>                                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __variance_epsilon__: A small float number to avoid dividing by 0.</span>
+<a name="line-14497"></a>                                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __t__: A 4D input Tensor.</span>
+<a name="line-14498"></a>                                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __m__: A 1D mean Tensor with size matching the last dimension of t.</span>
+<a name="line-14499"></a>                                                       <span class='hs-comment'>-- This is the first output from tf.nn.moments,</span>
+<a name="line-14500"></a>                                                       <span class='hs-comment'>-- or a saved moving average thereof.</span>
+<a name="line-14501"></a>                                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __v__: A 1D variance Tensor with size matching the last dimension of t.</span>
+<a name="line-14502"></a>                                                       <span class='hs-comment'>-- This is the second output from tf.nn.moments,</span>
+<a name="line-14503"></a>                                                       <span class='hs-comment'>-- or a saved moving average thereof.</span>
+<a name="line-14504"></a>                                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __gamma__: A 1D gamma Tensor with size matching the last dimension of t.</span>
+<a name="line-14505"></a>                                                       <span class='hs-comment'>-- If "scale_after_normalization" is true, this Tensor will be multiplied</span>
+<a name="line-14506"></a>                                                       <span class='hs-comment'>-- with the normalized Tensor.</span>
+<a name="line-14507"></a>                                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __backprop__: 4D backprop Tensor.</span>
+<a name="line-14508"></a>                                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-14509"></a>                                            <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-14510"></a>                                            <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-14511"></a>                                        <span class='hs-comment'>-- ^ (__dx__, __dm__, __dv__, __db__, __dg__)</span>
+<a name="line-14512"></a>                                        <span class='hs-comment'>--</span>
+<a name="line-14513"></a>                                        <span class='hs-comment'>-- * __dx__: 4D backprop tensor for input.</span>
+<a name="line-14514"></a>                                        <span class='hs-comment'>--</span>
+<a name="line-14515"></a>                                        <span class='hs-comment'>-- * __dm__: 1D backprop tensor for mean.</span>
+<a name="line-14516"></a>                                        <span class='hs-comment'>--</span>
+<a name="line-14517"></a>                                        <span class='hs-comment'>-- * __dv__: 1D backprop tensor for variance.</span>
+<a name="line-14518"></a>                                        <span class='hs-comment'>--</span>
+<a name="line-14519"></a>                                        <span class='hs-comment'>-- * __db__: 1D backprop tensor for beta.</span>
+<a name="line-14520"></a>                                        <span class='hs-comment'>--</span>
+<a name="line-14521"></a>                                        <span class='hs-comment'>-- * __dg__: 1D backprop tensor for gamma.</span>
+<a name="line-14522"></a><span class='hs-definition'>batchNormWithGlobalNormalizationGrad</span> <span class='hs-varid'>scale_after_normalization</span> <span class='hs-varid'>variance_epsilon</span>
+<a name="line-14523"></a>                                     <span class='hs-varid'>t</span> <span class='hs-varid'>m</span> <span class='hs-varid'>v</span> <span class='hs-varid'>gamma</span> <span class='hs-varid'>backprop</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-14524"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchNormWithGlobalNormalizationGrad"</span>
+<a name="line-14525"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-14526"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"scale_after_normalization"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>scale_after_normalization</span>
+<a name="line-14527"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"variance_epsilon"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>variance_epsilon</span><span class='hs-layout'>)</span>
+<a name="line-14528"></a>        <span class='hs-varid'>t</span> <span class='hs-varid'>m</span> <span class='hs-varid'>v</span> <span class='hs-varid'>gamma</span> <span class='hs-varid'>backprop</span>
+<a name="line-14529"></a><span class='hs-comment'>{-
+<a name="line-14530"></a>attr {
+<a name="line-14531"></a>  allowed_values {
+<a name="line-14532"></a>    list {
+<a name="line-14533"></a>      type: DT_FLOAT
+<a name="line-14534"></a>      type: DT_DOUBLE
+<a name="line-14535"></a>      type: DT_INT64
+<a name="line-14536"></a>      type: DT_INT32
+<a name="line-14537"></a>      type: DT_UINT8
+<a name="line-14538"></a>      type: DT_UINT16
+<a name="line-14539"></a>      type: DT_INT16
+<a name="line-14540"></a>      type: DT_INT8
+<a name="line-14541"></a>      type: DT_COMPLEX64
+<a name="line-14542"></a>      type: DT_COMPLEX128
+<a name="line-14543"></a>      type: DT_QINT8
+<a name="line-14544"></a>      type: DT_QUINT8
+<a name="line-14545"></a>      type: DT_QINT32
+<a name="line-14546"></a>      type: DT_HALF
+<a name="line-14547"></a>    }
+<a name="line-14548"></a>  }
+<a name="line-14549"></a>  name: "T"
+<a name="line-14550"></a>  type: "type"
+<a name="line-14551"></a>}
+<a name="line-14552"></a>attr {
+<a name="line-14553"></a>  description: "A small float number to avoid dividing by 0."
+<a name="line-14554"></a>  name: "variance_epsilon"
+<a name="line-14555"></a>  type: "float"
+<a name="line-14556"></a>}
+<a name="line-14557"></a>attr {
+<a name="line-14558"></a>  description: "A bool indicating whether the resulted tensor\nneeds to be multiplied with gamma."
+<a name="line-14559"></a>  name: "scale_after_normalization"
+<a name="line-14560"></a>  type: "bool"
+<a name="line-14561"></a>}
+<a name="line-14562"></a>input_arg {
+<a name="line-14563"></a>  description: "A 4D input Tensor." name: "t" type_attr: "T"
+<a name="line-14564"></a>}
+<a name="line-14565"></a>input_arg {
+<a name="line-14566"></a>  description: "A 1D mean Tensor with size matching the last dimension of t.\nThis is the first output from tf.nn.moments,\nor a saved moving average thereof."
+<a name="line-14567"></a>  name: "m"
+<a name="line-14568"></a>  type_attr: "T"
+<a name="line-14569"></a>}
+<a name="line-14570"></a>input_arg {
+<a name="line-14571"></a>  description: "A 1D variance Tensor with size matching the last dimension of t.\nThis is the second output from tf.nn.moments,\nor a saved moving average thereof."
+<a name="line-14572"></a>  name: "v"
+<a name="line-14573"></a>  type_attr: "T"
+<a name="line-14574"></a>}
+<a name="line-14575"></a>input_arg {
+<a name="line-14576"></a>  description: "A 1D gamma Tensor with size matching the last dimension of t.\nIf \"scale_after_normalization\" is true, this Tensor will be multiplied\nwith the normalized Tensor."
+<a name="line-14577"></a>  name: "gamma"
+<a name="line-14578"></a>  type_attr: "T"
+<a name="line-14579"></a>}
+<a name="line-14580"></a>input_arg {
+<a name="line-14581"></a>  description: "4D backprop Tensor." name: "backprop" type_attr: "T"
+<a name="line-14582"></a>}
+<a name="line-14583"></a>output_arg {
+<a name="line-14584"></a>  description: "4D backprop tensor for input."
+<a name="line-14585"></a>  name: "dx"
+<a name="line-14586"></a>  type_attr: "T"
+<a name="line-14587"></a>}
+<a name="line-14588"></a>output_arg {
+<a name="line-14589"></a>  description: "1D backprop tensor for mean."
+<a name="line-14590"></a>  name: "dm"
+<a name="line-14591"></a>  type_attr: "T"
+<a name="line-14592"></a>}
+<a name="line-14593"></a>output_arg {
+<a name="line-14594"></a>  description: "1D backprop tensor for variance."
+<a name="line-14595"></a>  name: "dv"
+<a name="line-14596"></a>  type_attr: "T"
+<a name="line-14597"></a>}
+<a name="line-14598"></a>output_arg {
+<a name="line-14599"></a>  description: "1D backprop tensor for beta."
+<a name="line-14600"></a>  name: "db"
+<a name="line-14601"></a>  type_attr: "T"
+<a name="line-14602"></a>}
+<a name="line-14603"></a>output_arg {
+<a name="line-14604"></a>  description: "1D backprop tensor for gamma."
+<a name="line-14605"></a>  name: "dg"
+<a name="line-14606"></a>  type_attr: "T"
+<a name="line-14607"></a>}
+<a name="line-14608"></a>-}</span>
+<a name="line-14609"></a>
+<a name="line-14610"></a><a name="biasAddV1"></a><span class='hs-comment'>-- | Adds `bias` to `value`.</span>
+<a name="line-14611"></a><span class='hs-comment'>--</span>
+<a name="line-14612"></a><span class='hs-comment'>-- This is a deprecated version of BiasAdd and will be soon removed.</span>
+<a name="line-14613"></a><span class='hs-comment'>-- </span>
+<a name="line-14614"></a><span class='hs-comment'>-- This is a special case of `tf.add` where `bias` is restricted to be 1-D.</span>
+<a name="line-14615"></a><span class='hs-comment'>-- Broadcasting is supported, so `value` may have any number of dimensions.</span>
+<a name="line-14616"></a><span class='hs-definition'>biasAddV1</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-14617"></a>                               <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-14618"></a>                                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-14619"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-14620"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-14621"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-14622"></a>                                       <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-14623"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: Any number of dimensions.</span>
+<a name="line-14624"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __bias__: 1-D with size the last dimension of `value`.</span>
+<a name="line-14625"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Broadcasted sum of `value` and `bias`.</span>
+<a name="line-14626"></a><span class='hs-definition'>biasAddV1</span> <span class='hs-varid'>value</span> <span class='hs-varid'>bias</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-14627"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BiasAddV1"</span>
+<a name="line-14628"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-14629"></a>        <span class='hs-varid'>value</span> <span class='hs-varid'>bias</span>
+<a name="line-14630"></a><span class='hs-comment'>{-
+<a name="line-14631"></a>attr {
+<a name="line-14632"></a>  allowed_values {
+<a name="line-14633"></a>    list {
+<a name="line-14634"></a>      type: DT_FLOAT
+<a name="line-14635"></a>      type: DT_DOUBLE
+<a name="line-14636"></a>      type: DT_INT64
+<a name="line-14637"></a>      type: DT_INT32
+<a name="line-14638"></a>      type: DT_UINT8
+<a name="line-14639"></a>      type: DT_UINT16
+<a name="line-14640"></a>      type: DT_INT16
+<a name="line-14641"></a>      type: DT_INT8
+<a name="line-14642"></a>      type: DT_COMPLEX64
+<a name="line-14643"></a>      type: DT_COMPLEX128
+<a name="line-14644"></a>      type: DT_QINT8
+<a name="line-14645"></a>      type: DT_QUINT8
+<a name="line-14646"></a>      type: DT_QINT32
+<a name="line-14647"></a>      type: DT_HALF
+<a name="line-14648"></a>    }
+<a name="line-14649"></a>  }
+<a name="line-14650"></a>  name: "T"
+<a name="line-14651"></a>  type: "type"
+<a name="line-14652"></a>}
+<a name="line-14653"></a>input_arg {
+<a name="line-14654"></a>  description: "Any number of dimensions."
+<a name="line-14655"></a>  name: "value"
+<a name="line-14656"></a>  type_attr: "T"
+<a name="line-14657"></a>}
+<a name="line-14658"></a>input_arg {
+<a name="line-14659"></a>  description: "1-D with size the last dimension of `value`."
+<a name="line-14660"></a>  name: "bias"
+<a name="line-14661"></a>  type_attr: "T"
+<a name="line-14662"></a>}
+<a name="line-14663"></a>output_arg {
+<a name="line-14664"></a>  description: "Broadcasted sum of `value` and `bias`."
+<a name="line-14665"></a>  name: "output"
+<a name="line-14666"></a>  type_attr: "T"
+<a name="line-14667"></a>}
+<a name="line-14668"></a>-}</span>
+<a name="line-14669"></a>
+<a name="line-14670"></a><a name="invertPermutation"></a><span class='hs-comment'>-- | Computes the inverse permutation of a tensor.</span>
+<a name="line-14671"></a><span class='hs-comment'>--</span>
+<a name="line-14672"></a><span class='hs-comment'>-- This operation computes the inverse of an index permutation. It takes a 1-D</span>
+<a name="line-14673"></a><span class='hs-comment'>-- integer tensor `x`, which represents the indices of a zero-based array, and</span>
+<a name="line-14674"></a><span class='hs-comment'>-- swaps each value with its index position. In other words, for an output tensor</span>
+<a name="line-14675"></a><span class='hs-comment'>-- `y` and an input tensor `x`, this operation computes the following:</span>
+<a name="line-14676"></a><span class='hs-comment'>-- </span>
+<a name="line-14677"></a><span class='hs-comment'>-- `y[x[i]] = i for i in [0, 1, ..., len(x) - 1]`</span>
+<a name="line-14678"></a><span class='hs-comment'>-- </span>
+<a name="line-14679"></a><span class='hs-comment'>-- The values must include 0. There can be no duplicate values or negative values.</span>
+<a name="line-14680"></a><span class='hs-comment'>-- </span>
+<a name="line-14681"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-14682"></a><span class='hs-comment'>-- </span>
+<a name="line-14683"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-14684"></a><span class='hs-comment'>-- # tensor `x` is [3, 4, 0, 2, 1]</span>
+<a name="line-14685"></a><span class='hs-comment'>-- invert_permutation(x) ==&gt; [2, 4, 3, 0, 1]</span>
+<a name="line-14686"></a><span class='hs-comment'>-- ```</span>
+<a name="line-14687"></a><span class='hs-definition'>invertPermutation</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-14688"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-14689"></a>                     <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__: 1-D.</span>
+<a name="line-14690"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__: 1-D.</span>
+<a name="line-14691"></a><span class='hs-definition'>invertPermutation</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-14692"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"InvertPermutation"</span>
+<a name="line-14693"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-14694"></a>        <span class='hs-varid'>x</span>
+<a name="line-14695"></a><span class='hs-comment'>{-
+<a name="line-14696"></a>attr {
+<a name="line-14697"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-14698"></a>  default_value { type: DT_INT32 }
+<a name="line-14699"></a>  name: "T"
+<a name="line-14700"></a>  type: "type"
+<a name="line-14701"></a>}
+<a name="line-14702"></a>input_arg { description: "1-D." name: "x" type_attr: "T" }
+<a name="line-14703"></a>output_arg { description: "1-D." name: "y" type_attr: "T" }
+<a name="line-14704"></a>-}</span>
+<a name="line-14705"></a>
+<a name="line-14706"></a><a name="mirrorPadGrad"></a><span class='hs-comment'>-- | Gradient op for `MirrorPad` op. This op folds a mirror-padded tensor.</span>
+<a name="line-14707"></a><span class='hs-comment'>--</span>
+<a name="line-14708"></a><span class='hs-comment'>-- This operation folds the padded areas of `input` by `MirrorPad` according to the</span>
+<a name="line-14709"></a><span class='hs-comment'>-- `paddings` you specify. `paddings` must be the same as `paddings` argument</span>
+<a name="line-14710"></a><span class='hs-comment'>-- given to the corresponding `MirrorPad` op.</span>
+<a name="line-14711"></a><span class='hs-comment'>-- </span>
+<a name="line-14712"></a><span class='hs-comment'>-- The folded size of each dimension D of the output is:</span>
+<a name="line-14713"></a><span class='hs-comment'>-- </span>
+<a name="line-14714"></a><span class='hs-comment'>-- `input.dim_size(D) - paddings(D, 0) - paddings(D, 1)`</span>
+<a name="line-14715"></a><span class='hs-comment'>-- </span>
+<a name="line-14716"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-14717"></a><span class='hs-comment'>-- </span>
+<a name="line-14718"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-14719"></a><span class='hs-comment'>-- # 't' is [[1, 2, 3], [4, 5, 6], [7, 8, 9]].</span>
+<a name="line-14720"></a><span class='hs-comment'>-- # 'paddings' is [[0, 1]], [0, 1]].</span>
+<a name="line-14721"></a><span class='hs-comment'>-- # 'mode' is SYMMETRIC.</span>
+<a name="line-14722"></a><span class='hs-comment'>-- # rank of 't' is 2.</span>
+<a name="line-14723"></a><span class='hs-comment'>-- pad(t, paddings) ==&gt; [[ 1,  5]</span>
+<a name="line-14724"></a><span class='hs-comment'>--                       [11, 28]]</span>
+<a name="line-14725"></a><span class='hs-comment'>-- ```</span>
+<a name="line-14726"></a><span class='hs-definition'>mirrorPadGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tpaddings</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tpaddings</span><span class='hs-layout'>,</span>
+<a name="line-14727"></a>                                             <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-14728"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tpaddings</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-14729"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: The input tensor to be folded.</span>
+<a name="line-14730"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tpaddings</span> <span class='hs-comment'>-- ^ __paddings__: A two-column matrix specifying the padding sizes. The number of</span>
+<a name="line-14731"></a>                                        <span class='hs-comment'>-- rows must be the same as the rank of `input`.</span>
+<a name="line-14732"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The folded tensor.</span>
+<a name="line-14733"></a><span class='hs-definition'>mirrorPadGrad</span> <span class='hs-varid'>input</span> <span class='hs-varid'>paddings</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-14734"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MirrorPadGrad"</span>
+<a name="line-14735"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-14736"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tpaddings"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tpaddings</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-14737"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>paddings</span>
+<a name="line-14738"></a><span class='hs-comment'>{-
+<a name="line-14739"></a>attr { name: "T" type: "type" }
+<a name="line-14740"></a>attr {
+<a name="line-14741"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-14742"></a>  default_value { type: DT_INT32 }
+<a name="line-14743"></a>  name: "Tpaddings"
+<a name="line-14744"></a>  type: "type"
+<a name="line-14745"></a>}
+<a name="line-14746"></a>attr {
+<a name="line-14747"></a>  allowed_values { list { s: "REFLECT" s: "SYMMETRIC" } }
+<a name="line-14748"></a>  description: "The mode used in the `MirrorPad` op."
+<a name="line-14749"></a>  name: "mode"
+<a name="line-14750"></a>  type: "string"
+<a name="line-14751"></a>}
+<a name="line-14752"></a>input_arg {
+<a name="line-14753"></a>  description: "The input tensor to be folded."
+<a name="line-14754"></a>  name: "input"
+<a name="line-14755"></a>  type_attr: "T"
+<a name="line-14756"></a>}
+<a name="line-14757"></a>input_arg {
+<a name="line-14758"></a>  description: "A two-column matrix specifying the padding sizes. The number of\nrows must be the same as the rank of `input`."
+<a name="line-14759"></a>  name: "paddings"
+<a name="line-14760"></a>  type_attr: "Tpaddings"
+<a name="line-14761"></a>}
+<a name="line-14762"></a>output_arg {
+<a name="line-14763"></a>  description: "The folded tensor." name: "output" type_attr: "T"
+<a name="line-14764"></a>}
+<a name="line-14765"></a>-}</span>
+<a name="line-14766"></a>
+<a name="line-14767"></a><a name="reverse"></a><span class='hs-comment'>-- | Reverses specific dimensions of a tensor.</span>
+<a name="line-14768"></a><span class='hs-comment'>--</span>
+<a name="line-14769"></a><span class='hs-comment'>-- Given a `tensor`, and a `bool` tensor `dims` representing the dimensions</span>
+<a name="line-14770"></a><span class='hs-comment'>-- of `tensor`, this operation reverses each dimension i of `tensor` where</span>
+<a name="line-14771"></a><span class='hs-comment'>-- `dims[i]` is `True`.</span>
+<a name="line-14772"></a><span class='hs-comment'>-- </span>
+<a name="line-14773"></a><span class='hs-comment'>-- `tensor` can have up to 8 dimensions. The number of dimensions</span>
+<a name="line-14774"></a><span class='hs-comment'>-- of `tensor` must equal the number of elements in `dims`. In other words:</span>
+<a name="line-14775"></a><span class='hs-comment'>-- </span>
+<a name="line-14776"></a><span class='hs-comment'>-- `rank(tensor) = size(dims)`</span>
+<a name="line-14777"></a><span class='hs-comment'>-- </span>
+<a name="line-14778"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-14779"></a><span class='hs-comment'>-- </span>
+<a name="line-14780"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-14781"></a><span class='hs-comment'>-- # tensor 't' is [[[[ 0,  1,  2,  3],</span>
+<a name="line-14782"></a><span class='hs-comment'>-- #                  [ 4,  5,  6,  7],</span>
+<a name="line-14783"></a><span class='hs-comment'>-- #                  [ 8,  9, 10, 11]],</span>
+<a name="line-14784"></a><span class='hs-comment'>-- #                 [[12, 13, 14, 15],</span>
+<a name="line-14785"></a><span class='hs-comment'>-- #                  [16, 17, 18, 19],</span>
+<a name="line-14786"></a><span class='hs-comment'>-- #                  [20, 21, 22, 23]]]]</span>
+<a name="line-14787"></a><span class='hs-comment'>-- # tensor 't' shape is [1, 2, 3, 4]</span>
+<a name="line-14788"></a><span class='hs-comment'>-- </span>
+<a name="line-14789"></a><span class='hs-comment'>-- # 'dims' is [False, False, False, True]</span>
+<a name="line-14790"></a><span class='hs-comment'>-- reverse(t, dims) ==&gt; [[[[ 3,  2,  1,  0],</span>
+<a name="line-14791"></a><span class='hs-comment'>--                         [ 7,  6,  5,  4],</span>
+<a name="line-14792"></a><span class='hs-comment'>--                         [ 11, 10, 9, 8]],</span>
+<a name="line-14793"></a><span class='hs-comment'>--                        [[15, 14, 13, 12],</span>
+<a name="line-14794"></a><span class='hs-comment'>--                         [19, 18, 17, 16],</span>
+<a name="line-14795"></a><span class='hs-comment'>--                         [23, 22, 21, 20]]]]</span>
+<a name="line-14796"></a><span class='hs-comment'>-- </span>
+<a name="line-14797"></a><span class='hs-comment'>-- # 'dims' is [False, True, False, False]</span>
+<a name="line-14798"></a><span class='hs-comment'>-- reverse(t, dims) ==&gt; [[[[12, 13, 14, 15],</span>
+<a name="line-14799"></a><span class='hs-comment'>--                         [16, 17, 18, 19],</span>
+<a name="line-14800"></a><span class='hs-comment'>--                         [20, 21, 22, 23]</span>
+<a name="line-14801"></a><span class='hs-comment'>--                        [[ 0,  1,  2,  3],</span>
+<a name="line-14802"></a><span class='hs-comment'>--                         [ 4,  5,  6,  7],</span>
+<a name="line-14803"></a><span class='hs-comment'>--                         [ 8,  9, 10, 11]]]]</span>
+<a name="line-14804"></a><span class='hs-comment'>-- </span>
+<a name="line-14805"></a><span class='hs-comment'>-- # 'dims' is [False, False, True, False]</span>
+<a name="line-14806"></a><span class='hs-comment'>-- reverse(t, dims) ==&gt; [[[[8, 9, 10, 11],</span>
+<a name="line-14807"></a><span class='hs-comment'>--                         [4, 5, 6, 7],</span>
+<a name="line-14808"></a><span class='hs-comment'>--                         [0, 1, 2, 3]]</span>
+<a name="line-14809"></a><span class='hs-comment'>--                        [[20, 21, 22, 23],</span>
+<a name="line-14810"></a><span class='hs-comment'>--                         [16, 17, 18, 19],</span>
+<a name="line-14811"></a><span class='hs-comment'>--                         [12, 13, 14, 15]]]]</span>
+<a name="line-14812"></a><span class='hs-comment'>-- ```</span>
+<a name="line-14813"></a><span class='hs-definition'>reverse</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-14814"></a>                             <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-14815"></a>                                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-14816"></a>                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-14817"></a>                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-14818"></a>                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-14819"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __tensor__: Up to 8-D.</span>
+<a name="line-14820"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __dims__: 1-D. The dimensions to reverse.</span>
+<a name="line-14821"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The same shape as `tensor`.</span>
+<a name="line-14822"></a><span class='hs-definition'>reverse</span> <span class='hs-varid'>tensor</span> <span class='hs-varid'>dims</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-14823"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Reverse"</span>
+<a name="line-14824"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-14825"></a>        <span class='hs-varid'>tensor</span> <span class='hs-varid'>dims</span>
+<a name="line-14826"></a><span class='hs-comment'>{-
+<a name="line-14827"></a>attr {
+<a name="line-14828"></a>  allowed_values {
+<a name="line-14829"></a>    list {
+<a name="line-14830"></a>      type: DT_UINT8
+<a name="line-14831"></a>      type: DT_INT8
+<a name="line-14832"></a>      type: DT_INT32
+<a name="line-14833"></a>      type: DT_INT64
+<a name="line-14834"></a>      type: DT_BOOL
+<a name="line-14835"></a>      type: DT_HALF
+<a name="line-14836"></a>      type: DT_FLOAT
+<a name="line-14837"></a>      type: DT_DOUBLE
+<a name="line-14838"></a>      type: DT_COMPLEX64
+<a name="line-14839"></a>      type: DT_COMPLEX128
+<a name="line-14840"></a>    }
+<a name="line-14841"></a>  }
+<a name="line-14842"></a>  name: "T"
+<a name="line-14843"></a>  type: "type"
+<a name="line-14844"></a>}
+<a name="line-14845"></a>input_arg {
+<a name="line-14846"></a>  description: "Up to 8-D." name: "tensor" type_attr: "T"
+<a name="line-14847"></a>}
+<a name="line-14848"></a>input_arg {
+<a name="line-14849"></a>  description: "1-D. The dimensions to reverse."
+<a name="line-14850"></a>  name: "dims"
+<a name="line-14851"></a>  type: DT_BOOL
+<a name="line-14852"></a>}
+<a name="line-14853"></a>output_arg {
+<a name="line-14854"></a>  description: "The same shape as `tensor`."
+<a name="line-14855"></a>  name: "output"
+<a name="line-14856"></a>  type_attr: "T"
+<a name="line-14857"></a>}
+<a name="line-14858"></a>-}</span>
+<a name="line-14859"></a>
+<a name="line-14860"></a><a name="conv2D"></a><span class='hs-comment'>-- | Computes a 2-D convolution given 4-D `input` and `filter` tensors.</span>
+<a name="line-14861"></a><span class='hs-comment'>--</span>
+<a name="line-14862"></a><span class='hs-comment'>-- Given an input tensor of shape `[batch, in_height, in_width, in_channels]`</span>
+<a name="line-14863"></a><span class='hs-comment'>-- and a filter / kernel tensor of shape</span>
+<a name="line-14864"></a><span class='hs-comment'>-- `[filter_height, filter_width, in_channels, out_channels]`, this op</span>
+<a name="line-14865"></a><span class='hs-comment'>-- performs the following:</span>
+<a name="line-14866"></a><span class='hs-comment'>-- </span>
+<a name="line-14867"></a><span class='hs-comment'>-- 1. Flattens the filter to a 2-D matrix with shape</span>
+<a name="line-14868"></a><span class='hs-comment'>--    `[filter_height * filter_width * in_channels, output_channels]`.</span>
+<a name="line-14869"></a><span class='hs-comment'>-- 2. Extracts image patches from the input tensor to form a *virtual*</span>
+<a name="line-14870"></a><span class='hs-comment'>--    tensor of shape `[batch, out_height, out_width,</span>
+<a name="line-14871"></a><span class='hs-comment'>--    filter_height * filter_width * in_channels]`.</span>
+<a name="line-14872"></a><span class='hs-comment'>-- 3. For each patch, right-multiplies the filter matrix and the image patch</span>
+<a name="line-14873"></a><span class='hs-comment'>--    vector.</span>
+<a name="line-14874"></a><span class='hs-comment'>-- </span>
+<a name="line-14875"></a><span class='hs-comment'>-- In detail, with the default NHWC format,</span>
+<a name="line-14876"></a><span class='hs-comment'>-- </span>
+<a name="line-14877"></a><span class='hs-comment'>--     output[b, i, j, k] =</span>
+<a name="line-14878"></a><span class='hs-comment'>--         sum_{di, dj, q} input[b, strides[1] * i + di, strides[2] * j + dj, q] *</span>
+<a name="line-14879"></a><span class='hs-comment'>--                         filter[di, dj, q, k]</span>
+<a name="line-14880"></a><span class='hs-comment'>-- </span>
+<a name="line-14881"></a><span class='hs-comment'>-- Must have `strides[0] = strides[3] = 1`.  For the most common case of the same</span>
+<a name="line-14882"></a><span class='hs-comment'>-- horizontal and vertices strides, `strides = [1, stride, stride, 1]`.</span>
+<a name="line-14883"></a><span class='hs-definition'>conv2D</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-14884"></a>                                                  <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-14885"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-14886"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __filter__</span>
+<a name="line-14887"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-14888"></a><span class='hs-definition'>conv2D</span> <span class='hs-varid'>input</span> <span class='hs-varid'>filter</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-14889"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Conv2D"</span>
+<a name="line-14890"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-14891"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>filter</span>
+<a name="line-14892"></a><span class='hs-comment'>{-
+<a name="line-14893"></a>attr {
+<a name="line-14894"></a>  allowed_values {
+<a name="line-14895"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-14896"></a>  }
+<a name="line-14897"></a>  name: "T"
+<a name="line-14898"></a>  type: "type"
+<a name="line-14899"></a>}
+<a name="line-14900"></a>attr {
+<a name="line-14901"></a>  description: "1-D of length 4.  The stride of the sliding window for each dimension\nof `input`. Must be in the same order as the dimension specified with format."
+<a name="line-14902"></a>  name: "strides"
+<a name="line-14903"></a>  type: "list(int)"
+<a name="line-14904"></a>}
+<a name="line-14905"></a>attr {
+<a name="line-14906"></a>  default_value { b: true } name: "use_cudnn_on_gpu" type: "bool"
+<a name="line-14907"></a>}
+<a name="line-14908"></a>attr {
+<a name="line-14909"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-14910"></a>  description: "The type of padding algorithm to use."
+<a name="line-14911"></a>  name: "padding"
+<a name="line-14912"></a>  type: "string"
+<a name="line-14913"></a>}
+<a name="line-14914"></a>attr {
+<a name="line-14915"></a>  allowed_values { list { s: "NHWC" s: "NCHW" } }
+<a name="line-14916"></a>  default_value { s: "NHWC" }
+<a name="line-14917"></a>  description: "Specify the data format of the input and output data. With the\ndefault format \"NHWC\", the data is stored in the order of:\n    [batch, in_height, in_width, in_channels].\nAlternatively, the format could be \"NCHW\", the data storage order of:\n    [batch, in_channels, in_height, in_width]."
+<a name="line-14918"></a>  name: "data_format"
+<a name="line-14919"></a>  type: "string"
+<a name="line-14920"></a>}
+<a name="line-14921"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-14922"></a>input_arg { name: "filter" type_attr: "T" }
+<a name="line-14923"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-14924"></a>-}</span>
+<a name="line-14925"></a>
+<a name="line-14926"></a><span class='hs-comment'>-- | Computes the gradients of convolution with respect to the input.</span>
+<a name="line-14927"></a>
+<a name="line-14928"></a><a name="conv2DBackpropInput"></a><span class='hs-definition'>conv2DBackpropInput</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-14929"></a>                                            <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-14930"></a>                                                    <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-14931"></a>                       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __input_sizes__: An integer vector representing the shape of `input`,</span>
+<a name="line-14932"></a>                                                <span class='hs-comment'>-- where `input` is a 4-D `[batch, height, width, channels]` tensor.</span>
+<a name="line-14933"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __filter__: 4-D with shape</span>
+<a name="line-14934"></a>                                      <span class='hs-comment'>-- `[filter_height, filter_width, in_channels, out_channels]`.</span>
+<a name="line-14935"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out_backprop__: 4-D with shape `[batch, out_height, out_width, out_channels]`.</span>
+<a name="line-14936"></a>                                      <span class='hs-comment'>-- Gradients w.r.t. the output of the convolution.</span>
+<a name="line-14937"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 4-D with shape `[batch, in_height, in_width, in_channels]`.  Gradient</span>
+<a name="line-14938"></a>                       <span class='hs-comment'>-- w.r.t. the input of the convolution.</span>
+<a name="line-14939"></a><span class='hs-definition'>conv2DBackpropInput</span> <span class='hs-varid'>input_sizes</span> <span class='hs-varid'>filter</span> <span class='hs-varid'>out_backprop</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-14940"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Conv2DBackpropInput"</span>
+<a name="line-14941"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-14942"></a>        <span class='hs-varid'>input_sizes</span> <span class='hs-varid'>filter</span> <span class='hs-varid'>out_backprop</span>
+<a name="line-14943"></a><span class='hs-comment'>{-
+<a name="line-14944"></a>attr {
+<a name="line-14945"></a>  allowed_values {
+<a name="line-14946"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-14947"></a>  }
+<a name="line-14948"></a>  name: "T"
+<a name="line-14949"></a>  type: "type"
+<a name="line-14950"></a>}
+<a name="line-14951"></a>attr {
+<a name="line-14952"></a>  description: "The stride of the sliding window for each dimension of the input\nof the convolution. Must be in the same order as the dimension specified with\nformat."
+<a name="line-14953"></a>  name: "strides"
+<a name="line-14954"></a>  type: "list(int)"
+<a name="line-14955"></a>}
+<a name="line-14956"></a>attr {
+<a name="line-14957"></a>  default_value { b: true } name: "use_cudnn_on_gpu" type: "bool"
+<a name="line-14958"></a>}
+<a name="line-14959"></a>attr {
+<a name="line-14960"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-14961"></a>  description: "The type of padding algorithm to use."
+<a name="line-14962"></a>  name: "padding"
+<a name="line-14963"></a>  type: "string"
+<a name="line-14964"></a>}
+<a name="line-14965"></a>attr {
+<a name="line-14966"></a>  allowed_values { list { s: "NHWC" s: "NCHW" } }
+<a name="line-14967"></a>  default_value { s: "NHWC" }
+<a name="line-14968"></a>  description: "Specify the data format of the input and output data. With the\ndefault format \"NHWC\", the data is stored in the order of:\n    [batch, in_height, in_width, in_channels].\nAlternatively, the format could be \"NCHW\", the data storage order of:\n    [batch, in_channels, in_height, in_width]."
+<a name="line-14969"></a>  name: "data_format"
+<a name="line-14970"></a>  type: "string"
+<a name="line-14971"></a>}
+<a name="line-14972"></a>input_arg {
+<a name="line-14973"></a>  description: "An integer vector representing the shape of `input`,\nwhere `input` is a 4-D `[batch, height, width, channels]` tensor."
+<a name="line-14974"></a>  name: "input_sizes"
+<a name="line-14975"></a>  type: DT_INT32
+<a name="line-14976"></a>}
+<a name="line-14977"></a>input_arg {
+<a name="line-14978"></a>  description: "4-D with shape\n`[filter_height, filter_width, in_channels, out_channels]`."
+<a name="line-14979"></a>  name: "filter"
+<a name="line-14980"></a>  type_attr: "T"
+<a name="line-14981"></a>}
+<a name="line-14982"></a>input_arg {
+<a name="line-14983"></a>  description: "4-D with shape `[batch, out_height, out_width, out_channels]`.\nGradients w.r.t. the output of the convolution."
+<a name="line-14984"></a>  name: "out_backprop"
+<a name="line-14985"></a>  type_attr: "T"
+<a name="line-14986"></a>}
+<a name="line-14987"></a>output_arg {
+<a name="line-14988"></a>  description: "4-D with shape `[batch, in_height, in_width, in_channels]`.  Gradient\nw.r.t. the input of the convolution."
+<a name="line-14989"></a>  name: "output"
+<a name="line-14990"></a>  type_attr: "T"
+<a name="line-14991"></a>}
+<a name="line-14992"></a>-}</span>
+<a name="line-14993"></a>
+<a name="line-14994"></a><a name="readerSerializeState"></a><span class='hs-comment'>-- | Produce a string tensor that encodes the state of a Reader.</span>
+<a name="line-14995"></a><span class='hs-comment'>--</span>
+<a name="line-14996"></a><span class='hs-comment'>-- Not all Readers support being serialized, so this can produce an</span>
+<a name="line-14997"></a><span class='hs-comment'>-- Unimplemented error.</span>
+<a name="line-14998"></a><span class='hs-definition'>readerSerializeState</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __reader_handle__: Handle to a Reader.</span>
+<a name="line-14999"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __state__</span>
+<a name="line-15000"></a><span class='hs-definition'>readerSerializeState</span> <span class='hs-varid'>reader_handle</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-15001"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ReaderSerializeState"</span><span class='hs-layout'>)</span>
+<a name="line-15002"></a>        <span class='hs-varid'>reader_handle</span>
+<a name="line-15003"></a><span class='hs-comment'>{-
+<a name="line-15004"></a>input_arg {
+<a name="line-15005"></a>  description: "Handle to a Reader."
+<a name="line-15006"></a>  is_ref: true
+<a name="line-15007"></a>  name: "reader_handle"
+<a name="line-15008"></a>  type: DT_STRING
+<a name="line-15009"></a>}
+<a name="line-15010"></a>output_arg { name: "state" type: DT_STRING }
+<a name="line-15011"></a>-}</span>
+<a name="line-15012"></a>
+<a name="line-15013"></a><a name="temporaryVariable"></a><span class='hs-comment'>-- | Returns a tensor that may be mutated, but only persists within a single step.</span>
+<a name="line-15014"></a><span class='hs-comment'>--</span>
+<a name="line-15015"></a><span class='hs-comment'>-- This is an experimental op for internal use only and it is possible to use this</span>
+<a name="line-15016"></a><span class='hs-comment'>-- op in unsafe ways.  DO NOT USE unless you fully understand the risks.</span>
+<a name="line-15017"></a><span class='hs-comment'>-- </span>
+<a name="line-15018"></a><span class='hs-comment'>-- It is the caller's responsibility to ensure that 'ref' is eventually passed to a</span>
+<a name="line-15019"></a><span class='hs-comment'>-- matching 'DestroyTemporaryVariable' op after all other uses have completed.</span>
+<a name="line-15020"></a><span class='hs-comment'>-- </span>
+<a name="line-15021"></a><span class='hs-comment'>-- Outputs a ref to the tensor state so it may be read or modified.</span>
+<a name="line-15022"></a><span class='hs-comment'>-- </span>
+<a name="line-15023"></a><span class='hs-comment'>--   E.g.</span>
+<a name="line-15024"></a><span class='hs-comment'>--       var = state_ops._temporary_variable([1, 2], types.float_)</span>
+<a name="line-15025"></a><span class='hs-comment'>--       var_name = var.op.name</span>
+<a name="line-15026"></a><span class='hs-comment'>--       var = state_ops.assign(var, [[4.0, 5.0]])</span>
+<a name="line-15027"></a><span class='hs-comment'>--       var = state_ops.assign_add(var, [[6.0, 7.0]])</span>
+<a name="line-15028"></a><span class='hs-comment'>--       final = state_ops._destroy_temporary_variable(var, var_name=var_name)</span>
+<a name="line-15029"></a><span class='hs-definition'>temporaryVariable</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-15030"></a>                     <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __ref__: A reference to the variable tensor.</span>
+<a name="line-15031"></a><span class='hs-definition'>temporaryVariable</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-15032"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TemporaryVariable"</span>
+<a name="line-15033"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-15034"></a>        
+<a name="line-15035"></a><span class='hs-comment'>{-
+<a name="line-15036"></a>attr {
+<a name="line-15037"></a>  description: "The shape of the variable tensor."
+<a name="line-15038"></a>  name: "shape"
+<a name="line-15039"></a>  type: "shape"
+<a name="line-15040"></a>}
+<a name="line-15041"></a>attr {
+<a name="line-15042"></a>  description: "The type of elements in the variable tensor."
+<a name="line-15043"></a>  name: "dtype"
+<a name="line-15044"></a>  type: "type"
+<a name="line-15045"></a>}
+<a name="line-15046"></a>attr {
+<a name="line-15047"></a>  default_value { s: "" }
+<a name="line-15048"></a>  description: "Overrides the name used for the temporary variable resource. Default\nvalue is the name of the \'TemporaryVariable\' op (which is guaranteed unique)."
+<a name="line-15049"></a>  name: "var_name"
+<a name="line-15050"></a>  type: "string"
+<a name="line-15051"></a>}
+<a name="line-15052"></a>output_arg {
+<a name="line-15053"></a>  description: "A reference to the variable tensor."
+<a name="line-15054"></a>  is_ref: true
+<a name="line-15055"></a>  name: "ref"
+<a name="line-15056"></a>  type_attr: "dtype"
+<a name="line-15057"></a>}
+<a name="line-15058"></a>-}</span>
+<a name="line-15059"></a>
+<a name="line-15060"></a><a name="cropAndResize"></a><span class='hs-comment'>-- | Extracts crops from the input image tensor and bilinearly resizes them (possibly</span>
+<a name="line-15061"></a><span class='hs-comment'>--</span>
+<a name="line-15062"></a><span class='hs-comment'>-- with aspect ratio change) to a common output size specified by `crop_size`. This</span>
+<a name="line-15063"></a><span class='hs-comment'>-- is more general than the `crop_to_bounding_box` op which extracts a fixed size</span>
+<a name="line-15064"></a><span class='hs-comment'>-- slice from the input image and does not allow resizing or aspect ratio change.</span>
+<a name="line-15065"></a><span class='hs-comment'>-- </span>
+<a name="line-15066"></a><span class='hs-comment'>-- Returns a tensor with `crops` from the input `image` at positions defined at the</span>
+<a name="line-15067"></a><span class='hs-comment'>-- bounding box locations in `boxes`. The cropped boxes are all resized (with</span>
+<a name="line-15068"></a><span class='hs-comment'>-- bilinear interpolation) to a fixed `size = [crop_height, crop_width]`. The</span>
+<a name="line-15069"></a><span class='hs-comment'>-- result is a 4-D tensor `[num_boxes, crop_height, crop_width, depth]`.</span>
+<a name="line-15070"></a><span class='hs-definition'>cropAndResize</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-15071"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-15072"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-15073"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-15074"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-15075"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-15076"></a>                                                               <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-15077"></a>                                                               <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-15078"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __image__: A 4-D tensor of shape `[batch, image_height, image_width, depth]`.</span>
+<a name="line-15079"></a>                             <span class='hs-comment'>-- Both `image_height` and `image_width` need to be positive.</span>
+<a name="line-15080"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __boxes__: A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor</span>
+<a name="line-15081"></a>                                    <span class='hs-comment'>-- specifies the coordinates of a box in the `box_ind[i]` image and is specified</span>
+<a name="line-15082"></a>                                    <span class='hs-comment'>-- in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of</span>
+<a name="line-15083"></a>                                    <span class='hs-comment'>-- `y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the</span>
+<a name="line-15084"></a>                                    <span class='hs-comment'>-- `[0, 1]` interval of normalized image height is mapped to</span>
+<a name="line-15085"></a>                                    <span class='hs-comment'>-- `[0, image_height - 1] in image height coordinates. We do allow y1 &gt; y2, in</span>
+<a name="line-15086"></a>                                    <span class='hs-comment'>-- which case the sampled crop is an up-down flipped version of the original</span>
+<a name="line-15087"></a>                                    <span class='hs-comment'>-- image. The width dimension is treated similarly. Normalized coordinates</span>
+<a name="line-15088"></a>                                    <span class='hs-comment'>-- outside the `[0, 1]` range are allowed, in which case we use</span>
+<a name="line-15089"></a>                                    <span class='hs-comment'>-- `extrapolation_value` to extrapolate the input image values.</span>
+<a name="line-15090"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __box_ind__: A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.</span>
+<a name="line-15091"></a>                                             <span class='hs-comment'>-- The value of `box_ind[i]` specifies the image that the `i`-th box refers to.</span>
+<a name="line-15092"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __crop_size__: A 1-D tensor of 2 elements, `size = [crop_height, crop_width]`. All</span>
+<a name="line-15093"></a>                                             <span class='hs-comment'>-- cropped image patches are resized to this size. The aspect ratio of the image</span>
+<a name="line-15094"></a>                                             <span class='hs-comment'>-- content is not preserved. Both `crop_height` and `crop_width` need to be</span>
+<a name="line-15095"></a>                                             <span class='hs-comment'>-- positive.</span>
+<a name="line-15096"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __crops__: A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`.</span>
+<a name="line-15097"></a><span class='hs-definition'>cropAndResize</span> <span class='hs-varid'>image</span> <span class='hs-varid'>boxes</span> <span class='hs-varid'>box_ind</span> <span class='hs-varid'>crop_size</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-15098"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"CropAndResize"</span>
+<a name="line-15099"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-15100"></a>        <span class='hs-varid'>image</span> <span class='hs-varid'>boxes</span> <span class='hs-varid'>box_ind</span> <span class='hs-varid'>crop_size</span>
+<a name="line-15101"></a><span class='hs-comment'>{-
+<a name="line-15102"></a>attr {
+<a name="line-15103"></a>  allowed_values {
+<a name="line-15104"></a>    list {
+<a name="line-15105"></a>      type: DT_UINT8
+<a name="line-15106"></a>      type: DT_INT8
+<a name="line-15107"></a>      type: DT_INT16
+<a name="line-15108"></a>      type: DT_INT32
+<a name="line-15109"></a>      type: DT_INT64
+<a name="line-15110"></a>      type: DT_HALF
+<a name="line-15111"></a>      type: DT_FLOAT
+<a name="line-15112"></a>      type: DT_DOUBLE
+<a name="line-15113"></a>    }
+<a name="line-15114"></a>  }
+<a name="line-15115"></a>  name: "T"
+<a name="line-15116"></a>  type: "type"
+<a name="line-15117"></a>}
+<a name="line-15118"></a>attr {
+<a name="line-15119"></a>  allowed_values { list { s: "bilinear" } }
+<a name="line-15120"></a>  default_value { s: "bilinear" }
+<a name="line-15121"></a>  description: "A string specifying the interpolation method. Only \'bilinear\' is\nsupported for now."
+<a name="line-15122"></a>  name: "method"
+<a name="line-15123"></a>  type: "string"
+<a name="line-15124"></a>}
+<a name="line-15125"></a>attr {
+<a name="line-15126"></a>  default_value { f: 0.0 }
+<a name="line-15127"></a>  description: "Value used for extrapolation, when applicable."
+<a name="line-15128"></a>  name: "extrapolation_value"
+<a name="line-15129"></a>  type: "float"
+<a name="line-15130"></a>}
+<a name="line-15131"></a>input_arg {
+<a name="line-15132"></a>  description: "A 4-D tensor of shape `[batch, image_height, image_width, depth]`.\nBoth `image_height` and `image_width` need to be positive."
+<a name="line-15133"></a>  name: "image"
+<a name="line-15134"></a>  type_attr: "T"
+<a name="line-15135"></a>}
+<a name="line-15136"></a>input_arg {
+<a name="line-15137"></a>  description: "A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor\nspecifies the coordinates of a box in the `box_ind[i]` image and is specified\nin normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of\n`y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the\n`[0, 1]` interval of normalized image height is mapped to\n`[0, image_height - 1] in image height coordinates. We do allow y1 &gt; y2, in\nwhich case the sampled crop is an up-down flipped version of the original\nimage. The width dimension is treated similarly. Normalized coordinates\noutside the `[0, 1]` range are allowed, in which case we use\n`extrapolation_value` to extrapolate the input image values."
+<a name="line-15138"></a>  name: "boxes"
+<a name="line-15139"></a>  type: DT_FLOAT
+<a name="line-15140"></a>}
+<a name="line-15141"></a>input_arg {
+<a name="line-15142"></a>  description: "A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.\nThe value of `box_ind[i]` specifies the image that the `i`-th box refers to."
+<a name="line-15143"></a>  name: "box_ind"
+<a name="line-15144"></a>  type: DT_INT32
+<a name="line-15145"></a>}
+<a name="line-15146"></a>input_arg {
+<a name="line-15147"></a>  description: "A 1-D tensor of 2 elements, `size = [crop_height, crop_width]`. All\ncropped image patches are resized to this size. The aspect ratio of the image\ncontent is not preserved. Both `crop_height` and `crop_width` need to be\npositive."
+<a name="line-15148"></a>  name: "crop_size"
+<a name="line-15149"></a>  type: DT_INT32
+<a name="line-15150"></a>}
+<a name="line-15151"></a>output_arg {
+<a name="line-15152"></a>  description: "A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`."
+<a name="line-15153"></a>  name: "crops"
+<a name="line-15154"></a>  type: DT_FLOAT
+<a name="line-15155"></a>}
+<a name="line-15156"></a>-}</span>
+<a name="line-15157"></a>
+<a name="line-15158"></a><span class='hs-comment'>-- | Computes gradients of the maxpooling function.</span>
+<a name="line-15159"></a>
+<a name="line-15160"></a><a name="maxPoolGrad"></a><span class='hs-definition'>maxPoolGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-15161"></a>                                                          <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-15162"></a>               <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __orig_input__: The original input tensor.</span>
+<a name="line-15163"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __orig_output__: The original output tensor.</span>
+<a name="line-15164"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: 4-D.  Gradients w.r.t. the output of `max_pool`.</span>
+<a name="line-15165"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Gradients w.r.t. the input to `max_pool`.</span>
+<a name="line-15166"></a><span class='hs-definition'>maxPoolGrad</span> <span class='hs-varid'>orig_input</span> <span class='hs-varid'>orig_output</span> <span class='hs-varid'>grad</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-15167"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MaxPoolGrad"</span>
+<a name="line-15168"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-15169"></a>        <span class='hs-varid'>orig_input</span> <span class='hs-varid'>orig_output</span> <span class='hs-varid'>grad</span>
+<a name="line-15170"></a><span class='hs-comment'>{-
+<a name="line-15171"></a>attr {
+<a name="line-15172"></a>  description: "The size of the window for each dimension of the input tensor."
+<a name="line-15173"></a>  has_minimum: true
+<a name="line-15174"></a>  minimum: 4
+<a name="line-15175"></a>  name: "ksize"
+<a name="line-15176"></a>  type: "list(int)"
+<a name="line-15177"></a>}
+<a name="line-15178"></a>attr {
+<a name="line-15179"></a>  description: "The stride of the sliding window for each dimension of the\ninput tensor."
+<a name="line-15180"></a>  has_minimum: true
+<a name="line-15181"></a>  minimum: 4
+<a name="line-15182"></a>  name: "strides"
+<a name="line-15183"></a>  type: "list(int)"
+<a name="line-15184"></a>}
+<a name="line-15185"></a>attr {
+<a name="line-15186"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-15187"></a>  description: "The type of padding algorithm to use."
+<a name="line-15188"></a>  name: "padding"
+<a name="line-15189"></a>  type: "string"
+<a name="line-15190"></a>}
+<a name="line-15191"></a>attr {
+<a name="line-15192"></a>  allowed_values { list { s: "NHWC" s: "NCHW" } }
+<a name="line-15193"></a>  default_value { s: "NHWC" }
+<a name="line-15194"></a>  description: "Specify the data format of the input and output data. With the\ndefault format \"NHWC\", the data is stored in the order of:\n    [batch, in_height, in_width, in_channels].\nAlternatively, the format could be \"NCHW\", the data storage order of:\n    [batch, in_channels, in_height, in_width]."
+<a name="line-15195"></a>  name: "data_format"
+<a name="line-15196"></a>  type: "string"
+<a name="line-15197"></a>}
+<a name="line-15198"></a>attr {
+<a name="line-15199"></a>  allowed_values { list { type: DT_FLOAT type: DT_HALF } }
+<a name="line-15200"></a>  default_value { type: DT_FLOAT }
+<a name="line-15201"></a>  name: "T"
+<a name="line-15202"></a>  type: "type"
+<a name="line-15203"></a>}
+<a name="line-15204"></a>input_arg {
+<a name="line-15205"></a>  description: "The original input tensor."
+<a name="line-15206"></a>  name: "orig_input"
+<a name="line-15207"></a>  type_attr: "T"
+<a name="line-15208"></a>}
+<a name="line-15209"></a>input_arg {
+<a name="line-15210"></a>  description: "The original output tensor."
+<a name="line-15211"></a>  name: "orig_output"
+<a name="line-15212"></a>  type_attr: "T"
+<a name="line-15213"></a>}
+<a name="line-15214"></a>input_arg {
+<a name="line-15215"></a>  description: "4-D.  Gradients w.r.t. the output of `max_pool`."
+<a name="line-15216"></a>  name: "grad"
+<a name="line-15217"></a>  type_attr: "T"
+<a name="line-15218"></a>}
+<a name="line-15219"></a>output_arg {
+<a name="line-15220"></a>  description: "Gradients w.r.t. the input to `max_pool`."
+<a name="line-15221"></a>  name: "output"
+<a name="line-15222"></a>  type_attr: "T"
+<a name="line-15223"></a>}
+<a name="line-15224"></a>-}</span>
+<a name="line-15225"></a>
+<a name="line-15226"></a><a name="fusedResizeAndPadConv2D"></a><span class='hs-comment'>-- | Performs a resize and padding as a preprocess during a convolution.</span>
+<a name="line-15227"></a><span class='hs-comment'>--</span>
+<a name="line-15228"></a><span class='hs-comment'>-- It's often possible to do spatial transformations more efficiently as part of</span>
+<a name="line-15229"></a><span class='hs-comment'>-- the packing stage of a convolution, so this op allows for an optimized</span>
+<a name="line-15230"></a><span class='hs-comment'>-- implementation where these stages are fused together. This prevents the need to</span>
+<a name="line-15231"></a><span class='hs-comment'>-- write out the intermediate results as whole tensors, reducing memory pressure,</span>
+<a name="line-15232"></a><span class='hs-comment'>-- and we can get some latency gains by merging the transformation calculations.</span>
+<a name="line-15233"></a><span class='hs-comment'>-- The data_format attribute for Conv2D isn't supported by this op, and defaults to</span>
+<a name="line-15234"></a><span class='hs-comment'>-- 'NHWC' order.</span>
+<a name="line-15235"></a><span class='hs-comment'>-- Internally this op uses a single per-graph scratch buffer, which means that it</span>
+<a name="line-15236"></a><span class='hs-comment'>-- will block if multiple versions are being run in parallel. This is because this</span>
+<a name="line-15237"></a><span class='hs-comment'>-- operator is primarily an optimization to minimize memory usage.</span>
+<a name="line-15238"></a><span class='hs-definition'>fusedResizeAndPadConv2D</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-15239"></a>                                                   <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-15240"></a>                                                           <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-15241"></a>                           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: 4-D with shape `[batch, in_height, in_width, in_channels]`.</span>
+<a name="line-15242"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __size__: A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The</span>
+<a name="line-15243"></a>                                                       <span class='hs-comment'>-- new size for the images.</span>
+<a name="line-15244"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __paddings__: A two-column matrix specifying the padding sizes. The number of</span>
+<a name="line-15245"></a>                                                       <span class='hs-comment'>-- rows must be the same as the rank of `input`.</span>
+<a name="line-15246"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __filter__: 4-D with shape</span>
+<a name="line-15247"></a>                                          <span class='hs-comment'>-- `[filter_height, filter_width, in_channels, out_channels]`.</span>
+<a name="line-15248"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-15249"></a><span class='hs-definition'>fusedResizeAndPadConv2D</span> <span class='hs-varid'>input</span> <span class='hs-varid'>size</span> <span class='hs-varid'>paddings</span> <span class='hs-varid'>filter</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-15250"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"FusedResizeAndPadConv2D"</span>
+<a name="line-15251"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-15252"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>size</span> <span class='hs-varid'>paddings</span> <span class='hs-varid'>filter</span>
+<a name="line-15253"></a><span class='hs-comment'>{-
+<a name="line-15254"></a>attr {
+<a name="line-15255"></a>  allowed_values {
+<a name="line-15256"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-15257"></a>  }
+<a name="line-15258"></a>  name: "T"
+<a name="line-15259"></a>  type: "type"
+<a name="line-15260"></a>}
+<a name="line-15261"></a>attr {
+<a name="line-15262"></a>  default_value { b: false }
+<a name="line-15263"></a>  description: "If true, rescale input by (new_height - 1) / (height - 1),\nwhich exactly aligns the 4 corners of images and resized images. If false, rescale\nby new_height / height. Treat similarly the width dimension."
+<a name="line-15264"></a>  name: "resize_align_corners"
+<a name="line-15265"></a>  type: "bool"
+<a name="line-15266"></a>}
+<a name="line-15267"></a>attr {
+<a name="line-15268"></a>  allowed_values { list { s: "REFLECT" s: "SYMMETRIC" } }
+<a name="line-15269"></a>  name: "mode"
+<a name="line-15270"></a>  type: "string"
+<a name="line-15271"></a>}
+<a name="line-15272"></a>attr {
+<a name="line-15273"></a>  description: "1-D of length 4.  The stride of the sliding window for each dimension\nof `input`. Must be in the same order as the dimension specified with format."
+<a name="line-15274"></a>  name: "strides"
+<a name="line-15275"></a>  type: "list(int)"
+<a name="line-15276"></a>}
+<a name="line-15277"></a>attr {
+<a name="line-15278"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-15279"></a>  description: "The type of padding algorithm to use."
+<a name="line-15280"></a>  name: "padding"
+<a name="line-15281"></a>  type: "string"
+<a name="line-15282"></a>}
+<a name="line-15283"></a>input_arg {
+<a name="line-15284"></a>  description: "4-D with shape `[batch, in_height, in_width, in_channels]`."
+<a name="line-15285"></a>  name: "input"
+<a name="line-15286"></a>  type_attr: "T"
+<a name="line-15287"></a>}
+<a name="line-15288"></a>input_arg {
+<a name="line-15289"></a>  description: "A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The\nnew size for the images."
+<a name="line-15290"></a>  name: "size"
+<a name="line-15291"></a>  type: DT_INT32
+<a name="line-15292"></a>}
+<a name="line-15293"></a>input_arg {
+<a name="line-15294"></a>  description: "A two-column matrix specifying the padding sizes. The number of\nrows must be the same as the rank of `input`."
+<a name="line-15295"></a>  name: "paddings"
+<a name="line-15296"></a>  type: DT_INT32
+<a name="line-15297"></a>}
+<a name="line-15298"></a>input_arg {
+<a name="line-15299"></a>  description: "4-D with shape\n`[filter_height, filter_width, in_channels, out_channels]`."
+<a name="line-15300"></a>  name: "filter"
+<a name="line-15301"></a>  type_attr: "T"
+<a name="line-15302"></a>}
+<a name="line-15303"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-15304"></a>-}</span>
+<a name="line-15305"></a>
+<a name="line-15306"></a><a name="randomUniform"></a><span class='hs-comment'>-- | Outputs random values from a uniform distribution.</span>
+<a name="line-15307"></a><span class='hs-comment'>--</span>
+<a name="line-15308"></a><span class='hs-comment'>-- The generated values follow a uniform distribution in the range `[0, 1)`. The</span>
+<a name="line-15309"></a><span class='hs-comment'>-- lower bound 0 is included in the range, while the upper bound 1 is excluded.</span>
+<a name="line-15310"></a><span class='hs-definition'>randomUniform</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-15311"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-15312"></a>                                      <span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>,</span>
+<a name="line-15313"></a>                                      <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-15314"></a>                                              <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-15315"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __shape__: The shape of the output tensor.</span>
+<a name="line-15316"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __output__: A tensor of the specified shape filled with uniform random values.</span>
+<a name="line-15317"></a><span class='hs-definition'>randomUniform</span> <span class='hs-varid'>shape</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-15318"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RandomUniform"</span>
+<a name="line-15319"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-15320"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-15321"></a>        <span class='hs-varid'>shape</span>
+<a name="line-15322"></a><span class='hs-comment'>{-
+<a name="line-15323"></a>attr {
+<a name="line-15324"></a>  default_value { i: 0 }
+<a name="line-15325"></a>  description: "If either `seed` or `seed2` are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-15326"></a>  name: "seed"
+<a name="line-15327"></a>  type: "int"
+<a name="line-15328"></a>}
+<a name="line-15329"></a>attr {
+<a name="line-15330"></a>  default_value { i: 0 }
+<a name="line-15331"></a>  description: "A second seed to avoid seed collision."
+<a name="line-15332"></a>  name: "seed2"
+<a name="line-15333"></a>  type: "int"
+<a name="line-15334"></a>}
+<a name="line-15335"></a>attr {
+<a name="line-15336"></a>  allowed_values {
+<a name="line-15337"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-15338"></a>  }
+<a name="line-15339"></a>  description: "The type of the output."
+<a name="line-15340"></a>  name: "dtype"
+<a name="line-15341"></a>  type: "type"
+<a name="line-15342"></a>}
+<a name="line-15343"></a>attr {
+<a name="line-15344"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-15345"></a>  name: "T"
+<a name="line-15346"></a>  type: "type"
+<a name="line-15347"></a>}
+<a name="line-15348"></a>input_arg {
+<a name="line-15349"></a>  description: "The shape of the output tensor."
+<a name="line-15350"></a>  name: "shape"
+<a name="line-15351"></a>  type_attr: "T"
+<a name="line-15352"></a>}
+<a name="line-15353"></a>output_arg {
+<a name="line-15354"></a>  description: "A tensor of the specified shape filled with uniform random values."
+<a name="line-15355"></a>  name: "output"
+<a name="line-15356"></a>  type_attr: "dtype"
+<a name="line-15357"></a>}
+<a name="line-15358"></a>-}</span>
+<a name="line-15359"></a>
+<a name="line-15360"></a><a name="depthwiseConv2dNative"></a><span class='hs-comment'>-- | Computes a 2-D depthwise convolution given 4-D `input` and `filter` tensors.</span>
+<a name="line-15361"></a><span class='hs-comment'>--</span>
+<a name="line-15362"></a><span class='hs-comment'>-- Given an input tensor of shape `[batch, in_height, in_width, in_channels]`</span>
+<a name="line-15363"></a><span class='hs-comment'>-- and a filter / kernel tensor of shape</span>
+<a name="line-15364"></a><span class='hs-comment'>-- `[filter_height, filter_width, in_channels, channel_multiplier]`, containing</span>
+<a name="line-15365"></a><span class='hs-comment'>-- `in_channels` convolutional filters of depth 1, `depthwise_conv2d` applies</span>
+<a name="line-15366"></a><span class='hs-comment'>-- a different filter to each input channel (expanding from 1 channel to</span>
+<a name="line-15367"></a><span class='hs-comment'>-- `channel_multiplier` channels for each), then concatenates the results</span>
+<a name="line-15368"></a><span class='hs-comment'>-- together. Thus, the output has `in_channels * channel_multiplier` channels.</span>
+<a name="line-15369"></a><span class='hs-comment'>-- </span>
+<a name="line-15370"></a><span class='hs-comment'>-- for k in 0..in_channels-1</span>
+<a name="line-15371"></a><span class='hs-comment'>--   for q in 0..channel_multiplier-1</span>
+<a name="line-15372"></a><span class='hs-comment'>--     output[b, i, j, k * channel_multiplier + q] =</span>
+<a name="line-15373"></a><span class='hs-comment'>--       sum_{di, dj} input[b, strides[1] * i + di, strides[2] * j + dj, k] *</span>
+<a name="line-15374"></a><span class='hs-comment'>--                         filter[di, dj, k, q]</span>
+<a name="line-15375"></a><span class='hs-comment'>-- </span>
+<a name="line-15376"></a><span class='hs-comment'>-- Must have `strides[0] = strides[3] = 1`.  For the most common case of the same</span>
+<a name="line-15377"></a><span class='hs-comment'>-- horizontal and vertices strides, `strides = [1, stride, stride, 1]`.</span>
+<a name="line-15378"></a><span class='hs-definition'>depthwiseConv2dNative</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-15379"></a>                                                                 <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-15380"></a>                         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-15381"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __filter__</span>
+<a name="line-15382"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-15383"></a><span class='hs-definition'>depthwiseConv2dNative</span> <span class='hs-varid'>input</span> <span class='hs-varid'>filter</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-15384"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DepthwiseConv2dNative"</span>
+<a name="line-15385"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-15386"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>filter</span>
+<a name="line-15387"></a><span class='hs-comment'>{-
+<a name="line-15388"></a>attr {
+<a name="line-15389"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-15390"></a>  name: "T"
+<a name="line-15391"></a>  type: "type"
+<a name="line-15392"></a>}
+<a name="line-15393"></a>attr {
+<a name="line-15394"></a>  description: "1-D of length 4.  The stride of the sliding window for each dimension\nof `input`."
+<a name="line-15395"></a>  name: "strides"
+<a name="line-15396"></a>  type: "list(int)"
+<a name="line-15397"></a>}
+<a name="line-15398"></a>attr {
+<a name="line-15399"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-15400"></a>  description: "The type of padding algorithm to use."
+<a name="line-15401"></a>  name: "padding"
+<a name="line-15402"></a>  type: "string"
+<a name="line-15403"></a>}
+<a name="line-15404"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-15405"></a>input_arg { name: "filter" type_attr: "T" }
+<a name="line-15406"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-15407"></a>-}</span>
+<a name="line-15408"></a>
+<a name="line-15409"></a><span class='hs-comment'>-- | var: Should be from a Variable().</span>
+<a name="line-15410"></a>
+<a name="line-15411"></a><a name="sparseApplyAdadelta"></a><span class='hs-definition'>sparseApplyAdadelta</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>v8</span> <span class='hs-varid'>t</span>
+<a name="line-15412"></a>                       <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-15413"></a>                                   <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-15414"></a>                                           <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-15415"></a>                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-15416"></a>                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-15417"></a>                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-15418"></a>                                           <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-15419"></a>                                   <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-15420"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-15421"></a>                       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__</span>
+<a name="line-15422"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __accum__: Should be from a Variable().</span>
+<a name="line-15423"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __accum_update__: : Should be from a Variable().</span>
+<a name="line-15424"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr__: Learning rate. Must be a scalar.</span>
+<a name="line-15425"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __rho__: Decay factor. Must be a scalar.</span>
+<a name="line-15426"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __epsilon__: Constant factor. Must be a scalar.</span>
+<a name="line-15427"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: The gradient.</span>
+<a name="line-15428"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v8</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __indices__: A vector of indices into the first dimension of var and accum.</span>
+<a name="line-15429"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-15430"></a><span class='hs-definition'>sparseApplyAdadelta</span> <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>accum_update</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>rho</span> <span class='hs-varid'>epsilon</span> <span class='hs-varid'>grad</span>
+<a name="line-15431"></a>                    <span class='hs-varid'>indices</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-15432"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseApplyAdadelta"</span>
+<a name="line-15433"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-15434"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-15435"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>accum_update</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>rho</span> <span class='hs-varid'>epsilon</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>indices</span>
+<a name="line-15436"></a><span class='hs-comment'>{-
+<a name="line-15437"></a>attr {
+<a name="line-15438"></a>  allowed_values {
+<a name="line-15439"></a>    list {
+<a name="line-15440"></a>      type: DT_FLOAT
+<a name="line-15441"></a>      type: DT_DOUBLE
+<a name="line-15442"></a>      type: DT_INT64
+<a name="line-15443"></a>      type: DT_INT32
+<a name="line-15444"></a>      type: DT_UINT8
+<a name="line-15445"></a>      type: DT_UINT16
+<a name="line-15446"></a>      type: DT_INT16
+<a name="line-15447"></a>      type: DT_INT8
+<a name="line-15448"></a>      type: DT_COMPLEX64
+<a name="line-15449"></a>      type: DT_COMPLEX128
+<a name="line-15450"></a>      type: DT_QINT8
+<a name="line-15451"></a>      type: DT_QUINT8
+<a name="line-15452"></a>      type: DT_QINT32
+<a name="line-15453"></a>      type: DT_HALF
+<a name="line-15454"></a>    }
+<a name="line-15455"></a>  }
+<a name="line-15456"></a>  name: "T"
+<a name="line-15457"></a>  type: "type"
+<a name="line-15458"></a>}
+<a name="line-15459"></a>attr {
+<a name="line-15460"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-15461"></a>  name: "Tindices"
+<a name="line-15462"></a>  type: "type"
+<a name="line-15463"></a>}
+<a name="line-15464"></a>attr {
+<a name="line-15465"></a>  default_value { b: false }
+<a name="line-15466"></a>  description: "If True, updating of the var and accum tensors will be protected by\na lock; otherwise the behavior is undefined, but may exhibit less contention."
+<a name="line-15467"></a>  name: "use_locking"
+<a name="line-15468"></a>  type: "bool"
+<a name="line-15469"></a>}
+<a name="line-15470"></a>input_arg { is_ref: true name: "var" type_attr: "T" }
+<a name="line-15471"></a>input_arg {
+<a name="line-15472"></a>  description: "Should be from a Variable()."
+<a name="line-15473"></a>  is_ref: true
+<a name="line-15474"></a>  name: "accum"
+<a name="line-15475"></a>  type_attr: "T"
+<a name="line-15476"></a>}
+<a name="line-15477"></a>input_arg {
+<a name="line-15478"></a>  description: ": Should be from a Variable()."
+<a name="line-15479"></a>  is_ref: true
+<a name="line-15480"></a>  name: "accum_update"
+<a name="line-15481"></a>  type_attr: "T"
+<a name="line-15482"></a>}
+<a name="line-15483"></a>input_arg {
+<a name="line-15484"></a>  description: "Learning rate. Must be a scalar."
+<a name="line-15485"></a>  name: "lr"
+<a name="line-15486"></a>  type_attr: "T"
+<a name="line-15487"></a>}
+<a name="line-15488"></a>input_arg {
+<a name="line-15489"></a>  description: "Decay factor. Must be a scalar."
+<a name="line-15490"></a>  name: "rho"
+<a name="line-15491"></a>  type_attr: "T"
+<a name="line-15492"></a>}
+<a name="line-15493"></a>input_arg {
+<a name="line-15494"></a>  description: "Constant factor. Must be a scalar."
+<a name="line-15495"></a>  name: "epsilon"
+<a name="line-15496"></a>  type_attr: "T"
+<a name="line-15497"></a>}
+<a name="line-15498"></a>input_arg {
+<a name="line-15499"></a>  description: "The gradient." name: "grad" type_attr: "T"
+<a name="line-15500"></a>}
+<a name="line-15501"></a>input_arg {
+<a name="line-15502"></a>  description: "A vector of indices into the first dimension of var and accum."
+<a name="line-15503"></a>  name: "indices"
+<a name="line-15504"></a>  type_attr: "Tindices"
+<a name="line-15505"></a>}
+<a name="line-15506"></a>output_arg {
+<a name="line-15507"></a>  description: "Same as \"var\"."
+<a name="line-15508"></a>  is_ref: true
+<a name="line-15509"></a>  name: "out"
+<a name="line-15510"></a>  type_attr: "T"
+<a name="line-15511"></a>}
+<a name="line-15512"></a>-}</span>
+<a name="line-15513"></a>
+<a name="line-15514"></a><span class='hs-comment'>-- | Computes the gradients of depthwise convolution with respect to the filter.</span>
+<a name="line-15515"></a>
+<a name="line-15516"></a><a name="depthwiseConv2dNativeBackpropFilter"></a><span class='hs-definition'>depthwiseConv2dNativeBackpropFilter</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-15517"></a>                                                            <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-15518"></a>                                                                    <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-15519"></a>                                       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: 4-D with shape `[batch, in_height, in_width, in_channels]`.</span>
+<a name="line-15520"></a>                                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __filter_sizes__: An integer vector representing the tensor shape of `filter`,</span>
+<a name="line-15521"></a>                                                                   <span class='hs-comment'>-- where `filter` is a 4-D</span>
+<a name="line-15522"></a>                                                                   <span class='hs-comment'>-- `[filter_height, filter_width, in_channels, depthwise_multiplier]` tensor.</span>
+<a name="line-15523"></a>                                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out_backprop__: 4-D with shape `[batch, out_height, out_width, out_channels]`.</span>
+<a name="line-15524"></a>                                                      <span class='hs-comment'>-- Gradients w.r.t. the output of the convolution.</span>
+<a name="line-15525"></a>                                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 4-D with shape</span>
+<a name="line-15526"></a>                                       <span class='hs-comment'>-- `[filter_height, filter_width, in_channels, out_channels]`.  Gradient w.r.t.</span>
+<a name="line-15527"></a>                                       <span class='hs-comment'>-- the `filter` input of the convolution.</span>
+<a name="line-15528"></a><span class='hs-definition'>depthwiseConv2dNativeBackpropFilter</span> <span class='hs-varid'>input</span> <span class='hs-varid'>filter_sizes</span>
+<a name="line-15529"></a>                                    <span class='hs-varid'>out_backprop</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-15530"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DepthwiseConv2dNativeBackpropFilter"</span>
+<a name="line-15531"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-15532"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>filter_sizes</span> <span class='hs-varid'>out_backprop</span>
+<a name="line-15533"></a><span class='hs-comment'>{-
+<a name="line-15534"></a>attr {
+<a name="line-15535"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-15536"></a>  name: "T"
+<a name="line-15537"></a>  type: "type"
+<a name="line-15538"></a>}
+<a name="line-15539"></a>attr {
+<a name="line-15540"></a>  description: "The stride of the sliding window for each dimension of the input\nof the convolution."
+<a name="line-15541"></a>  name: "strides"
+<a name="line-15542"></a>  type: "list(int)"
+<a name="line-15543"></a>}
+<a name="line-15544"></a>attr {
+<a name="line-15545"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-15546"></a>  description: "The type of padding algorithm to use."
+<a name="line-15547"></a>  name: "padding"
+<a name="line-15548"></a>  type: "string"
+<a name="line-15549"></a>}
+<a name="line-15550"></a>input_arg {
+<a name="line-15551"></a>  description: "4-D with shape `[batch, in_height, in_width, in_channels]`."
+<a name="line-15552"></a>  name: "input"
+<a name="line-15553"></a>  type_attr: "T"
+<a name="line-15554"></a>}
+<a name="line-15555"></a>input_arg {
+<a name="line-15556"></a>  description: "An integer vector representing the tensor shape of `filter`,\nwhere `filter` is a 4-D\n`[filter_height, filter_width, in_channels, depthwise_multiplier]` tensor."
+<a name="line-15557"></a>  name: "filter_sizes"
+<a name="line-15558"></a>  type: DT_INT32
+<a name="line-15559"></a>}
+<a name="line-15560"></a>input_arg {
+<a name="line-15561"></a>  description: "4-D with shape `[batch, out_height, out_width, out_channels]`.\nGradients w.r.t. the output of the convolution."
+<a name="line-15562"></a>  name: "out_backprop"
+<a name="line-15563"></a>  type_attr: "T"
+<a name="line-15564"></a>}
+<a name="line-15565"></a>output_arg {
+<a name="line-15566"></a>  description: "4-D with shape\n`[filter_height, filter_width, in_channels, out_channels]`.  Gradient w.r.t.\nthe `filter` input of the convolution."
+<a name="line-15567"></a>  name: "output"
+<a name="line-15568"></a>  type_attr: "T"
+<a name="line-15569"></a>}
+<a name="line-15570"></a>-}</span>
+<a name="line-15571"></a>
+<a name="line-15572"></a><a name="conv3D"></a><span class='hs-comment'>-- | Computes a 3-D convolution given 5-D `input` and `filter` tensors.</span>
+<a name="line-15573"></a><span class='hs-comment'>--</span>
+<a name="line-15574"></a><span class='hs-comment'>-- In signal processing, cross-correlation is a measure of similarity of</span>
+<a name="line-15575"></a><span class='hs-comment'>-- two waveforms as a function of a time-lag applied to one of them. This</span>
+<a name="line-15576"></a><span class='hs-comment'>-- is also known as a sliding dot product or sliding inner-product.</span>
+<a name="line-15577"></a><span class='hs-comment'>-- </span>
+<a name="line-15578"></a><span class='hs-comment'>-- Our Conv3D implements a form of cross-correlation.</span>
+<a name="line-15579"></a><span class='hs-definition'>conv3D</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-15580"></a>                                                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-15581"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-15582"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-15583"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-15584"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-15585"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-15586"></a>                                                  <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-15587"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Shape `[batch, in_depth, in_height, in_width, in_channels]`.</span>
+<a name="line-15588"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __filter__: Shape `[filter_depth, filter_height, filter_width, in_channels,</span>
+<a name="line-15589"></a>                         <span class='hs-comment'>-- out_channels]`. `in_channels` must match between `input` and `filter`.</span>
+<a name="line-15590"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-15591"></a><span class='hs-definition'>conv3D</span> <span class='hs-varid'>input</span> <span class='hs-varid'>filter</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-15592"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Conv3D"</span>
+<a name="line-15593"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-15594"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>filter</span>
+<a name="line-15595"></a><span class='hs-comment'>{-
+<a name="line-15596"></a>attr {
+<a name="line-15597"></a>  allowed_values {
+<a name="line-15598"></a>    list {
+<a name="line-15599"></a>      type: DT_FLOAT
+<a name="line-15600"></a>      type: DT_DOUBLE
+<a name="line-15601"></a>      type: DT_INT64
+<a name="line-15602"></a>      type: DT_INT32
+<a name="line-15603"></a>      type: DT_UINT8
+<a name="line-15604"></a>      type: DT_UINT16
+<a name="line-15605"></a>      type: DT_INT16
+<a name="line-15606"></a>      type: DT_INT8
+<a name="line-15607"></a>      type: DT_COMPLEX64
+<a name="line-15608"></a>      type: DT_COMPLEX128
+<a name="line-15609"></a>      type: DT_QINT8
+<a name="line-15610"></a>      type: DT_QUINT8
+<a name="line-15611"></a>      type: DT_QINT32
+<a name="line-15612"></a>      type: DT_HALF
+<a name="line-15613"></a>    }
+<a name="line-15614"></a>  }
+<a name="line-15615"></a>  name: "T"
+<a name="line-15616"></a>  type: "type"
+<a name="line-15617"></a>}
+<a name="line-15618"></a>attr {
+<a name="line-15619"></a>  description: "1-D tensor of length 5. The stride of the sliding window for each\ndimension of `input`. Must have `strides[0] = strides[4] = 1`."
+<a name="line-15620"></a>  has_minimum: true
+<a name="line-15621"></a>  minimum: 5
+<a name="line-15622"></a>  name: "strides"
+<a name="line-15623"></a>  type: "list(int)"
+<a name="line-15624"></a>}
+<a name="line-15625"></a>attr {
+<a name="line-15626"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-15627"></a>  description: "The type of padding algorithm to use."
+<a name="line-15628"></a>  name: "padding"
+<a name="line-15629"></a>  type: "string"
+<a name="line-15630"></a>}
+<a name="line-15631"></a>input_arg {
+<a name="line-15632"></a>  description: "Shape `[batch, in_depth, in_height, in_width, in_channels]`."
+<a name="line-15633"></a>  name: "input"
+<a name="line-15634"></a>  type_attr: "T"
+<a name="line-15635"></a>}
+<a name="line-15636"></a>input_arg {
+<a name="line-15637"></a>  description: "Shape `[filter_depth, filter_height, filter_width, in_channels,\nout_channels]`. `in_channels` must match between `input` and `filter`."
+<a name="line-15638"></a>  name: "filter"
+<a name="line-15639"></a>  type_attr: "T"
+<a name="line-15640"></a>}
+<a name="line-15641"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-15642"></a>-}</span>
+<a name="line-15643"></a>
+<a name="line-15644"></a><a name="greaterEqual"></a><span class='hs-comment'>-- | Returns the truth value of (x &gt;= y) element-wise.</span>
+<a name="line-15645"></a><span class='hs-comment'>--</span>
+<a name="line-15646"></a><span class='hs-comment'>-- *NOTE*: `GreaterEqual` supports broadcasting. More about broadcasting</span>
+<a name="line-15647"></a><span class='hs-comment'>-- [here](<a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)">http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)</a></span>
+<a name="line-15648"></a><span class='hs-definition'>greaterEqual</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-15649"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-15650"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-15651"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-15652"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-15653"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-15654"></a>                                                        <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-15655"></a>                <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-15656"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-15657"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-15658"></a><span class='hs-definition'>greaterEqual</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-15659"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"GreaterEqual"</span>
+<a name="line-15660"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-15661"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-15662"></a><span class='hs-comment'>{-
+<a name="line-15663"></a>attr {
+<a name="line-15664"></a>  allowed_values {
+<a name="line-15665"></a>    list {
+<a name="line-15666"></a>      type: DT_FLOAT
+<a name="line-15667"></a>      type: DT_DOUBLE
+<a name="line-15668"></a>      type: DT_INT32
+<a name="line-15669"></a>      type: DT_INT64
+<a name="line-15670"></a>      type: DT_UINT8
+<a name="line-15671"></a>      type: DT_INT16
+<a name="line-15672"></a>      type: DT_INT8
+<a name="line-15673"></a>      type: DT_UINT16
+<a name="line-15674"></a>      type: DT_HALF
+<a name="line-15675"></a>    }
+<a name="line-15676"></a>  }
+<a name="line-15677"></a>  name: "T"
+<a name="line-15678"></a>  type: "type"
+<a name="line-15679"></a>}
+<a name="line-15680"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-15681"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-15682"></a>output_arg { name: "z" type: DT_BOOL }
+<a name="line-15683"></a>-}</span>
+<a name="line-15684"></a>
+<a name="line-15685"></a><a name="sparseDenseCwiseAdd"></a><span class='hs-comment'>-- | Adds up a SparseTensor and a dense Tensor, using these special rules:</span>
+<a name="line-15686"></a><span class='hs-comment'>--</span>
+<a name="line-15687"></a><span class='hs-comment'>-- (1) Broadcasts the dense side to have the same shape as the sparse side, if</span>
+<a name="line-15688"></a><span class='hs-comment'>--     eligible;</span>
+<a name="line-15689"></a><span class='hs-comment'>-- (2) Then, only the dense values pointed to by the indices of the SparseTensor</span>
+<a name="line-15690"></a><span class='hs-comment'>--     participate in the cwise addition.</span>
+<a name="line-15691"></a><span class='hs-comment'>-- </span>
+<a name="line-15692"></a><span class='hs-comment'>-- By these rules, the result is a logical SparseTensor with exactly the same</span>
+<a name="line-15693"></a><span class='hs-comment'>-- indices and shape, but possibly with different non-zero values.  The output of</span>
+<a name="line-15694"></a><span class='hs-comment'>-- this Op is the resultant non-zero values.</span>
+<a name="line-15695"></a><span class='hs-definition'>sparseDenseCwiseAdd</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-15696"></a>                                               <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-15697"></a>                                                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-15698"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-15699"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-15700"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-15701"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-15702"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-15703"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-15704"></a>                                                       <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-15705"></a>                       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __sp_indices__: 2-D.  `N x R` matrix with the indices of non-empty values in a</span>
+<a name="line-15706"></a>                                                <span class='hs-comment'>-- SparseTensor, possibly not in canonical ordering.</span>
+<a name="line-15707"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __sp_values__: 1-D.  `N` non-empty values corresponding to `sp_indices`.</span>
+<a name="line-15708"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __sp_shape__: 1-D.  Shape of the input SparseTensor.</span>
+<a name="line-15709"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __dense__: `R`-D.  The dense Tensor operand.</span>
+<a name="line-15710"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 1-D.  The `N` values that are operated on.</span>
+<a name="line-15711"></a><span class='hs-definition'>sparseDenseCwiseAdd</span> <span class='hs-varid'>sp_indices</span> <span class='hs-varid'>sp_values</span> <span class='hs-varid'>sp_shape</span> <span class='hs-varid'>dense</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-15712"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseDenseCwiseAdd"</span>
+<a name="line-15713"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-15714"></a>        <span class='hs-varid'>sp_indices</span> <span class='hs-varid'>sp_values</span> <span class='hs-varid'>sp_shape</span> <span class='hs-varid'>dense</span>
+<a name="line-15715"></a><span class='hs-comment'>{-
+<a name="line-15716"></a>attr {
+<a name="line-15717"></a>  allowed_values {
+<a name="line-15718"></a>    list {
+<a name="line-15719"></a>      type: DT_FLOAT
+<a name="line-15720"></a>      type: DT_DOUBLE
+<a name="line-15721"></a>      type: DT_INT64
+<a name="line-15722"></a>      type: DT_INT32
+<a name="line-15723"></a>      type: DT_UINT8
+<a name="line-15724"></a>      type: DT_UINT16
+<a name="line-15725"></a>      type: DT_INT16
+<a name="line-15726"></a>      type: DT_INT8
+<a name="line-15727"></a>      type: DT_COMPLEX64
+<a name="line-15728"></a>      type: DT_COMPLEX128
+<a name="line-15729"></a>      type: DT_QINT8
+<a name="line-15730"></a>      type: DT_QUINT8
+<a name="line-15731"></a>      type: DT_QINT32
+<a name="line-15732"></a>      type: DT_HALF
+<a name="line-15733"></a>    }
+<a name="line-15734"></a>  }
+<a name="line-15735"></a>  name: "T"
+<a name="line-15736"></a>  type: "type"
+<a name="line-15737"></a>}
+<a name="line-15738"></a>input_arg {
+<a name="line-15739"></a>  description: "2-D.  `N x R` matrix with the indices of non-empty values in a\nSparseTensor, possibly not in canonical ordering."
+<a name="line-15740"></a>  name: "sp_indices"
+<a name="line-15741"></a>  type: DT_INT64
+<a name="line-15742"></a>}
+<a name="line-15743"></a>input_arg {
+<a name="line-15744"></a>  description: "1-D.  `N` non-empty values corresponding to `sp_indices`."
+<a name="line-15745"></a>  name: "sp_values"
+<a name="line-15746"></a>  type_attr: "T"
+<a name="line-15747"></a>}
+<a name="line-15748"></a>input_arg {
+<a name="line-15749"></a>  description: "1-D.  Shape of the input SparseTensor."
+<a name="line-15750"></a>  name: "sp_shape"
+<a name="line-15751"></a>  type: DT_INT64
+<a name="line-15752"></a>}
+<a name="line-15753"></a>input_arg {
+<a name="line-15754"></a>  description: "`R`-D.  The dense Tensor operand."
+<a name="line-15755"></a>  name: "dense"
+<a name="line-15756"></a>  type_attr: "T"
+<a name="line-15757"></a>}
+<a name="line-15758"></a>output_arg {
+<a name="line-15759"></a>  description: "1-D.  The `N` values that are operated on."
+<a name="line-15760"></a>  name: "output"
+<a name="line-15761"></a>  type_attr: "T"
+<a name="line-15762"></a>}
+<a name="line-15763"></a>-}</span>
+<a name="line-15764"></a>
+<a name="line-15765"></a><span class='hs-comment'>-- | Computes the gradients of 3-D convolution with respect to the filter.</span>
+<a name="line-15766"></a>
+<a name="line-15767"></a><a name="conv3DBackpropFilter"></a><span class='hs-definition'>conv3DBackpropFilter</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-15768"></a>                                             <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-15769"></a>                                                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-15770"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-15771"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-15772"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-15773"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-15774"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-15775"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-15776"></a>                                                     <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-15777"></a>                        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Shape `[batch, depth, rows, cols, in_channels]`.</span>
+<a name="line-15778"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __filter__: Shape `[depth, rows, cols, in_channels, out_channels]`.</span>
+<a name="line-15779"></a>                                       <span class='hs-comment'>-- `in_channels` must match between `input` and `filter`.</span>
+<a name="line-15780"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out_backprop__: Backprop signal of shape `[batch, out_depth, out_rows, out_cols,</span>
+<a name="line-15781"></a>                                       <span class='hs-comment'>-- out_channels]`.</span>
+<a name="line-15782"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-15783"></a><span class='hs-definition'>conv3DBackpropFilter</span> <span class='hs-varid'>input</span> <span class='hs-varid'>filter</span> <span class='hs-varid'>out_backprop</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-15784"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Conv3DBackpropFilter"</span>
+<a name="line-15785"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-15786"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>filter</span> <span class='hs-varid'>out_backprop</span>
+<a name="line-15787"></a><span class='hs-comment'>{-
+<a name="line-15788"></a>attr {
+<a name="line-15789"></a>  allowed_values {
+<a name="line-15790"></a>    list {
+<a name="line-15791"></a>      type: DT_FLOAT
+<a name="line-15792"></a>      type: DT_DOUBLE
+<a name="line-15793"></a>      type: DT_INT64
+<a name="line-15794"></a>      type: DT_INT32
+<a name="line-15795"></a>      type: DT_UINT8
+<a name="line-15796"></a>      type: DT_UINT16
+<a name="line-15797"></a>      type: DT_INT16
+<a name="line-15798"></a>      type: DT_INT8
+<a name="line-15799"></a>      type: DT_COMPLEX64
+<a name="line-15800"></a>      type: DT_COMPLEX128
+<a name="line-15801"></a>      type: DT_QINT8
+<a name="line-15802"></a>      type: DT_QUINT8
+<a name="line-15803"></a>      type: DT_QINT32
+<a name="line-15804"></a>      type: DT_HALF
+<a name="line-15805"></a>    }
+<a name="line-15806"></a>  }
+<a name="line-15807"></a>  name: "T"
+<a name="line-15808"></a>  type: "type"
+<a name="line-15809"></a>}
+<a name="line-15810"></a>attr {
+<a name="line-15811"></a>  description: "1-D tensor of length 5. The stride of the sliding window for each\ndimension of `input`. Must have `strides[0] = strides[4] = 1`."
+<a name="line-15812"></a>  has_minimum: true
+<a name="line-15813"></a>  minimum: 5
+<a name="line-15814"></a>  name: "strides"
+<a name="line-15815"></a>  type: "list(int)"
+<a name="line-15816"></a>}
+<a name="line-15817"></a>attr {
+<a name="line-15818"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-15819"></a>  description: "The type of padding algorithm to use."
+<a name="line-15820"></a>  name: "padding"
+<a name="line-15821"></a>  type: "string"
+<a name="line-15822"></a>}
+<a name="line-15823"></a>input_arg {
+<a name="line-15824"></a>  description: "Shape `[batch, depth, rows, cols, in_channels]`."
+<a name="line-15825"></a>  name: "input"
+<a name="line-15826"></a>  type_attr: "T"
+<a name="line-15827"></a>}
+<a name="line-15828"></a>input_arg {
+<a name="line-15829"></a>  description: "Shape `[depth, rows, cols, in_channels, out_channels]`.\n`in_channels` must match between `input` and `filter`."
+<a name="line-15830"></a>  name: "filter"
+<a name="line-15831"></a>  type_attr: "T"
+<a name="line-15832"></a>}
+<a name="line-15833"></a>input_arg {
+<a name="line-15834"></a>  description: "Backprop signal of shape `[batch, out_depth, out_rows, out_cols,\nout_channels]`."
+<a name="line-15835"></a>  name: "out_backprop"
+<a name="line-15836"></a>  type_attr: "T"
+<a name="line-15837"></a>}
+<a name="line-15838"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-15839"></a>-}</span>
+<a name="line-15840"></a>
+<a name="line-15841"></a><span class='hs-comment'>-- | Computes the gradients of 3-D convolution with respect to the input.</span>
+<a name="line-15842"></a>
+<a name="line-15843"></a><a name="conv3DBackpropInputV2"></a><span class='hs-definition'>conv3DBackpropInputV2</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-15844"></a>                                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-15845"></a>                                                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-15846"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-15847"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-15848"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-15849"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-15850"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-15851"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-15852"></a>                                                      <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-15853"></a>                         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __input_sizes__: An integer vector representing the tensor shape of `input`,</span>
+<a name="line-15854"></a>                                                  <span class='hs-comment'>-- where `input` is a 5-D</span>
+<a name="line-15855"></a>                                                  <span class='hs-comment'>-- `[batch, depth, rows, cols, in_channels]` tensor.</span>
+<a name="line-15856"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __filter__: Shape `[depth, rows, cols, in_channels, out_channels]`.</span>
+<a name="line-15857"></a>                                        <span class='hs-comment'>-- `in_channels` must match between `input` and `filter`.</span>
+<a name="line-15858"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out_backprop__: Backprop signal of shape `[batch, out_depth, out_rows, out_cols,</span>
+<a name="line-15859"></a>                                        <span class='hs-comment'>-- out_channels]`.</span>
+<a name="line-15860"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-15861"></a><span class='hs-definition'>conv3DBackpropInputV2</span> <span class='hs-varid'>input_sizes</span> <span class='hs-varid'>filter</span> <span class='hs-varid'>out_backprop</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-15862"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Conv3DBackpropInputV2"</span>
+<a name="line-15863"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-15864"></a>        <span class='hs-varid'>input_sizes</span> <span class='hs-varid'>filter</span> <span class='hs-varid'>out_backprop</span>
+<a name="line-15865"></a><span class='hs-comment'>{-
+<a name="line-15866"></a>attr {
+<a name="line-15867"></a>  allowed_values {
+<a name="line-15868"></a>    list {
+<a name="line-15869"></a>      type: DT_FLOAT
+<a name="line-15870"></a>      type: DT_DOUBLE
+<a name="line-15871"></a>      type: DT_INT64
+<a name="line-15872"></a>      type: DT_INT32
+<a name="line-15873"></a>      type: DT_UINT8
+<a name="line-15874"></a>      type: DT_UINT16
+<a name="line-15875"></a>      type: DT_INT16
+<a name="line-15876"></a>      type: DT_INT8
+<a name="line-15877"></a>      type: DT_COMPLEX64
+<a name="line-15878"></a>      type: DT_COMPLEX128
+<a name="line-15879"></a>      type: DT_QINT8
+<a name="line-15880"></a>      type: DT_QUINT8
+<a name="line-15881"></a>      type: DT_QINT32
+<a name="line-15882"></a>      type: DT_HALF
+<a name="line-15883"></a>    }
+<a name="line-15884"></a>  }
+<a name="line-15885"></a>  name: "T"
+<a name="line-15886"></a>  type: "type"
+<a name="line-15887"></a>}
+<a name="line-15888"></a>attr {
+<a name="line-15889"></a>  description: "1-D tensor of length 5. The stride of the sliding window for each\ndimension of `input`. Must have `strides[0] = strides[4] = 1`."
+<a name="line-15890"></a>  has_minimum: true
+<a name="line-15891"></a>  minimum: 5
+<a name="line-15892"></a>  name: "strides"
+<a name="line-15893"></a>  type: "list(int)"
+<a name="line-15894"></a>}
+<a name="line-15895"></a>attr {
+<a name="line-15896"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-15897"></a>  description: "The type of padding algorithm to use."
+<a name="line-15898"></a>  name: "padding"
+<a name="line-15899"></a>  type: "string"
+<a name="line-15900"></a>}
+<a name="line-15901"></a>input_arg {
+<a name="line-15902"></a>  description: "An integer vector representing the tensor shape of `input`,\nwhere `input` is a 5-D\n`[batch, depth, rows, cols, in_channels]` tensor."
+<a name="line-15903"></a>  name: "input_sizes"
+<a name="line-15904"></a>  type: DT_INT32
+<a name="line-15905"></a>}
+<a name="line-15906"></a>input_arg {
+<a name="line-15907"></a>  description: "Shape `[depth, rows, cols, in_channels, out_channels]`.\n`in_channels` must match between `input` and `filter`."
+<a name="line-15908"></a>  name: "filter"
+<a name="line-15909"></a>  type_attr: "T"
+<a name="line-15910"></a>}
+<a name="line-15911"></a>input_arg {
+<a name="line-15912"></a>  description: "Backprop signal of shape `[batch, out_depth, out_rows, out_cols,\nout_channels]`."
+<a name="line-15913"></a>  name: "out_backprop"
+<a name="line-15914"></a>  type_attr: "T"
+<a name="line-15915"></a>}
+<a name="line-15916"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-15917"></a>-}</span>
+<a name="line-15918"></a>
+<a name="line-15919"></a><a name="mod"></a><span class='hs-comment'>-- | Returns element-wise remainder of division.</span>
+<a name="line-15920"></a><span class='hs-comment'>--</span>
+<a name="line-15921"></a><span class='hs-comment'>-- *NOTE*: `Mod` supports broadcasting. More about broadcasting</span>
+<a name="line-15922"></a><span class='hs-comment'>-- [here](<a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)">http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)</a></span>
+<a name="line-15923"></a><span class='hs-definition'>mod</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-15924"></a>                                               <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-15925"></a>       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-15926"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-15927"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-15928"></a><span class='hs-definition'>mod</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-15929"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Mod"</span>
+<a name="line-15930"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-15931"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-15932"></a><span class='hs-comment'>{-
+<a name="line-15933"></a>attr {
+<a name="line-15934"></a>  allowed_values {
+<a name="line-15935"></a>    list {
+<a name="line-15936"></a>      type: DT_INT32 type: DT_INT64 type: DT_FLOAT type: DT_DOUBLE
+<a name="line-15937"></a>    }
+<a name="line-15938"></a>  }
+<a name="line-15939"></a>  name: "T"
+<a name="line-15940"></a>  type: "type"
+<a name="line-15941"></a>}
+<a name="line-15942"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-15943"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-15944"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-15945"></a>-}</span>
+<a name="line-15946"></a>
+<a name="line-15947"></a><a name="refMerge"></a><span class='hs-comment'>-- | Forwards the value of an available tensor from `inputs` to `output`.</span>
+<a name="line-15948"></a><span class='hs-comment'>--</span>
+<a name="line-15949"></a><span class='hs-comment'>-- `Merge` waits for at least one of the tensors in `inputs` to become available.</span>
+<a name="line-15950"></a><span class='hs-comment'>-- It is usually combined with `Switch` to implement branching.</span>
+<a name="line-15951"></a><span class='hs-comment'>-- </span>
+<a name="line-15952"></a><span class='hs-comment'>-- `Merge` forwards the first tensor for become available to `output`, and sets</span>
+<a name="line-15953"></a><span class='hs-comment'>-- `value_index` to its index in `inputs`.</span>
+<a name="line-15954"></a><span class='hs-definition'>refMerge</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-15955"></a>            <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __inputs__: The input tensors, exactly one of which will become available.</span>
+<a name="line-15956"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-15957"></a>            <span class='hs-comment'>-- ^ (__output__, __value_index__)</span>
+<a name="line-15958"></a>            <span class='hs-comment'>--</span>
+<a name="line-15959"></a>            <span class='hs-comment'>-- * __output__: Will be set to the available input tensor.</span>
+<a name="line-15960"></a>            <span class='hs-comment'>--</span>
+<a name="line-15961"></a>            <span class='hs-comment'>-- * __value_index__: The index of the chosen input tensor in `inputs`.</span>
+<a name="line-15962"></a><span class='hs-definition'>refMerge</span> <span class='hs-varid'>inputs</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"N"</span><span class='hs-layout'>,</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"inputs"</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>inputs</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-15963"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RefMerge"</span>
+<a name="line-15964"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-15965"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"N"</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>inputs</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-15966"></a>        <span class='hs-varid'>inputs</span>
+<a name="line-15967"></a><span class='hs-comment'>{-
+<a name="line-15968"></a>attr { name: "T" type: "type" }
+<a name="line-15969"></a>attr { has_minimum: true minimum: 1 name: "N" type: "int" }
+<a name="line-15970"></a>input_arg {
+<a name="line-15971"></a>  description: "The input tensors, exactly one of which will become available."
+<a name="line-15972"></a>  is_ref: true
+<a name="line-15973"></a>  name: "inputs"
+<a name="line-15974"></a>  number_attr: "N"
+<a name="line-15975"></a>  type_attr: "T"
+<a name="line-15976"></a>}
+<a name="line-15977"></a>output_arg {
+<a name="line-15978"></a>  description: "Will be set to the available input tensor."
+<a name="line-15979"></a>  is_ref: true
+<a name="line-15980"></a>  name: "output"
+<a name="line-15981"></a>  type_attr: "T"
+<a name="line-15982"></a>}
+<a name="line-15983"></a>output_arg {
+<a name="line-15984"></a>  description: "The index of the chosen input tensor in `inputs`."
+<a name="line-15985"></a>  name: "value_index"
+<a name="line-15986"></a>  type: DT_INT32
+<a name="line-15987"></a>}
+<a name="line-15988"></a>-}</span>
+<a name="line-15989"></a>
+<a name="line-15990"></a><span class='hs-comment'>-- | Computes the gradients of 3-D convolution with respect to the filter.</span>
+<a name="line-15991"></a>
+<a name="line-15992"></a><a name="conv3DBackpropFilterV2"></a><span class='hs-definition'>conv3DBackpropFilterV2</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-15993"></a>                                               <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-15994"></a>                                                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-15995"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-15996"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-15997"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-15998"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-15999"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-16000"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-16001"></a>                                                       <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16002"></a>                          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Shape `[batch, depth, rows, cols, in_channels]`.</span>
+<a name="line-16003"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __filter_sizes__: An integer vector representing the tensor shape of `filter`,</span>
+<a name="line-16004"></a>                                                      <span class='hs-comment'>-- where `filter` is a 5-D</span>
+<a name="line-16005"></a>                                                      <span class='hs-comment'>-- `[filter_depth, filter_height, filter_width, in_channels, out_channels]`</span>
+<a name="line-16006"></a>                                                      <span class='hs-comment'>-- tensor.</span>
+<a name="line-16007"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out_backprop__: Backprop signal of shape `[batch, out_depth, out_rows, out_cols,</span>
+<a name="line-16008"></a>                                         <span class='hs-comment'>-- out_channels]`.</span>
+<a name="line-16009"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-16010"></a><span class='hs-definition'>conv3DBackpropFilterV2</span> <span class='hs-varid'>input</span> <span class='hs-varid'>filter_sizes</span> <span class='hs-varid'>out_backprop</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16011"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Conv3DBackpropFilterV2"</span>
+<a name="line-16012"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16013"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>filter_sizes</span> <span class='hs-varid'>out_backprop</span>
+<a name="line-16014"></a><span class='hs-comment'>{-
+<a name="line-16015"></a>attr {
+<a name="line-16016"></a>  allowed_values {
+<a name="line-16017"></a>    list {
+<a name="line-16018"></a>      type: DT_FLOAT
+<a name="line-16019"></a>      type: DT_DOUBLE
+<a name="line-16020"></a>      type: DT_INT64
+<a name="line-16021"></a>      type: DT_INT32
+<a name="line-16022"></a>      type: DT_UINT8
+<a name="line-16023"></a>      type: DT_UINT16
+<a name="line-16024"></a>      type: DT_INT16
+<a name="line-16025"></a>      type: DT_INT8
+<a name="line-16026"></a>      type: DT_COMPLEX64
+<a name="line-16027"></a>      type: DT_COMPLEX128
+<a name="line-16028"></a>      type: DT_QINT8
+<a name="line-16029"></a>      type: DT_QUINT8
+<a name="line-16030"></a>      type: DT_QINT32
+<a name="line-16031"></a>      type: DT_HALF
+<a name="line-16032"></a>    }
+<a name="line-16033"></a>  }
+<a name="line-16034"></a>  name: "T"
+<a name="line-16035"></a>  type: "type"
+<a name="line-16036"></a>}
+<a name="line-16037"></a>attr {
+<a name="line-16038"></a>  description: "1-D tensor of length 5. The stride of the sliding window for each\ndimension of `input`. Must have `strides[0] = strides[4] = 1`."
+<a name="line-16039"></a>  has_minimum: true
+<a name="line-16040"></a>  minimum: 5
+<a name="line-16041"></a>  name: "strides"
+<a name="line-16042"></a>  type: "list(int)"
+<a name="line-16043"></a>}
+<a name="line-16044"></a>attr {
+<a name="line-16045"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-16046"></a>  description: "The type of padding algorithm to use."
+<a name="line-16047"></a>  name: "padding"
+<a name="line-16048"></a>  type: "string"
+<a name="line-16049"></a>}
+<a name="line-16050"></a>input_arg {
+<a name="line-16051"></a>  description: "Shape `[batch, depth, rows, cols, in_channels]`."
+<a name="line-16052"></a>  name: "input"
+<a name="line-16053"></a>  type_attr: "T"
+<a name="line-16054"></a>}
+<a name="line-16055"></a>input_arg {
+<a name="line-16056"></a>  description: "An integer vector representing the tensor shape of `filter`,\nwhere `filter` is a 5-D\n`[filter_depth, filter_height, filter_width, in_channels, out_channels]`\ntensor."
+<a name="line-16057"></a>  name: "filter_sizes"
+<a name="line-16058"></a>  type: DT_INT32
+<a name="line-16059"></a>}
+<a name="line-16060"></a>input_arg {
+<a name="line-16061"></a>  description: "Backprop signal of shape `[batch, out_depth, out_rows, out_cols,\nout_channels]`."
+<a name="line-16062"></a>  name: "out_backprop"
+<a name="line-16063"></a>  type_attr: "T"
+<a name="line-16064"></a>}
+<a name="line-16065"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-16066"></a>-}</span>
+<a name="line-16067"></a>
+<a name="line-16068"></a><a name="serializeManySparse"></a><span class='hs-comment'>-- | Serialize an `N`-minibatch `SparseTensor` into an `[N, 3]` string `Tensor`.</span>
+<a name="line-16069"></a><span class='hs-comment'>--</span>
+<a name="line-16070"></a><span class='hs-comment'>-- The `SparseTensor` must have rank `R` greater than 1, and the first dimension</span>
+<a name="line-16071"></a><span class='hs-comment'>-- is treated as the minibatch dimension.  Elements of the `SparseTensor`</span>
+<a name="line-16072"></a><span class='hs-comment'>-- must be sorted in increasing order of this first dimension.  The serialized</span>
+<a name="line-16073"></a><span class='hs-comment'>-- `SparseTensor` objects going into each row of `serialized_sparse` will have</span>
+<a name="line-16074"></a><span class='hs-comment'>-- rank `R-1`.</span>
+<a name="line-16075"></a><span class='hs-comment'>-- </span>
+<a name="line-16076"></a><span class='hs-comment'>-- The minibatch size `N` is extracted from `sparse_shape[0]`.</span>
+<a name="line-16077"></a><span class='hs-definition'>serializeManySparse</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16078"></a>                       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __sparse_indices__: 2-D.  The `indices` of the minibatch `SparseTensor`.</span>
+<a name="line-16079"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __sparse_values__: 1-D.  The `values` of the minibatch `SparseTensor`.</span>
+<a name="line-16080"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __sparse_shape__: 1-D.  The `shape` of the minibatch `SparseTensor`.</span>
+<a name="line-16081"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __serialized_sparse__</span>
+<a name="line-16082"></a><span class='hs-definition'>serializeManySparse</span> <span class='hs-varid'>sparse_indices</span> <span class='hs-varid'>sparse_values</span>
+<a name="line-16083"></a>                    <span class='hs-varid'>sparse_shape</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16084"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SerializeManySparse"</span>
+<a name="line-16085"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16086"></a>        <span class='hs-varid'>sparse_indices</span> <span class='hs-varid'>sparse_values</span> <span class='hs-varid'>sparse_shape</span>
+<a name="line-16087"></a><span class='hs-comment'>{-
+<a name="line-16088"></a>attr { name: "T" type: "type" }
+<a name="line-16089"></a>input_arg {
+<a name="line-16090"></a>  description: "2-D.  The `indices` of the minibatch `SparseTensor`."
+<a name="line-16091"></a>  name: "sparse_indices"
+<a name="line-16092"></a>  type: DT_INT64
+<a name="line-16093"></a>}
+<a name="line-16094"></a>input_arg {
+<a name="line-16095"></a>  description: "1-D.  The `values` of the minibatch `SparseTensor`."
+<a name="line-16096"></a>  name: "sparse_values"
+<a name="line-16097"></a>  type_attr: "T"
+<a name="line-16098"></a>}
+<a name="line-16099"></a>input_arg {
+<a name="line-16100"></a>  description: "1-D.  The `shape` of the minibatch `SparseTensor`."
+<a name="line-16101"></a>  name: "sparse_shape"
+<a name="line-16102"></a>  type: DT_INT64
+<a name="line-16103"></a>}
+<a name="line-16104"></a>output_arg { name: "serialized_sparse" type: DT_STRING }
+<a name="line-16105"></a>-}</span>
+<a name="line-16106"></a>
+<a name="line-16107"></a><span class='hs-comment'>-- | Computes gradients of average pooling function.</span>
+<a name="line-16108"></a>
+<a name="line-16109"></a><a name="avgPool3DGrad"></a><span class='hs-definition'>avgPool3DGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-16110"></a>                                   <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-16111"></a>                                           <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-16112"></a>                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-16113"></a>                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-16114"></a>                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-16115"></a>                                           <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16116"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __orig_input_shape__: The original input dimensions.</span>
+<a name="line-16117"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: Output backprop of shape `[batch, depth, rows, cols, channels]`.</span>
+<a name="line-16118"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The backprop for input.</span>
+<a name="line-16119"></a><span class='hs-definition'>avgPool3DGrad</span> <span class='hs-varid'>orig_input_shape</span> <span class='hs-varid'>grad</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16120"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"AvgPool3DGrad"</span>
+<a name="line-16121"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16122"></a>        <span class='hs-varid'>orig_input_shape</span> <span class='hs-varid'>grad</span>
+<a name="line-16123"></a><span class='hs-comment'>{-
+<a name="line-16124"></a>attr {
+<a name="line-16125"></a>  description: "1-D tensor of length 5. The size of the window for each dimension of\nthe input tensor. Must have `ksize[0] = ksize[4] = 1`."
+<a name="line-16126"></a>  has_minimum: true
+<a name="line-16127"></a>  minimum: 5
+<a name="line-16128"></a>  name: "ksize"
+<a name="line-16129"></a>  type: "list(int)"
+<a name="line-16130"></a>}
+<a name="line-16131"></a>attr {
+<a name="line-16132"></a>  description: "1-D tensor of length 5. The stride of the sliding window for each\ndimension of `input`. Must have `strides[0] = strides[4] = 1`."
+<a name="line-16133"></a>  has_minimum: true
+<a name="line-16134"></a>  minimum: 5
+<a name="line-16135"></a>  name: "strides"
+<a name="line-16136"></a>  type: "list(int)"
+<a name="line-16137"></a>}
+<a name="line-16138"></a>attr {
+<a name="line-16139"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-16140"></a>  description: "The type of padding algorithm to use."
+<a name="line-16141"></a>  name: "padding"
+<a name="line-16142"></a>  type: "string"
+<a name="line-16143"></a>}
+<a name="line-16144"></a>attr {
+<a name="line-16145"></a>  allowed_values {
+<a name="line-16146"></a>    list {
+<a name="line-16147"></a>      type: DT_FLOAT
+<a name="line-16148"></a>      type: DT_DOUBLE
+<a name="line-16149"></a>      type: DT_INT64
+<a name="line-16150"></a>      type: DT_INT32
+<a name="line-16151"></a>      type: DT_UINT8
+<a name="line-16152"></a>      type: DT_UINT16
+<a name="line-16153"></a>      type: DT_INT16
+<a name="line-16154"></a>      type: DT_INT8
+<a name="line-16155"></a>      type: DT_COMPLEX64
+<a name="line-16156"></a>      type: DT_COMPLEX128
+<a name="line-16157"></a>      type: DT_QINT8
+<a name="line-16158"></a>      type: DT_QUINT8
+<a name="line-16159"></a>      type: DT_QINT32
+<a name="line-16160"></a>      type: DT_HALF
+<a name="line-16161"></a>    }
+<a name="line-16162"></a>  }
+<a name="line-16163"></a>  name: "T"
+<a name="line-16164"></a>  type: "type"
+<a name="line-16165"></a>}
+<a name="line-16166"></a>input_arg {
+<a name="line-16167"></a>  description: "The original input dimensions."
+<a name="line-16168"></a>  name: "orig_input_shape"
+<a name="line-16169"></a>  type: DT_INT32
+<a name="line-16170"></a>}
+<a name="line-16171"></a>input_arg {
+<a name="line-16172"></a>  description: "Output backprop of shape `[batch, depth, rows, cols, channels]`."
+<a name="line-16173"></a>  name: "grad"
+<a name="line-16174"></a>  type_attr: "T"
+<a name="line-16175"></a>}
+<a name="line-16176"></a>output_arg {
+<a name="line-16177"></a>  description: "The backprop for input."
+<a name="line-16178"></a>  name: "output"
+<a name="line-16179"></a>  type_attr: "T"
+<a name="line-16180"></a>}
+<a name="line-16181"></a>-}</span>
+<a name="line-16182"></a>
+<a name="line-16183"></a><span class='hs-comment'>-- | Computes gradients of max pooling function.</span>
+<a name="line-16184"></a>
+<a name="line-16185"></a><a name="maxPool3DGrad"></a><span class='hs-definition'>maxPool3DGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-16186"></a>                                      <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-16187"></a>                                              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-16188"></a>                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-16189"></a>                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-16190"></a>                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-16191"></a>                                              <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16192"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __orig_input__: The original input tensor.</span>
+<a name="line-16193"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __orig_output__: The original output tensor.</span>
+<a name="line-16194"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: Output backprop of shape `[batch, depth, rows, cols, channels]`.</span>
+<a name="line-16195"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-16196"></a><span class='hs-definition'>maxPool3DGrad</span> <span class='hs-varid'>orig_input</span> <span class='hs-varid'>orig_output</span> <span class='hs-varid'>grad</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16197"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MaxPool3DGrad"</span>
+<a name="line-16198"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16199"></a>        <span class='hs-varid'>orig_input</span> <span class='hs-varid'>orig_output</span> <span class='hs-varid'>grad</span>
+<a name="line-16200"></a><span class='hs-comment'>{-
+<a name="line-16201"></a>attr {
+<a name="line-16202"></a>  description: "1-D tensor of length 5. The size of the window for each dimension of\nthe input tensor. Must have `ksize[0] = ksize[4] = 1`."
+<a name="line-16203"></a>  has_minimum: true
+<a name="line-16204"></a>  minimum: 5
+<a name="line-16205"></a>  name: "ksize"
+<a name="line-16206"></a>  type: "list(int)"
+<a name="line-16207"></a>}
+<a name="line-16208"></a>attr {
+<a name="line-16209"></a>  description: "1-D tensor of length 5. The stride of the sliding window for each\ndimension of `input`. Must have `strides[0] = strides[4] = 1`."
+<a name="line-16210"></a>  has_minimum: true
+<a name="line-16211"></a>  minimum: 5
+<a name="line-16212"></a>  name: "strides"
+<a name="line-16213"></a>  type: "list(int)"
+<a name="line-16214"></a>}
+<a name="line-16215"></a>attr {
+<a name="line-16216"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-16217"></a>  description: "The type of padding algorithm to use."
+<a name="line-16218"></a>  name: "padding"
+<a name="line-16219"></a>  type: "string"
+<a name="line-16220"></a>}
+<a name="line-16221"></a>attr {
+<a name="line-16222"></a>  allowed_values {
+<a name="line-16223"></a>    list {
+<a name="line-16224"></a>      type: DT_FLOAT
+<a name="line-16225"></a>      type: DT_DOUBLE
+<a name="line-16226"></a>      type: DT_INT64
+<a name="line-16227"></a>      type: DT_INT32
+<a name="line-16228"></a>      type: DT_UINT8
+<a name="line-16229"></a>      type: DT_UINT16
+<a name="line-16230"></a>      type: DT_INT16
+<a name="line-16231"></a>      type: DT_INT8
+<a name="line-16232"></a>      type: DT_COMPLEX64
+<a name="line-16233"></a>      type: DT_COMPLEX128
+<a name="line-16234"></a>      type: DT_QINT8
+<a name="line-16235"></a>      type: DT_QUINT8
+<a name="line-16236"></a>      type: DT_QINT32
+<a name="line-16237"></a>      type: DT_HALF
+<a name="line-16238"></a>    }
+<a name="line-16239"></a>  }
+<a name="line-16240"></a>  name: "T"
+<a name="line-16241"></a>  type: "type"
+<a name="line-16242"></a>}
+<a name="line-16243"></a>input_arg {
+<a name="line-16244"></a>  description: "The original input tensor."
+<a name="line-16245"></a>  name: "orig_input"
+<a name="line-16246"></a>  type: DT_FLOAT
+<a name="line-16247"></a>}
+<a name="line-16248"></a>input_arg {
+<a name="line-16249"></a>  description: "The original output tensor."
+<a name="line-16250"></a>  name: "orig_output"
+<a name="line-16251"></a>  type: DT_FLOAT
+<a name="line-16252"></a>}
+<a name="line-16253"></a>input_arg {
+<a name="line-16254"></a>  description: "Output backprop of shape `[batch, depth, rows, cols, channels]`."
+<a name="line-16255"></a>  name: "grad"
+<a name="line-16256"></a>  type_attr: "T"
+<a name="line-16257"></a>}
+<a name="line-16258"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-16259"></a>-}</span>
+<a name="line-16260"></a>
+<a name="line-16261"></a><a name="sparseReduceSum"></a><span class='hs-comment'>-- | Computes the sum of elements across dimensions of a SparseTensor.</span>
+<a name="line-16262"></a><span class='hs-comment'>--</span>
+<a name="line-16263"></a><span class='hs-comment'>-- This Op takes a SparseTensor and is the sparse counterpart to</span>
+<a name="line-16264"></a><span class='hs-comment'>-- `tf.reduce_sum()`.  In particular, this Op also returns a dense `Tensor`</span>
+<a name="line-16265"></a><span class='hs-comment'>-- instead of a sparse one.</span>
+<a name="line-16266"></a><span class='hs-comment'>-- </span>
+<a name="line-16267"></a><span class='hs-comment'>-- Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless</span>
+<a name="line-16268"></a><span class='hs-comment'>-- `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in</span>
+<a name="line-16269"></a><span class='hs-comment'>-- `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained</span>
+<a name="line-16270"></a><span class='hs-comment'>-- with length 1.</span>
+<a name="line-16271"></a><span class='hs-comment'>-- </span>
+<a name="line-16272"></a><span class='hs-comment'>-- If `reduction_axes` has no entries, all dimensions are reduced, and a tensor</span>
+<a name="line-16273"></a><span class='hs-comment'>-- with a single element is returned.  Additionally, the axes can be negative,</span>
+<a name="line-16274"></a><span class='hs-comment'>-- which are interpreted according to the indexing rules in Python.</span>
+<a name="line-16275"></a><span class='hs-definition'>sparseReduceSum</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-16276"></a>                                           <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-16277"></a>                                                   <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-16278"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-16279"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-16280"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-16281"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-16282"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-16283"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-16284"></a>                                                   <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16285"></a>                   <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __input_indices__: 2-D.  `N x R` matrix with the indices of non-empty values in a</span>
+<a name="line-16286"></a>                                            <span class='hs-comment'>-- SparseTensor, possibly not in canonical ordering.</span>
+<a name="line-16287"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input_values__: 1-D.  `N` non-empty values corresponding to `input_indices`.</span>
+<a name="line-16288"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __input_shape__: 1-D.  Shape of the input SparseTensor.</span>
+<a name="line-16289"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __reduction_axes__: 1-D.  Length-`K` vector containing the reduction axes.</span>
+<a name="line-16290"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: `R-K`-D.  The reduced Tensor.</span>
+<a name="line-16291"></a><span class='hs-definition'>sparseReduceSum</span> <span class='hs-varid'>input_indices</span> <span class='hs-varid'>input_values</span> <span class='hs-varid'>input_shape</span>
+<a name="line-16292"></a>                <span class='hs-varid'>reduction_axes</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16293"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseReduceSum"</span>
+<a name="line-16294"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16295"></a>        <span class='hs-varid'>input_indices</span> <span class='hs-varid'>input_values</span> <span class='hs-varid'>input_shape</span> <span class='hs-varid'>reduction_axes</span>
+<a name="line-16296"></a><span class='hs-comment'>{-
+<a name="line-16297"></a>attr {
+<a name="line-16298"></a>  default_value { b: false }
+<a name="line-16299"></a>  description: "If true, retain reduced dimensions with length 1."
+<a name="line-16300"></a>  name: "keep_dims"
+<a name="line-16301"></a>  type: "bool"
+<a name="line-16302"></a>}
+<a name="line-16303"></a>attr {
+<a name="line-16304"></a>  allowed_values {
+<a name="line-16305"></a>    list {
+<a name="line-16306"></a>      type: DT_FLOAT
+<a name="line-16307"></a>      type: DT_DOUBLE
+<a name="line-16308"></a>      type: DT_INT64
+<a name="line-16309"></a>      type: DT_INT32
+<a name="line-16310"></a>      type: DT_UINT8
+<a name="line-16311"></a>      type: DT_UINT16
+<a name="line-16312"></a>      type: DT_INT16
+<a name="line-16313"></a>      type: DT_INT8
+<a name="line-16314"></a>      type: DT_COMPLEX64
+<a name="line-16315"></a>      type: DT_COMPLEX128
+<a name="line-16316"></a>      type: DT_QINT8
+<a name="line-16317"></a>      type: DT_QUINT8
+<a name="line-16318"></a>      type: DT_QINT32
+<a name="line-16319"></a>      type: DT_HALF
+<a name="line-16320"></a>    }
+<a name="line-16321"></a>  }
+<a name="line-16322"></a>  name: "T"
+<a name="line-16323"></a>  type: "type"
+<a name="line-16324"></a>}
+<a name="line-16325"></a>input_arg {
+<a name="line-16326"></a>  description: "2-D.  `N x R` matrix with the indices of non-empty values in a\nSparseTensor, possibly not in canonical ordering."
+<a name="line-16327"></a>  name: "input_indices"
+<a name="line-16328"></a>  type: DT_INT64
+<a name="line-16329"></a>}
+<a name="line-16330"></a>input_arg {
+<a name="line-16331"></a>  description: "1-D.  `N` non-empty values corresponding to `input_indices`."
+<a name="line-16332"></a>  name: "input_values"
+<a name="line-16333"></a>  type_attr: "T"
+<a name="line-16334"></a>}
+<a name="line-16335"></a>input_arg {
+<a name="line-16336"></a>  description: "1-D.  Shape of the input SparseTensor."
+<a name="line-16337"></a>  name: "input_shape"
+<a name="line-16338"></a>  type: DT_INT64
+<a name="line-16339"></a>}
+<a name="line-16340"></a>input_arg {
+<a name="line-16341"></a>  description: "1-D.  Length-`K` vector containing the reduction axes."
+<a name="line-16342"></a>  name: "reduction_axes"
+<a name="line-16343"></a>  type: DT_INT32
+<a name="line-16344"></a>}
+<a name="line-16345"></a>output_arg {
+<a name="line-16346"></a>  description: "`R-K`-D.  The reduced Tensor."
+<a name="line-16347"></a>  name: "output"
+<a name="line-16348"></a>  type_attr: "T"
+<a name="line-16349"></a>}
+<a name="line-16350"></a>-}</span>
+<a name="line-16351"></a>
+<a name="line-16352"></a><span class='hs-comment'>-- | Computes rectified linear: `max(features, 0)`.</span>
+<a name="line-16353"></a>
+<a name="line-16354"></a><a name="relu"></a><span class='hs-definition'>relu</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-16355"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-16356"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-16357"></a>                                             <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16358"></a>        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __features__</span>
+<a name="line-16359"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __activations__</span>
+<a name="line-16360"></a><span class='hs-definition'>relu</span> <span class='hs-varid'>features</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16361"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Relu"</span>
+<a name="line-16362"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16363"></a>        <span class='hs-varid'>features</span>
+<a name="line-16364"></a><span class='hs-comment'>{-
+<a name="line-16365"></a>attr {
+<a name="line-16366"></a>  allowed_values {
+<a name="line-16367"></a>    list {
+<a name="line-16368"></a>      type: DT_FLOAT
+<a name="line-16369"></a>      type: DT_DOUBLE
+<a name="line-16370"></a>      type: DT_INT32
+<a name="line-16371"></a>      type: DT_INT64
+<a name="line-16372"></a>      type: DT_UINT8
+<a name="line-16373"></a>      type: DT_INT16
+<a name="line-16374"></a>      type: DT_INT8
+<a name="line-16375"></a>      type: DT_UINT16
+<a name="line-16376"></a>      type: DT_HALF
+<a name="line-16377"></a>    }
+<a name="line-16378"></a>  }
+<a name="line-16379"></a>  name: "T"
+<a name="line-16380"></a>  type: "type"
+<a name="line-16381"></a>}
+<a name="line-16382"></a>input_arg { name: "features" type_attr: "T" }
+<a name="line-16383"></a>output_arg { name: "activations" type_attr: "T" }
+<a name="line-16384"></a>-}</span>
+<a name="line-16385"></a>
+<a name="line-16386"></a><a name="l2Loss"></a><span class='hs-comment'>-- | L2 Loss.</span>
+<a name="line-16387"></a><span class='hs-comment'>--</span>
+<a name="line-16388"></a><span class='hs-comment'>-- Computes half the L2 norm of a tensor without the `sqrt`:</span>
+<a name="line-16389"></a><span class='hs-comment'>-- </span>
+<a name="line-16390"></a><span class='hs-comment'>--     output = sum(t ** 2) / 2</span>
+<a name="line-16391"></a><span class='hs-definition'>l2Loss</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-16392"></a>                                               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-16393"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-16394"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-16395"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-16396"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-16397"></a>                                               <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16398"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __t__: Typically 2-D, but may have any dimensions.</span>
+<a name="line-16399"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 0-D.</span>
+<a name="line-16400"></a><span class='hs-definition'>l2Loss</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16401"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"L2Loss"</span>
+<a name="line-16402"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16403"></a>        <span class='hs-varid'>t</span>
+<a name="line-16404"></a><span class='hs-comment'>{-
+<a name="line-16405"></a>attr {
+<a name="line-16406"></a>  allowed_values {
+<a name="line-16407"></a>    list {
+<a name="line-16408"></a>      type: DT_FLOAT
+<a name="line-16409"></a>      type: DT_DOUBLE
+<a name="line-16410"></a>      type: DT_INT64
+<a name="line-16411"></a>      type: DT_INT32
+<a name="line-16412"></a>      type: DT_UINT8
+<a name="line-16413"></a>      type: DT_UINT16
+<a name="line-16414"></a>      type: DT_INT16
+<a name="line-16415"></a>      type: DT_INT8
+<a name="line-16416"></a>      type: DT_COMPLEX64
+<a name="line-16417"></a>      type: DT_COMPLEX128
+<a name="line-16418"></a>      type: DT_QINT8
+<a name="line-16419"></a>      type: DT_QUINT8
+<a name="line-16420"></a>      type: DT_QINT32
+<a name="line-16421"></a>      type: DT_HALF
+<a name="line-16422"></a>    }
+<a name="line-16423"></a>  }
+<a name="line-16424"></a>  name: "T"
+<a name="line-16425"></a>  type: "type"
+<a name="line-16426"></a>}
+<a name="line-16427"></a>input_arg {
+<a name="line-16428"></a>  description: "Typically 2-D, but may have any dimensions."
+<a name="line-16429"></a>  name: "t"
+<a name="line-16430"></a>  type_attr: "T"
+<a name="line-16431"></a>}
+<a name="line-16432"></a>output_arg { description: "0-D." name: "output" type_attr: "T" }
+<a name="line-16433"></a>-}</span>
+<a name="line-16434"></a>
+<a name="line-16435"></a><a name="readerRestoreState"></a><span class='hs-comment'>-- | Restore a reader to a previously saved state.</span>
+<a name="line-16436"></a><span class='hs-comment'>--</span>
+<a name="line-16437"></a><span class='hs-comment'>-- Not all Readers support being restored, so this can produce an</span>
+<a name="line-16438"></a><span class='hs-comment'>-- Unimplemented error.</span>
+<a name="line-16439"></a><span class='hs-definition'>readerRestoreState</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __reader_handle__: Handle to a Reader.</span>
+<a name="line-16440"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __state__: Result of a ReaderSerializeState of a Reader with type</span>
+<a name="line-16441"></a>                                                              <span class='hs-comment'>-- matching reader_handle.</span>
+<a name="line-16442"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-16443"></a><span class='hs-definition'>readerRestoreState</span> <span class='hs-varid'>reader_handle</span> <span class='hs-varid'>state</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16444"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ReaderRestoreState"</span><span class='hs-layout'>)</span>
+<a name="line-16445"></a>        <span class='hs-varid'>reader_handle</span> <span class='hs-varid'>state</span>
+<a name="line-16446"></a><span class='hs-comment'>{-
+<a name="line-16447"></a>input_arg {
+<a name="line-16448"></a>  description: "Handle to a Reader."
+<a name="line-16449"></a>  is_ref: true
+<a name="line-16450"></a>  name: "reader_handle"
+<a name="line-16451"></a>  type: DT_STRING
+<a name="line-16452"></a>}
+<a name="line-16453"></a>input_arg {
+<a name="line-16454"></a>  description: "Result of a ReaderSerializeState of a Reader with type\nmatching reader_handle."
+<a name="line-16455"></a>  name: "state"
+<a name="line-16456"></a>  type: DT_STRING
+<a name="line-16457"></a>}
+<a name="line-16458"></a>-}</span>
+<a name="line-16459"></a>
+<a name="line-16460"></a><a name="shape"></a><span class='hs-comment'>-- | Returns the shape of a tensor.</span>
+<a name="line-16461"></a><span class='hs-comment'>--</span>
+<a name="line-16462"></a><span class='hs-comment'>-- This operation returns a 1-D integer tensor representing the shape of `input`.</span>
+<a name="line-16463"></a><span class='hs-comment'>-- </span>
+<a name="line-16464"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-16465"></a><span class='hs-comment'>-- </span>
+<a name="line-16466"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-16467"></a><span class='hs-comment'>-- # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]</span>
+<a name="line-16468"></a><span class='hs-comment'>-- shape(t) ==&gt; [2, 2, 3]</span>
+<a name="line-16469"></a><span class='hs-comment'>-- ```</span>
+<a name="line-16470"></a><span class='hs-definition'>shape</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varid'>out_type</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>,</span>
+<a name="line-16471"></a>                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-16472"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16473"></a>         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-16474"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>out_type</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-16475"></a><span class='hs-definition'>shape</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16476"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Shape"</span>
+<a name="line-16477"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-16478"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"out_type"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16479"></a>        <span class='hs-varid'>input</span>
+<a name="line-16480"></a><span class='hs-comment'>{-
+<a name="line-16481"></a>attr { name: "T" type: "type" }
+<a name="line-16482"></a>attr {
+<a name="line-16483"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-16484"></a>  default_value { type: DT_INT32 }
+<a name="line-16485"></a>  name: "out_type"
+<a name="line-16486"></a>  type: "type"
+<a name="line-16487"></a>}
+<a name="line-16488"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-16489"></a>output_arg { name: "output" type_attr: "out_type" }
+<a name="line-16490"></a>-}</span>
+<a name="line-16491"></a>
+<a name="line-16492"></a><a name="softmaxCrossEntropyWithLogits"></a><span class='hs-comment'>-- | Computes softmax cross entropy cost and gradients to backpropagate.</span>
+<a name="line-16493"></a><span class='hs-comment'>--</span>
+<a name="line-16494"></a><span class='hs-comment'>-- Inputs are the logits, not probabilities.</span>
+<a name="line-16495"></a><span class='hs-definition'>softmaxCrossEntropyWithLogits</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-16496"></a>                                                   <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-16497"></a>                                                           <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16498"></a>                                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __features__: batch_size x num_classes matrix</span>
+<a name="line-16499"></a>                                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __labels__: batch_size x num_classes matrix</span>
+<a name="line-16500"></a>                                                <span class='hs-comment'>-- The caller must ensure that each batch of labels represents a valid</span>
+<a name="line-16501"></a>                                                <span class='hs-comment'>-- probability distribution.</span>
+<a name="line-16502"></a>                                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-16503"></a>                                 <span class='hs-comment'>-- ^ (__loss__, __backprop__)</span>
+<a name="line-16504"></a>                                 <span class='hs-comment'>--</span>
+<a name="line-16505"></a>                                 <span class='hs-comment'>-- * __loss__: Per example loss (batch_size vector).</span>
+<a name="line-16506"></a>                                 <span class='hs-comment'>--</span>
+<a name="line-16507"></a>                                 <span class='hs-comment'>-- * __backprop__: backpropagated gradients (batch_size x num_classes matrix).</span>
+<a name="line-16508"></a><span class='hs-definition'>softmaxCrossEntropyWithLogits</span> <span class='hs-varid'>features</span> <span class='hs-varid'>labels</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16509"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SoftmaxCrossEntropyWithLogits"</span>
+<a name="line-16510"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16511"></a>        <span class='hs-varid'>features</span> <span class='hs-varid'>labels</span>
+<a name="line-16512"></a><span class='hs-comment'>{-
+<a name="line-16513"></a>attr {
+<a name="line-16514"></a>  allowed_values {
+<a name="line-16515"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-16516"></a>  }
+<a name="line-16517"></a>  name: "T"
+<a name="line-16518"></a>  type: "type"
+<a name="line-16519"></a>}
+<a name="line-16520"></a>input_arg {
+<a name="line-16521"></a>  description: "batch_size x num_classes matrix"
+<a name="line-16522"></a>  name: "features"
+<a name="line-16523"></a>  type_attr: "T"
+<a name="line-16524"></a>}
+<a name="line-16525"></a>input_arg {
+<a name="line-16526"></a>  description: "batch_size x num_classes matrix\nThe caller must ensure that each batch of labels represents a valid\nprobability distribution."
+<a name="line-16527"></a>  name: "labels"
+<a name="line-16528"></a>  type_attr: "T"
+<a name="line-16529"></a>}
+<a name="line-16530"></a>output_arg {
+<a name="line-16531"></a>  description: "Per example loss (batch_size vector)."
+<a name="line-16532"></a>  name: "loss"
+<a name="line-16533"></a>  type_attr: "T"
+<a name="line-16534"></a>}
+<a name="line-16535"></a>output_arg {
+<a name="line-16536"></a>  description: "backpropagated gradients (batch_size x num_classes matrix)."
+<a name="line-16537"></a>  name: "backprop"
+<a name="line-16538"></a>  type_attr: "T"
+<a name="line-16539"></a>}
+<a name="line-16540"></a>-}</span>
+<a name="line-16541"></a>
+<a name="line-16542"></a><span class='hs-comment'>-- | Performs max pooling on the input.</span>
+<a name="line-16543"></a>
+<a name="line-16544"></a><a name="maxPool"></a><span class='hs-definition'>maxPool</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16545"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: 4-D input to pool over.</span>
+<a name="line-16546"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The max pooled output tensor.</span>
+<a name="line-16547"></a><span class='hs-definition'>maxPool</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16548"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MaxPool"</span>
+<a name="line-16549"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16550"></a>        <span class='hs-varid'>input</span>
+<a name="line-16551"></a><span class='hs-comment'>{-
+<a name="line-16552"></a>attr {
+<a name="line-16553"></a>  allowed_values { list { type: DT_FLOAT type: DT_HALF } }
+<a name="line-16554"></a>  default_value { type: DT_FLOAT }
+<a name="line-16555"></a>  name: "T"
+<a name="line-16556"></a>  type: "type"
+<a name="line-16557"></a>}
+<a name="line-16558"></a>attr {
+<a name="line-16559"></a>  description: "The size of the window for each dimension of the input tensor."
+<a name="line-16560"></a>  has_minimum: true
+<a name="line-16561"></a>  minimum: 4
+<a name="line-16562"></a>  name: "ksize"
+<a name="line-16563"></a>  type: "list(int)"
+<a name="line-16564"></a>}
+<a name="line-16565"></a>attr {
+<a name="line-16566"></a>  description: "The stride of the sliding window for each dimension of the\ninput tensor."
+<a name="line-16567"></a>  has_minimum: true
+<a name="line-16568"></a>  minimum: 4
+<a name="line-16569"></a>  name: "strides"
+<a name="line-16570"></a>  type: "list(int)"
+<a name="line-16571"></a>}
+<a name="line-16572"></a>attr {
+<a name="line-16573"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-16574"></a>  description: "The type of padding algorithm to use."
+<a name="line-16575"></a>  name: "padding"
+<a name="line-16576"></a>  type: "string"
+<a name="line-16577"></a>}
+<a name="line-16578"></a>attr {
+<a name="line-16579"></a>  allowed_values { list { s: "NHWC" s: "NCHW" } }
+<a name="line-16580"></a>  default_value { s: "NHWC" }
+<a name="line-16581"></a>  description: "Specify the data format of the input and output data. With the\ndefault format \"NHWC\", the data is stored in the order of:\n    [batch, in_height, in_width, in_channels].\nAlternatively, the format could be \"NCHW\", the data storage order of:\n    [batch, in_channels, in_height, in_width]."
+<a name="line-16582"></a>  name: "data_format"
+<a name="line-16583"></a>  type: "string"
+<a name="line-16584"></a>}
+<a name="line-16585"></a>input_arg {
+<a name="line-16586"></a>  description: "4-D input to pool over." name: "input" type_attr: "T"
+<a name="line-16587"></a>}
+<a name="line-16588"></a>output_arg {
+<a name="line-16589"></a>  description: "The max pooled output tensor."
+<a name="line-16590"></a>  name: "output"
+<a name="line-16591"></a>  type_attr: "T"
+<a name="line-16592"></a>}
+<a name="line-16593"></a>-}</span>
+<a name="line-16594"></a>
+<a name="line-16595"></a><span class='hs-comment'>-- | Computes the gradient of morphological 2-D dilation with respect to the input.</span>
+<a name="line-16596"></a>
+<a name="line-16597"></a><a name="dilation2DBackpropInput"></a><span class='hs-definition'>dilation2DBackpropInput</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-16598"></a>                                                <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-16599"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-16600"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-16601"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-16602"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-16603"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-16604"></a>                                                        <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16605"></a>                           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: 4-D with shape `[batch, in_height, in_width, depth]`.</span>
+<a name="line-16606"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __filter__: 3-D with shape `[filter_height, filter_width, depth]`.</span>
+<a name="line-16607"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out_backprop__: 4-D with shape `[batch, out_height, out_width, depth]`.</span>
+<a name="line-16608"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __in_backprop__: 4-D with shape `[batch, in_height, in_width, depth]`.</span>
+<a name="line-16609"></a><span class='hs-definition'>dilation2DBackpropInput</span> <span class='hs-varid'>input</span> <span class='hs-varid'>filter</span> <span class='hs-varid'>out_backprop</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16610"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Dilation2DBackpropInput"</span>
+<a name="line-16611"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16612"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>filter</span> <span class='hs-varid'>out_backprop</span>
+<a name="line-16613"></a><span class='hs-comment'>{-
+<a name="line-16614"></a>attr {
+<a name="line-16615"></a>  allowed_values {
+<a name="line-16616"></a>    list {
+<a name="line-16617"></a>      type: DT_FLOAT
+<a name="line-16618"></a>      type: DT_DOUBLE
+<a name="line-16619"></a>      type: DT_INT32
+<a name="line-16620"></a>      type: DT_INT64
+<a name="line-16621"></a>      type: DT_UINT8
+<a name="line-16622"></a>      type: DT_INT16
+<a name="line-16623"></a>      type: DT_INT8
+<a name="line-16624"></a>      type: DT_UINT16
+<a name="line-16625"></a>      type: DT_HALF
+<a name="line-16626"></a>    }
+<a name="line-16627"></a>  }
+<a name="line-16628"></a>  name: "T"
+<a name="line-16629"></a>  type: "type"
+<a name="line-16630"></a>}
+<a name="line-16631"></a>attr {
+<a name="line-16632"></a>  description: "1-D of length 4. The stride of the sliding window for each dimension of\nthe input tensor. Must be: `[1, stride_height, stride_width, 1]`."
+<a name="line-16633"></a>  has_minimum: true
+<a name="line-16634"></a>  minimum: 4
+<a name="line-16635"></a>  name: "strides"
+<a name="line-16636"></a>  type: "list(int)"
+<a name="line-16637"></a>}
+<a name="line-16638"></a>attr {
+<a name="line-16639"></a>  description: "1-D of length 4. The input stride for atrous morphological dilation.\nMust be: `[1, rate_height, rate_width, 1]`."
+<a name="line-16640"></a>  has_minimum: true
+<a name="line-16641"></a>  minimum: 4
+<a name="line-16642"></a>  name: "rates"
+<a name="line-16643"></a>  type: "list(int)"
+<a name="line-16644"></a>}
+<a name="line-16645"></a>attr {
+<a name="line-16646"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-16647"></a>  description: "The type of padding algorithm to use."
+<a name="line-16648"></a>  name: "padding"
+<a name="line-16649"></a>  type: "string"
+<a name="line-16650"></a>}
+<a name="line-16651"></a>input_arg {
+<a name="line-16652"></a>  description: "4-D with shape `[batch, in_height, in_width, depth]`."
+<a name="line-16653"></a>  name: "input"
+<a name="line-16654"></a>  type_attr: "T"
+<a name="line-16655"></a>}
+<a name="line-16656"></a>input_arg {
+<a name="line-16657"></a>  description: "3-D with shape `[filter_height, filter_width, depth]`."
+<a name="line-16658"></a>  name: "filter"
+<a name="line-16659"></a>  type_attr: "T"
+<a name="line-16660"></a>}
+<a name="line-16661"></a>input_arg {
+<a name="line-16662"></a>  description: "4-D with shape `[batch, out_height, out_width, depth]`."
+<a name="line-16663"></a>  name: "out_backprop"
+<a name="line-16664"></a>  type_attr: "T"
+<a name="line-16665"></a>}
+<a name="line-16666"></a>output_arg {
+<a name="line-16667"></a>  description: "4-D with shape `[batch, in_height, in_width, depth]`."
+<a name="line-16668"></a>  name: "in_backprop"
+<a name="line-16669"></a>  type_attr: "T"
+<a name="line-16670"></a>}
+<a name="line-16671"></a>-}</span>
+<a name="line-16672"></a>
+<a name="line-16673"></a><a name="equal"></a><span class='hs-comment'>-- | Returns the truth value of (x == y) element-wise.</span>
+<a name="line-16674"></a><span class='hs-comment'>--</span>
+<a name="line-16675"></a><span class='hs-comment'>-- *NOTE*: `Equal` supports broadcasting. More about broadcasting</span>
+<a name="line-16676"></a><span class='hs-comment'>-- [here](<a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)">http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)</a></span>
+<a name="line-16677"></a><span class='hs-definition'>equal</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-16678"></a>                                                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-16679"></a>                                                 <span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-16680"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>,</span>
+<a name="line-16681"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-16682"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-16683"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-16684"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-16685"></a>                                                 <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16686"></a>         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-16687"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-16688"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-16689"></a><span class='hs-definition'>equal</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16690"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Equal"</span>
+<a name="line-16691"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16692"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-16693"></a><span class='hs-comment'>{-
+<a name="line-16694"></a>attr {
+<a name="line-16695"></a>  allowed_values {
+<a name="line-16696"></a>    list {
+<a name="line-16697"></a>      type: DT_HALF
+<a name="line-16698"></a>      type: DT_FLOAT
+<a name="line-16699"></a>      type: DT_DOUBLE
+<a name="line-16700"></a>      type: DT_UINT8
+<a name="line-16701"></a>      type: DT_INT8
+<a name="line-16702"></a>      type: DT_INT16
+<a name="line-16703"></a>      type: DT_INT32
+<a name="line-16704"></a>      type: DT_INT64
+<a name="line-16705"></a>      type: DT_COMPLEX64
+<a name="line-16706"></a>      type: DT_QUINT8
+<a name="line-16707"></a>      type: DT_QINT8
+<a name="line-16708"></a>      type: DT_QINT32
+<a name="line-16709"></a>      type: DT_STRING
+<a name="line-16710"></a>      type: DT_BOOL
+<a name="line-16711"></a>      type: DT_COMPLEX128
+<a name="line-16712"></a>    }
+<a name="line-16713"></a>  }
+<a name="line-16714"></a>  name: "T"
+<a name="line-16715"></a>  type: "type"
+<a name="line-16716"></a>}
+<a name="line-16717"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-16718"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-16719"></a>output_arg { name: "z" type: DT_BOOL }
+<a name="line-16720"></a>-}</span>
+<a name="line-16721"></a>
+<a name="line-16722"></a><span class='hs-comment'>-- | Computes the gradient of morphological 2-D dilation with respect to the filter.</span>
+<a name="line-16723"></a>
+<a name="line-16724"></a><a name="dilation2DBackpropFilter"></a><span class='hs-definition'>dilation2DBackpropFilter</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-16725"></a>                                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-16726"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-16727"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-16728"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-16729"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-16730"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-16731"></a>                                                         <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16732"></a>                            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: 4-D with shape `[batch, in_height, in_width, depth]`.</span>
+<a name="line-16733"></a>                            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __filter__: 3-D with shape `[filter_height, filter_width, depth]`.</span>
+<a name="line-16734"></a>                            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out_backprop__: 4-D with shape `[batch, out_height, out_width, depth]`.</span>
+<a name="line-16735"></a>                            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __filter_backprop__: 3-D with shape `[filter_height, filter_width, depth]`.</span>
+<a name="line-16736"></a><span class='hs-definition'>dilation2DBackpropFilter</span> <span class='hs-varid'>input</span> <span class='hs-varid'>filter</span> <span class='hs-varid'>out_backprop</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16737"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Dilation2DBackpropFilter"</span>
+<a name="line-16738"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16739"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>filter</span> <span class='hs-varid'>out_backprop</span>
+<a name="line-16740"></a><span class='hs-comment'>{-
+<a name="line-16741"></a>attr {
+<a name="line-16742"></a>  allowed_values {
+<a name="line-16743"></a>    list {
+<a name="line-16744"></a>      type: DT_FLOAT
+<a name="line-16745"></a>      type: DT_DOUBLE
+<a name="line-16746"></a>      type: DT_INT32
+<a name="line-16747"></a>      type: DT_INT64
+<a name="line-16748"></a>      type: DT_UINT8
+<a name="line-16749"></a>      type: DT_INT16
+<a name="line-16750"></a>      type: DT_INT8
+<a name="line-16751"></a>      type: DT_UINT16
+<a name="line-16752"></a>      type: DT_HALF
+<a name="line-16753"></a>    }
+<a name="line-16754"></a>  }
+<a name="line-16755"></a>  name: "T"
+<a name="line-16756"></a>  type: "type"
+<a name="line-16757"></a>}
+<a name="line-16758"></a>attr {
+<a name="line-16759"></a>  description: "1-D of length 4. The stride of the sliding window for each dimension of\nthe input tensor. Must be: `[1, stride_height, stride_width, 1]`."
+<a name="line-16760"></a>  has_minimum: true
+<a name="line-16761"></a>  minimum: 4
+<a name="line-16762"></a>  name: "strides"
+<a name="line-16763"></a>  type: "list(int)"
+<a name="line-16764"></a>}
+<a name="line-16765"></a>attr {
+<a name="line-16766"></a>  description: "1-D of length 4. The input stride for atrous morphological dilation.\nMust be: `[1, rate_height, rate_width, 1]`."
+<a name="line-16767"></a>  has_minimum: true
+<a name="line-16768"></a>  minimum: 4
+<a name="line-16769"></a>  name: "rates"
+<a name="line-16770"></a>  type: "list(int)"
+<a name="line-16771"></a>}
+<a name="line-16772"></a>attr {
+<a name="line-16773"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-16774"></a>  description: "The type of padding algorithm to use."
+<a name="line-16775"></a>  name: "padding"
+<a name="line-16776"></a>  type: "string"
+<a name="line-16777"></a>}
+<a name="line-16778"></a>input_arg {
+<a name="line-16779"></a>  description: "4-D with shape `[batch, in_height, in_width, depth]`."
+<a name="line-16780"></a>  name: "input"
+<a name="line-16781"></a>  type_attr: "T"
+<a name="line-16782"></a>}
+<a name="line-16783"></a>input_arg {
+<a name="line-16784"></a>  description: "3-D with shape `[filter_height, filter_width, depth]`."
+<a name="line-16785"></a>  name: "filter"
+<a name="line-16786"></a>  type_attr: "T"
+<a name="line-16787"></a>}
+<a name="line-16788"></a>input_arg {
+<a name="line-16789"></a>  description: "4-D with shape `[batch, out_height, out_width, depth]`."
+<a name="line-16790"></a>  name: "out_backprop"
+<a name="line-16791"></a>  type_attr: "T"
+<a name="line-16792"></a>}
+<a name="line-16793"></a>output_arg {
+<a name="line-16794"></a>  description: "3-D with shape `[filter_height, filter_width, depth]`."
+<a name="line-16795"></a>  name: "filter_backprop"
+<a name="line-16796"></a>  type_attr: "T"
+<a name="line-16797"></a>}
+<a name="line-16798"></a>-}</span>
+<a name="line-16799"></a>
+<a name="line-16800"></a><span class='hs-comment'>-- | Computes rectified linear gradients for a Relu operation.</span>
+<a name="line-16801"></a>
+<a name="line-16802"></a><a name="reluGrad"></a><span class='hs-definition'>reluGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-16803"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-16804"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-16805"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-16806"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-16807"></a>                                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-16808"></a>                                                    <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16809"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __gradients__: The backpropagated gradients to the corresponding Relu operation.</span>
+<a name="line-16810"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __features__: The features passed as input to the corresponding Relu operation, OR</span>
+<a name="line-16811"></a>                           <span class='hs-comment'>-- the outputs of that operation (both work equivalently).</span>
+<a name="line-16812"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __backprops__: `gradients * (features &gt; 0)`.</span>
+<a name="line-16813"></a><span class='hs-definition'>reluGrad</span> <span class='hs-varid'>gradients</span> <span class='hs-varid'>features</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16814"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ReluGrad"</span>
+<a name="line-16815"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16816"></a>        <span class='hs-varid'>gradients</span> <span class='hs-varid'>features</span>
+<a name="line-16817"></a><span class='hs-comment'>{-
+<a name="line-16818"></a>attr {
+<a name="line-16819"></a>  allowed_values {
+<a name="line-16820"></a>    list {
+<a name="line-16821"></a>      type: DT_FLOAT
+<a name="line-16822"></a>      type: DT_DOUBLE
+<a name="line-16823"></a>      type: DT_INT32
+<a name="line-16824"></a>      type: DT_INT64
+<a name="line-16825"></a>      type: DT_UINT8
+<a name="line-16826"></a>      type: DT_INT16
+<a name="line-16827"></a>      type: DT_INT8
+<a name="line-16828"></a>      type: DT_UINT16
+<a name="line-16829"></a>      type: DT_HALF
+<a name="line-16830"></a>    }
+<a name="line-16831"></a>  }
+<a name="line-16832"></a>  name: "T"
+<a name="line-16833"></a>  type: "type"
+<a name="line-16834"></a>}
+<a name="line-16835"></a>input_arg {
+<a name="line-16836"></a>  description: "The backpropagated gradients to the corresponding Relu operation."
+<a name="line-16837"></a>  name: "gradients"
+<a name="line-16838"></a>  type_attr: "T"
+<a name="line-16839"></a>}
+<a name="line-16840"></a>input_arg {
+<a name="line-16841"></a>  description: "The features passed as input to the corresponding Relu operation, OR\nthe outputs of that operation (both work equivalently)."
+<a name="line-16842"></a>  name: "features"
+<a name="line-16843"></a>  type_attr: "T"
+<a name="line-16844"></a>}
+<a name="line-16845"></a>output_arg {
+<a name="line-16846"></a>  description: "`gradients * (features &gt; 0)`."
+<a name="line-16847"></a>  name: "backprops"
+<a name="line-16848"></a>  type_attr: "T"
+<a name="line-16849"></a>}
+<a name="line-16850"></a>-}</span>
+<a name="line-16851"></a>
+<a name="line-16852"></a><span class='hs-comment'>-- | Computes rectified linear 6: `min(max(features, 0), 6)`.</span>
+<a name="line-16853"></a>
+<a name="line-16854"></a><a name="relu6"></a><span class='hs-definition'>relu6</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-16855"></a>                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-16856"></a>                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-16857"></a>                                              <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16858"></a>         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __features__</span>
+<a name="line-16859"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __activations__</span>
+<a name="line-16860"></a><span class='hs-definition'>relu6</span> <span class='hs-varid'>features</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16861"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Relu6"</span>
+<a name="line-16862"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16863"></a>        <span class='hs-varid'>features</span>
+<a name="line-16864"></a><span class='hs-comment'>{-
+<a name="line-16865"></a>attr {
+<a name="line-16866"></a>  allowed_values {
+<a name="line-16867"></a>    list {
+<a name="line-16868"></a>      type: DT_FLOAT
+<a name="line-16869"></a>      type: DT_DOUBLE
+<a name="line-16870"></a>      type: DT_INT32
+<a name="line-16871"></a>      type: DT_INT64
+<a name="line-16872"></a>      type: DT_UINT8
+<a name="line-16873"></a>      type: DT_INT16
+<a name="line-16874"></a>      type: DT_INT8
+<a name="line-16875"></a>      type: DT_UINT16
+<a name="line-16876"></a>      type: DT_HALF
+<a name="line-16877"></a>    }
+<a name="line-16878"></a>  }
+<a name="line-16879"></a>  name: "T"
+<a name="line-16880"></a>  type: "type"
+<a name="line-16881"></a>}
+<a name="line-16882"></a>input_arg { name: "features" type_attr: "T" }
+<a name="line-16883"></a>output_arg { name: "activations" type_attr: "T" }
+<a name="line-16884"></a>-}</span>
+<a name="line-16885"></a>
+<a name="line-16886"></a><a name="resizeBicubic"></a><span class='hs-comment'>-- | Resize `images` to `size` using bicubic interpolation.</span>
+<a name="line-16887"></a><span class='hs-comment'>--</span>
+<a name="line-16888"></a><span class='hs-comment'>-- Input images can be of different types but output images are always float.</span>
+<a name="line-16889"></a><span class='hs-definition'>resizeBicubic</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-16890"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-16891"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-16892"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-16893"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-16894"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-16895"></a>                                                         <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16896"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __images__: 4-D with shape `[batch, height, width, channels]`.</span>
+<a name="line-16897"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __size__: = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The</span>
+<a name="line-16898"></a>                                             <span class='hs-comment'>-- new size for the images.</span>
+<a name="line-16899"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __resized_images__: 4-D with shape</span>
+<a name="line-16900"></a>                 <span class='hs-comment'>-- `[batch, new_height, new_width, channels]`.</span>
+<a name="line-16901"></a><span class='hs-definition'>resizeBicubic</span> <span class='hs-varid'>images</span> <span class='hs-varid'>size</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16902"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ResizeBicubic"</span>
+<a name="line-16903"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16904"></a>        <span class='hs-varid'>images</span> <span class='hs-varid'>size</span>
+<a name="line-16905"></a><span class='hs-comment'>{-
+<a name="line-16906"></a>attr {
+<a name="line-16907"></a>  allowed_values {
+<a name="line-16908"></a>    list {
+<a name="line-16909"></a>      type: DT_UINT8
+<a name="line-16910"></a>      type: DT_INT8
+<a name="line-16911"></a>      type: DT_INT16
+<a name="line-16912"></a>      type: DT_INT32
+<a name="line-16913"></a>      type: DT_INT64
+<a name="line-16914"></a>      type: DT_HALF
+<a name="line-16915"></a>      type: DT_FLOAT
+<a name="line-16916"></a>      type: DT_DOUBLE
+<a name="line-16917"></a>    }
+<a name="line-16918"></a>  }
+<a name="line-16919"></a>  name: "T"
+<a name="line-16920"></a>  type: "type"
+<a name="line-16921"></a>}
+<a name="line-16922"></a>attr {
+<a name="line-16923"></a>  default_value { b: false }
+<a name="line-16924"></a>  description: "If true, rescale input by (new_height - 1) / (height - 1), which\nexactly aligns the 4 corners of images and resized images. If false, rescale\nby new_height / height. Treat similarly the width dimension."
+<a name="line-16925"></a>  name: "align_corners"
+<a name="line-16926"></a>  type: "bool"
+<a name="line-16927"></a>}
+<a name="line-16928"></a>input_arg {
+<a name="line-16929"></a>  description: "4-D with shape `[batch, height, width, channels]`."
+<a name="line-16930"></a>  name: "images"
+<a name="line-16931"></a>  type_attr: "T"
+<a name="line-16932"></a>}
+<a name="line-16933"></a>input_arg {
+<a name="line-16934"></a>  description: "= A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The\nnew size for the images."
+<a name="line-16935"></a>  name: "size"
+<a name="line-16936"></a>  type: DT_INT32
+<a name="line-16937"></a>}
+<a name="line-16938"></a>output_arg {
+<a name="line-16939"></a>  description: "4-D with shape\n`[batch, new_height, new_width, channels]`."
+<a name="line-16940"></a>  name: "resized_images"
+<a name="line-16941"></a>  type: DT_FLOAT
+<a name="line-16942"></a>}
+<a name="line-16943"></a>-}</span>
+<a name="line-16944"></a>
+<a name="line-16945"></a><span class='hs-comment'>-- | Computes rectified linear 6 gradients for a Relu6 operation.</span>
+<a name="line-16946"></a>
+<a name="line-16947"></a><a name="relu6Grad"></a><span class='hs-definition'>relu6Grad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-16948"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-16949"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-16950"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-16951"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-16952"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-16953"></a>                                                     <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-16954"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __gradients__: The backpropagated gradients to the corresponding Relu6 operation.</span>
+<a name="line-16955"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __features__: The features passed as input to the corresponding Relu6 operation.</span>
+<a name="line-16956"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __backprops__: The gradients:</span>
+<a name="line-16957"></a>             <span class='hs-comment'>-- `gradients * features * (features &gt; 0) * (features &lt; 6)`.</span>
+<a name="line-16958"></a><span class='hs-definition'>relu6Grad</span> <span class='hs-varid'>gradients</span> <span class='hs-varid'>features</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-16959"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Relu6Grad"</span>
+<a name="line-16960"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-16961"></a>        <span class='hs-varid'>gradients</span> <span class='hs-varid'>features</span>
+<a name="line-16962"></a><span class='hs-comment'>{-
+<a name="line-16963"></a>attr {
+<a name="line-16964"></a>  allowed_values {
+<a name="line-16965"></a>    list {
+<a name="line-16966"></a>      type: DT_FLOAT
+<a name="line-16967"></a>      type: DT_DOUBLE
+<a name="line-16968"></a>      type: DT_INT32
+<a name="line-16969"></a>      type: DT_INT64
+<a name="line-16970"></a>      type: DT_UINT8
+<a name="line-16971"></a>      type: DT_INT16
+<a name="line-16972"></a>      type: DT_INT8
+<a name="line-16973"></a>      type: DT_UINT16
+<a name="line-16974"></a>      type: DT_HALF
+<a name="line-16975"></a>    }
+<a name="line-16976"></a>  }
+<a name="line-16977"></a>  name: "T"
+<a name="line-16978"></a>  type: "type"
+<a name="line-16979"></a>}
+<a name="line-16980"></a>input_arg {
+<a name="line-16981"></a>  description: "The backpropagated gradients to the corresponding Relu6 operation."
+<a name="line-16982"></a>  name: "gradients"
+<a name="line-16983"></a>  type_attr: "T"
+<a name="line-16984"></a>}
+<a name="line-16985"></a>input_arg {
+<a name="line-16986"></a>  description: "The features passed as input to the corresponding Relu6 operation."
+<a name="line-16987"></a>  name: "features"
+<a name="line-16988"></a>  type_attr: "T"
+<a name="line-16989"></a>}
+<a name="line-16990"></a>output_arg {
+<a name="line-16991"></a>  description: "The gradients:\n`gradients * features * (features &gt; 0) * (features &lt; 6)`."
+<a name="line-16992"></a>  name: "backprops"
+<a name="line-16993"></a>  type_attr: "T"
+<a name="line-16994"></a>}
+<a name="line-16995"></a>-}</span>
+<a name="line-16996"></a>
+<a name="line-16997"></a><a name="sparseTensorDenseMatMul"></a><span class='hs-comment'>-- | Multiply SparseTensor (of rank 2) "A" by dense matrix "B".</span>
+<a name="line-16998"></a><span class='hs-comment'>--</span>
+<a name="line-16999"></a><span class='hs-comment'>-- No validity checking is performed on the indices of A.  However, the following</span>
+<a name="line-17000"></a><span class='hs-comment'>-- input format is recommended for optimal behavior:</span>
+<a name="line-17001"></a><span class='hs-comment'>-- </span>
+<a name="line-17002"></a><span class='hs-comment'>-- if adjoint_a == false:</span>
+<a name="line-17003"></a><span class='hs-comment'>--   A should be sorted in lexicographically increasing order.  Use SparseReorder</span>
+<a name="line-17004"></a><span class='hs-comment'>--   if you're not sure.</span>
+<a name="line-17005"></a><span class='hs-comment'>-- if adjoint_a == true:</span>
+<a name="line-17006"></a><span class='hs-comment'>--   A should be sorted in order of increasing dimension 1 (i.e., "column major"</span>
+<a name="line-17007"></a><span class='hs-comment'>--   order instead of "row major" order).</span>
+<a name="line-17008"></a><span class='hs-definition'>sparseTensorDenseMatMul</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-17009"></a>                           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __a_indices__: 2-D.  The `indices` of the `SparseTensor`, size `[nnz, 2]` Matrix.</span>
+<a name="line-17010"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __a_values__: 1-D.  The `values` of the `SparseTensor`, size `[nnz]` Vector.</span>
+<a name="line-17011"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __a_shape__: 1-D.  The `shape` of the `SparseTensor`, size `[2]` Vector.</span>
+<a name="line-17012"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __b__: 2-D.  A dense Matrix.</span>
+<a name="line-17013"></a>                           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __product__</span>
+<a name="line-17014"></a><span class='hs-definition'>sparseTensorDenseMatMul</span> <span class='hs-varid'>a_indices</span> <span class='hs-varid'>a_values</span> <span class='hs-varid'>a_shape</span> <span class='hs-varid'>b</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17015"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseTensorDenseMatMul"</span>
+<a name="line-17016"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-17017"></a>        <span class='hs-varid'>a_indices</span> <span class='hs-varid'>a_values</span> <span class='hs-varid'>a_shape</span> <span class='hs-varid'>b</span>
+<a name="line-17018"></a><span class='hs-comment'>{-
+<a name="line-17019"></a>attr { name: "T" type: "type" }
+<a name="line-17020"></a>attr {
+<a name="line-17021"></a>  default_value { b: false }
+<a name="line-17022"></a>  description: "Use the adjoint of A in the matrix multiply.  If A is complex, this\nis transpose(conj(A)).  Otherwise it\'s transpose(A)."
+<a name="line-17023"></a>  name: "adjoint_a"
+<a name="line-17024"></a>  type: "bool"
+<a name="line-17025"></a>}
+<a name="line-17026"></a>attr {
+<a name="line-17027"></a>  default_value { b: false }
+<a name="line-17028"></a>  description: "Use the adjoint of B in the matrix multiply.  If B is complex, this\nis transpose(conj(B)).  Otherwise it\'s transpose(B)."
+<a name="line-17029"></a>  name: "adjoint_b"
+<a name="line-17030"></a>  type: "bool"
+<a name="line-17031"></a>}
+<a name="line-17032"></a>input_arg {
+<a name="line-17033"></a>  description: "2-D.  The `indices` of the `SparseTensor`, size `[nnz, 2]` Matrix."
+<a name="line-17034"></a>  name: "a_indices"
+<a name="line-17035"></a>  type: DT_INT64
+<a name="line-17036"></a>}
+<a name="line-17037"></a>input_arg {
+<a name="line-17038"></a>  description: "1-D.  The `values` of the `SparseTensor`, size `[nnz]` Vector."
+<a name="line-17039"></a>  name: "a_values"
+<a name="line-17040"></a>  type_attr: "T"
+<a name="line-17041"></a>}
+<a name="line-17042"></a>input_arg {
+<a name="line-17043"></a>  description: "1-D.  The `shape` of the `SparseTensor`, size `[2]` Vector."
+<a name="line-17044"></a>  name: "a_shape"
+<a name="line-17045"></a>  type: DT_INT64
+<a name="line-17046"></a>}
+<a name="line-17047"></a>input_arg {
+<a name="line-17048"></a>  description: "2-D.  A dense Matrix." name: "b" type_attr: "T"
+<a name="line-17049"></a>}
+<a name="line-17050"></a>output_arg { name: "product" type_attr: "T" }
+<a name="line-17051"></a>-}</span>
+<a name="line-17052"></a>
+<a name="line-17053"></a><span class='hs-comment'>-- | Computes softplus: `log(exp(features) + 1)`.</span>
+<a name="line-17054"></a>
+<a name="line-17055"></a><a name="softplus"></a><span class='hs-definition'>softplus</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-17056"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-17057"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-17058"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-17059"></a>                                                 <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-17060"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __features__</span>
+<a name="line-17061"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __activations__</span>
+<a name="line-17062"></a><span class='hs-definition'>softplus</span> <span class='hs-varid'>features</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17063"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Softplus"</span>
+<a name="line-17064"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-17065"></a>        <span class='hs-varid'>features</span>
+<a name="line-17066"></a><span class='hs-comment'>{-
+<a name="line-17067"></a>attr {
+<a name="line-17068"></a>  allowed_values {
+<a name="line-17069"></a>    list {
+<a name="line-17070"></a>      type: DT_FLOAT
+<a name="line-17071"></a>      type: DT_DOUBLE
+<a name="line-17072"></a>      type: DT_INT32
+<a name="line-17073"></a>      type: DT_INT64
+<a name="line-17074"></a>      type: DT_UINT8
+<a name="line-17075"></a>      type: DT_INT16
+<a name="line-17076"></a>      type: DT_INT8
+<a name="line-17077"></a>      type: DT_UINT16
+<a name="line-17078"></a>      type: DT_HALF
+<a name="line-17079"></a>    }
+<a name="line-17080"></a>  }
+<a name="line-17081"></a>  name: "T"
+<a name="line-17082"></a>  type: "type"
+<a name="line-17083"></a>}
+<a name="line-17084"></a>input_arg { name: "features" type_attr: "T" }
+<a name="line-17085"></a>output_arg { name: "activations" type_attr: "T" }
+<a name="line-17086"></a>-}</span>
+<a name="line-17087"></a>
+<a name="line-17088"></a><a name="batchMatMul"></a><span class='hs-comment'>-- | Multiplies slices of two tensors in batches.</span>
+<a name="line-17089"></a><span class='hs-comment'>--</span>
+<a name="line-17090"></a><span class='hs-comment'>-- Multiplies all slices of `Tensor` `x` and `y` (each slice can be</span>
+<a name="line-17091"></a><span class='hs-comment'>-- viewed as an element of a batch), and arranges the individual results</span>
+<a name="line-17092"></a><span class='hs-comment'>-- in a single output tensor of the same batch size. Each of the</span>
+<a name="line-17093"></a><span class='hs-comment'>-- individual slices can optionally be adjointed (to adjoint a matrix</span>
+<a name="line-17094"></a><span class='hs-comment'>-- means to transpose and conjugate it) before multiplication by setting</span>
+<a name="line-17095"></a><span class='hs-comment'>-- the `adj_x` or `adj_y` flag to `True`, which are by default `False`.</span>
+<a name="line-17096"></a><span class='hs-comment'>-- </span>
+<a name="line-17097"></a><span class='hs-comment'>-- The input tensors `x` and `y` are 3-D or higher with shape `[..., r_x, c_x]`</span>
+<a name="line-17098"></a><span class='hs-comment'>-- and `[..., r_y, c_y]`.</span>
+<a name="line-17099"></a><span class='hs-comment'>-- </span>
+<a name="line-17100"></a><span class='hs-comment'>-- The output tensor is 3-D or higher with shape `[..., r_o, c_o]`, where:</span>
+<a name="line-17101"></a><span class='hs-comment'>-- </span>
+<a name="line-17102"></a><span class='hs-comment'>--     r_o = c_x if adj_x else r_x</span>
+<a name="line-17103"></a><span class='hs-comment'>--     c_o = r_y if adj_y else c_y</span>
+<a name="line-17104"></a><span class='hs-comment'>-- </span>
+<a name="line-17105"></a><span class='hs-comment'>-- It is computed as:</span>
+<a name="line-17106"></a><span class='hs-comment'>-- </span>
+<a name="line-17107"></a><span class='hs-comment'>--     output[..., :, :] = matrix(x[..., :, :]) * matrix(y[..., :, :])</span>
+<a name="line-17108"></a><span class='hs-definition'>batchMatMul</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-17109"></a>                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-17110"></a>                                         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-17111"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-17112"></a>                                         <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-17113"></a>               <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__: 3-D or higher with shape `[..., r_x, c_x]`.</span>
+<a name="line-17114"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__: 3-D or higher with shape `[..., r_y, c_y]`.</span>
+<a name="line-17115"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 3-D or higher with shape `[..., r_o, c_o]`</span>
+<a name="line-17116"></a><span class='hs-definition'>batchMatMul</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17117"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchMatMul"</span>
+<a name="line-17118"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-17119"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-17120"></a><span class='hs-comment'>{-
+<a name="line-17121"></a>attr {
+<a name="line-17122"></a>  allowed_values {
+<a name="line-17123"></a>    list {
+<a name="line-17124"></a>      type: DT_HALF
+<a name="line-17125"></a>      type: DT_FLOAT
+<a name="line-17126"></a>      type: DT_DOUBLE
+<a name="line-17127"></a>      type: DT_INT32
+<a name="line-17128"></a>      type: DT_COMPLEX64
+<a name="line-17129"></a>      type: DT_COMPLEX128
+<a name="line-17130"></a>    }
+<a name="line-17131"></a>  }
+<a name="line-17132"></a>  name: "T"
+<a name="line-17133"></a>  type: "type"
+<a name="line-17134"></a>}
+<a name="line-17135"></a>attr {
+<a name="line-17136"></a>  default_value { b: false }
+<a name="line-17137"></a>  description: "If `True`, adjoint the slices of `x`. Defaults to `False`."
+<a name="line-17138"></a>  name: "adj_x"
+<a name="line-17139"></a>  type: "bool"
+<a name="line-17140"></a>}
+<a name="line-17141"></a>attr {
+<a name="line-17142"></a>  default_value { b: false }
+<a name="line-17143"></a>  description: "If `True`, adjoint the slices of `y`. Defaults to `False`."
+<a name="line-17144"></a>  name: "adj_y"
+<a name="line-17145"></a>  type: "bool"
+<a name="line-17146"></a>}
+<a name="line-17147"></a>input_arg {
+<a name="line-17148"></a>  description: "3-D or higher with shape `[..., r_x, c_x]`."
+<a name="line-17149"></a>  name: "x"
+<a name="line-17150"></a>  type_attr: "T"
+<a name="line-17151"></a>}
+<a name="line-17152"></a>input_arg {
+<a name="line-17153"></a>  description: "3-D or higher with shape `[..., r_y, c_y]`."
+<a name="line-17154"></a>  name: "y"
+<a name="line-17155"></a>  type_attr: "T"
+<a name="line-17156"></a>}
+<a name="line-17157"></a>output_arg {
+<a name="line-17158"></a>  description: "3-D or higher with shape `[..., r_o, c_o]`"
+<a name="line-17159"></a>  name: "output"
+<a name="line-17160"></a>  type_attr: "T"
+<a name="line-17161"></a>}
+<a name="line-17162"></a>-}</span>
+<a name="line-17163"></a>
+<a name="line-17164"></a><span class='hs-comment'>-- | Computes softsign gradients for a softsign operation.</span>
+<a name="line-17165"></a>
+<a name="line-17166"></a><a name="softsignGrad"></a><span class='hs-definition'>softsignGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-17167"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-17168"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-17169"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-17170"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-17171"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-17172"></a>                                                        <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-17173"></a>                <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __gradients__: The backpropagated gradients to the corresponding softsign operation.</span>
+<a name="line-17174"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __features__: The features passed as input to the corresponding softsign operation.</span>
+<a name="line-17175"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __backprops__: The gradients: `gradients / (1 + abs(-features)) ** 2`.</span>
+<a name="line-17176"></a><span class='hs-definition'>softsignGrad</span> <span class='hs-varid'>gradients</span> <span class='hs-varid'>features</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17177"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SoftsignGrad"</span>
+<a name="line-17178"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-17179"></a>        <span class='hs-varid'>gradients</span> <span class='hs-varid'>features</span>
+<a name="line-17180"></a><span class='hs-comment'>{-
+<a name="line-17181"></a>attr {
+<a name="line-17182"></a>  allowed_values {
+<a name="line-17183"></a>    list {
+<a name="line-17184"></a>      type: DT_FLOAT
+<a name="line-17185"></a>      type: DT_DOUBLE
+<a name="line-17186"></a>      type: DT_INT32
+<a name="line-17187"></a>      type: DT_INT64
+<a name="line-17188"></a>      type: DT_UINT8
+<a name="line-17189"></a>      type: DT_INT16
+<a name="line-17190"></a>      type: DT_INT8
+<a name="line-17191"></a>      type: DT_UINT16
+<a name="line-17192"></a>      type: DT_HALF
+<a name="line-17193"></a>    }
+<a name="line-17194"></a>  }
+<a name="line-17195"></a>  name: "T"
+<a name="line-17196"></a>  type: "type"
+<a name="line-17197"></a>}
+<a name="line-17198"></a>input_arg {
+<a name="line-17199"></a>  description: "The backpropagated gradients to the corresponding softsign operation."
+<a name="line-17200"></a>  name: "gradients"
+<a name="line-17201"></a>  type_attr: "T"
+<a name="line-17202"></a>}
+<a name="line-17203"></a>input_arg {
+<a name="line-17204"></a>  description: "The features passed as input to the corresponding softsign operation."
+<a name="line-17205"></a>  name: "features"
+<a name="line-17206"></a>  type_attr: "T"
+<a name="line-17207"></a>}
+<a name="line-17208"></a>output_arg {
+<a name="line-17209"></a>  description: "The gradients: `gradients / (1 + abs(-features)) ** 2`."
+<a name="line-17210"></a>  name: "backprops"
+<a name="line-17211"></a>  type_attr: "T"
+<a name="line-17212"></a>}
+<a name="line-17213"></a>-}</span>
+<a name="line-17214"></a>
+<a name="line-17215"></a><a name="lessEqual"></a><span class='hs-comment'>-- | Returns the truth value of (x &lt;= y) element-wise.</span>
+<a name="line-17216"></a><span class='hs-comment'>--</span>
+<a name="line-17217"></a><span class='hs-comment'>-- *NOTE*: `LessEqual` supports broadcasting. More about broadcasting</span>
+<a name="line-17218"></a><span class='hs-comment'>-- [here](<a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)">http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)</a></span>
+<a name="line-17219"></a><span class='hs-definition'>lessEqual</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-17220"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-17221"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-17222"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-17223"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-17224"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-17225"></a>                                                     <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-17226"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-17227"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-17228"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-17229"></a><span class='hs-definition'>lessEqual</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17230"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"LessEqual"</span>
+<a name="line-17231"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-17232"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-17233"></a><span class='hs-comment'>{-
+<a name="line-17234"></a>attr {
+<a name="line-17235"></a>  allowed_values {
+<a name="line-17236"></a>    list {
+<a name="line-17237"></a>      type: DT_FLOAT
+<a name="line-17238"></a>      type: DT_DOUBLE
+<a name="line-17239"></a>      type: DT_INT32
+<a name="line-17240"></a>      type: DT_INT64
+<a name="line-17241"></a>      type: DT_UINT8
+<a name="line-17242"></a>      type: DT_INT16
+<a name="line-17243"></a>      type: DT_INT8
+<a name="line-17244"></a>      type: DT_UINT16
+<a name="line-17245"></a>      type: DT_HALF
+<a name="line-17246"></a>    }
+<a name="line-17247"></a>  }
+<a name="line-17248"></a>  name: "T"
+<a name="line-17249"></a>  type: "type"
+<a name="line-17250"></a>}
+<a name="line-17251"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-17252"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-17253"></a>output_arg { name: "z" type: DT_BOOL }
+<a name="line-17254"></a>-}</span>
+<a name="line-17255"></a>
+<a name="line-17256"></a><a name="logSoftmax"></a><span class='hs-comment'>-- | Computes log softmax activations.</span>
+<a name="line-17257"></a><span class='hs-comment'>--</span>
+<a name="line-17258"></a><span class='hs-comment'>-- For each batch `i` and class `j` we have</span>
+<a name="line-17259"></a><span class='hs-comment'>-- </span>
+<a name="line-17260"></a><span class='hs-comment'>--     logsoftmax[i, j] = logits[i, j] - log(sum(exp(logits[i])))</span>
+<a name="line-17261"></a><span class='hs-definition'>logSoftmax</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-17262"></a>                                                   <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-17263"></a>              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __logits__: 2-D with shape `[batch_size, num_classes]`.</span>
+<a name="line-17264"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __logsoftmax__: Same shape as `logits`.</span>
+<a name="line-17265"></a><span class='hs-definition'>logSoftmax</span> <span class='hs-varid'>logits</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17266"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"LogSoftmax"</span>
+<a name="line-17267"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-17268"></a>        <span class='hs-varid'>logits</span>
+<a name="line-17269"></a><span class='hs-comment'>{-
+<a name="line-17270"></a>attr {
+<a name="line-17271"></a>  allowed_values {
+<a name="line-17272"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-17273"></a>  }
+<a name="line-17274"></a>  name: "T"
+<a name="line-17275"></a>  type: "type"
+<a name="line-17276"></a>}
+<a name="line-17277"></a>input_arg {
+<a name="line-17278"></a>  description: "2-D with shape `[batch_size, num_classes]`."
+<a name="line-17279"></a>  name: "logits"
+<a name="line-17280"></a>  type_attr: "T"
+<a name="line-17281"></a>}
+<a name="line-17282"></a>output_arg {
+<a name="line-17283"></a>  description: "Same shape as `logits`."
+<a name="line-17284"></a>  name: "logsoftmax"
+<a name="line-17285"></a>  type_attr: "T"
+<a name="line-17286"></a>}
+<a name="line-17287"></a>-}</span>
+<a name="line-17288"></a>
+<a name="line-17289"></a><a name="inTopK"></a><span class='hs-comment'>-- | Says whether the targets are in the top `K` predictions.</span>
+<a name="line-17290"></a><span class='hs-comment'>--</span>
+<a name="line-17291"></a><span class='hs-comment'>-- This outputs a `batch_size` bool array, an entry `out[i]` is `true` if the</span>
+<a name="line-17292"></a><span class='hs-comment'>-- prediction for the target class is among the top `k` predictions among</span>
+<a name="line-17293"></a><span class='hs-comment'>-- all predictions for example `i`. Note that the behavior of `InTopK` differs</span>
+<a name="line-17294"></a><span class='hs-comment'>-- from the `TopK` op in its handling of ties; if multiple classes have the</span>
+<a name="line-17295"></a><span class='hs-comment'>-- same prediction value and straddle the top-`k` boundary, all of those</span>
+<a name="line-17296"></a><span class='hs-comment'>-- classes are considered to be in the top `k`.</span>
+<a name="line-17297"></a><span class='hs-comment'>-- </span>
+<a name="line-17298"></a><span class='hs-comment'>-- More formally, let</span>
+<a name="line-17299"></a><span class='hs-comment'>-- </span>
+<a name="line-17300"></a><span class='hs-comment'>--   \\(predictions_i\\) be the predictions for all classes for example `i`,</span>
+<a name="line-17301"></a><span class='hs-comment'>--   \\(targets_i\\) be the target class for example `i`,</span>
+<a name="line-17302"></a><span class='hs-comment'>--   \\(out_i\\) be the output for example `i`,</span>
+<a name="line-17303"></a><span class='hs-comment'>-- </span>
+<a name="line-17304"></a><span class='hs-comment'>-- $$out_i = predictions_{i, targets_i} \in TopKIncludingTies(predictions_i)$$</span>
+<a name="line-17305"></a><span class='hs-definition'>inTopK</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-17306"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-17307"></a>          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __k__: Number of top elements to look at for computing precision.</span>
+<a name="line-17308"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __predictions__: A `batch_size` x `classes` tensor.</span>
+<a name="line-17309"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __targets__: A `batch_size` vector of class ids.</span>
+<a name="line-17310"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __precision__: Computed Precision at `k` as a `bool Tensor`.</span>
+<a name="line-17311"></a><span class='hs-definition'>inTopK</span> <span class='hs-varid'>k</span> <span class='hs-varid'>predictions</span> <span class='hs-varid'>targets</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17312"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"InTopK"</span>
+<a name="line-17313"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-17314"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"k"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>k</span><span class='hs-layout'>)</span>
+<a name="line-17315"></a>        <span class='hs-varid'>predictions</span> <span class='hs-varid'>targets</span>
+<a name="line-17316"></a><span class='hs-comment'>{-
+<a name="line-17317"></a>attr {
+<a name="line-17318"></a>  description: "Number of top elements to look at for computing precision."
+<a name="line-17319"></a>  name: "k"
+<a name="line-17320"></a>  type: "int"
+<a name="line-17321"></a>}
+<a name="line-17322"></a>attr {
+<a name="line-17323"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-17324"></a>  default_value { type: DT_INT32 }
+<a name="line-17325"></a>  name: "T"
+<a name="line-17326"></a>  type: "type"
+<a name="line-17327"></a>}
+<a name="line-17328"></a>input_arg {
+<a name="line-17329"></a>  description: "A `batch_size` x `classes` tensor."
+<a name="line-17330"></a>  name: "predictions"
+<a name="line-17331"></a>  type: DT_FLOAT
+<a name="line-17332"></a>}
+<a name="line-17333"></a>input_arg {
+<a name="line-17334"></a>  description: "A `batch_size` vector of class ids."
+<a name="line-17335"></a>  name: "targets"
+<a name="line-17336"></a>  type_attr: "T"
+<a name="line-17337"></a>}
+<a name="line-17338"></a>output_arg {
+<a name="line-17339"></a>  description: "Computed Precision at `k` as a `bool Tensor`."
+<a name="line-17340"></a>  name: "precision"
+<a name="line-17341"></a>  type: DT_BOOL
+<a name="line-17342"></a>}
+<a name="line-17343"></a>-}</span>
+<a name="line-17344"></a>
+<a name="line-17345"></a><a name="matrixDiag"></a><span class='hs-comment'>-- | Returns a batched diagonal tensor with a given batched diagonal values.</span>
+<a name="line-17346"></a><span class='hs-comment'>--</span>
+<a name="line-17347"></a><span class='hs-comment'>-- Given a `diagonal`, this operation returns a tensor with the `diagonal` and</span>
+<a name="line-17348"></a><span class='hs-comment'>-- everything else padded with zeros. The diagonal is computed as follows:</span>
+<a name="line-17349"></a><span class='hs-comment'>-- </span>
+<a name="line-17350"></a><span class='hs-comment'>-- Assume `diagonal` has `k` dimensions `[I, J, K, ..., N]`, then the output is a</span>
+<a name="line-17351"></a><span class='hs-comment'>-- tensor of rank `k+1` with dimensions [I, J, K, ..., N, N]` where:</span>
+<a name="line-17352"></a><span class='hs-comment'>-- </span>
+<a name="line-17353"></a><span class='hs-comment'>-- `output[i, j, k, ..., m, n] = 1{m=n} * diagonal[i, j, k, ..., n]`.</span>
+<a name="line-17354"></a><span class='hs-comment'>-- </span>
+<a name="line-17355"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-17356"></a><span class='hs-comment'>-- </span>
+<a name="line-17357"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-17358"></a><span class='hs-comment'>-- # 'diagonal' is [[1, 2, 3, 4], [5, 6, 7, 8]]</span>
+<a name="line-17359"></a><span class='hs-comment'>-- </span>
+<a name="line-17360"></a><span class='hs-comment'>-- and diagonal.shape = (2, 4)</span>
+<a name="line-17361"></a><span class='hs-comment'>-- </span>
+<a name="line-17362"></a><span class='hs-comment'>-- tf.matrix_diag(diagonal) ==&gt; [[[1, 0, 0, 0]</span>
+<a name="line-17363"></a><span class='hs-comment'>--                                      [0, 2, 0, 0]</span>
+<a name="line-17364"></a><span class='hs-comment'>--                                      [0, 0, 3, 0]</span>
+<a name="line-17365"></a><span class='hs-comment'>--                                      [0, 0, 0, 4]],</span>
+<a name="line-17366"></a><span class='hs-comment'>--                                     [[5, 0, 0, 0]</span>
+<a name="line-17367"></a><span class='hs-comment'>--                                      [0, 6, 0, 0]</span>
+<a name="line-17368"></a><span class='hs-comment'>--                                      [0, 0, 7, 0]</span>
+<a name="line-17369"></a><span class='hs-comment'>--                                      [0, 0, 0, 8]]]</span>
+<a name="line-17370"></a><span class='hs-comment'>-- </span>
+<a name="line-17371"></a><span class='hs-comment'>-- which has shape (2, 4, 4)</span>
+<a name="line-17372"></a><span class='hs-comment'>-- ```</span>
+<a name="line-17373"></a><span class='hs-definition'>matrixDiag</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-17374"></a>              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __diagonal__: Rank `k`, where `k &gt;= 1`.</span>
+<a name="line-17375"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Rank `k+1`, with `output.shape = diagonal.shape + [diagonal.shape[-1]]`.</span>
+<a name="line-17376"></a><span class='hs-definition'>matrixDiag</span> <span class='hs-varid'>diagonal</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17377"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MatrixDiag"</span>
+<a name="line-17378"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-17379"></a>        <span class='hs-varid'>diagonal</span>
+<a name="line-17380"></a><span class='hs-comment'>{-
+<a name="line-17381"></a>attr { name: "T" type: "type" }
+<a name="line-17382"></a>input_arg {
+<a name="line-17383"></a>  description: "Rank `k`, where `k &gt;= 1`."
+<a name="line-17384"></a>  name: "diagonal"
+<a name="line-17385"></a>  type_attr: "T"
+<a name="line-17386"></a>}
+<a name="line-17387"></a>output_arg {
+<a name="line-17388"></a>  description: "Rank `k+1`, with `output.shape = diagonal.shape + [diagonal.shape[-1]]`."
+<a name="line-17389"></a>  name: "output"
+<a name="line-17390"></a>  type_attr: "T"
+<a name="line-17391"></a>}
+<a name="line-17392"></a>-}</span>
+<a name="line-17393"></a>
+<a name="line-17394"></a><span class='hs-comment'>-- | Performs 3D max pooling on the input.</span>
+<a name="line-17395"></a>
+<a name="line-17396"></a><a name="maxPool3D"></a><span class='hs-definition'>maxPool3D</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-17397"></a>                                                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-17398"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-17399"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-17400"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-17401"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-17402"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-17403"></a>                                                  <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-17404"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Shape `[batch, depth, rows, cols, channels]` tensor to pool over.</span>
+<a name="line-17405"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The max pooled output tensor.</span>
+<a name="line-17406"></a><span class='hs-definition'>maxPool3D</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17407"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MaxPool3D"</span>
+<a name="line-17408"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-17409"></a>        <span class='hs-varid'>input</span>
+<a name="line-17410"></a><span class='hs-comment'>{-
+<a name="line-17411"></a>attr {
+<a name="line-17412"></a>  description: "1-D tensor of length 5. The size of the window for each dimension of\nthe input tensor. Must have `ksize[0] = ksize[4] = 1`."
+<a name="line-17413"></a>  has_minimum: true
+<a name="line-17414"></a>  minimum: 5
+<a name="line-17415"></a>  name: "ksize"
+<a name="line-17416"></a>  type: "list(int)"
+<a name="line-17417"></a>}
+<a name="line-17418"></a>attr {
+<a name="line-17419"></a>  description: "1-D tensor of length 5. The stride of the sliding window for each\ndimension of `input`. Must have `strides[0] = strides[4] = 1`."
+<a name="line-17420"></a>  has_minimum: true
+<a name="line-17421"></a>  minimum: 5
+<a name="line-17422"></a>  name: "strides"
+<a name="line-17423"></a>  type: "list(int)"
+<a name="line-17424"></a>}
+<a name="line-17425"></a>attr {
+<a name="line-17426"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-17427"></a>  description: "The type of padding algorithm to use."
+<a name="line-17428"></a>  name: "padding"
+<a name="line-17429"></a>  type: "string"
+<a name="line-17430"></a>}
+<a name="line-17431"></a>attr {
+<a name="line-17432"></a>  allowed_values {
+<a name="line-17433"></a>    list {
+<a name="line-17434"></a>      type: DT_FLOAT
+<a name="line-17435"></a>      type: DT_DOUBLE
+<a name="line-17436"></a>      type: DT_INT64
+<a name="line-17437"></a>      type: DT_INT32
+<a name="line-17438"></a>      type: DT_UINT8
+<a name="line-17439"></a>      type: DT_UINT16
+<a name="line-17440"></a>      type: DT_INT16
+<a name="line-17441"></a>      type: DT_INT8
+<a name="line-17442"></a>      type: DT_COMPLEX64
+<a name="line-17443"></a>      type: DT_COMPLEX128
+<a name="line-17444"></a>      type: DT_QINT8
+<a name="line-17445"></a>      type: DT_QUINT8
+<a name="line-17446"></a>      type: DT_QINT32
+<a name="line-17447"></a>      type: DT_HALF
+<a name="line-17448"></a>    }
+<a name="line-17449"></a>  }
+<a name="line-17450"></a>  name: "T"
+<a name="line-17451"></a>  type: "type"
+<a name="line-17452"></a>}
+<a name="line-17453"></a>input_arg {
+<a name="line-17454"></a>  description: "Shape `[batch, depth, rows, cols, channels]` tensor to pool over."
+<a name="line-17455"></a>  name: "input"
+<a name="line-17456"></a>  type_attr: "T"
+<a name="line-17457"></a>}
+<a name="line-17458"></a>output_arg {
+<a name="line-17459"></a>  description: "The max pooled output tensor."
+<a name="line-17460"></a>  name: "output"
+<a name="line-17461"></a>  type_attr: "T"
+<a name="line-17462"></a>}
+<a name="line-17463"></a>-}</span>
+<a name="line-17464"></a>
+<a name="line-17465"></a><a name="topK"></a><span class='hs-comment'>-- | Finds values and indices of the `k` largest elements for the last dimension.</span>
+<a name="line-17466"></a><span class='hs-comment'>--</span>
+<a name="line-17467"></a><span class='hs-comment'>-- If the input is a vector (rank-1), finds the `k` largest entries in the vector</span>
+<a name="line-17468"></a><span class='hs-comment'>-- and outputs their values and indices as vectors.  Thus `values[j]` is the</span>
+<a name="line-17469"></a><span class='hs-comment'>-- `j`-th largest entry in `input`, and its index is `indices[j]`.</span>
+<a name="line-17470"></a><span class='hs-comment'>-- </span>
+<a name="line-17471"></a><span class='hs-comment'>-- For matrices (resp. higher rank input), computes the top `k` entries in each</span>
+<a name="line-17472"></a><span class='hs-comment'>-- row (resp. vector along the last dimension).  Thus,</span>
+<a name="line-17473"></a><span class='hs-comment'>-- </span>
+<a name="line-17474"></a><span class='hs-comment'>--     values.shape = indices.shape = input.shape[:-1] + [k]</span>
+<a name="line-17475"></a><span class='hs-comment'>-- </span>
+<a name="line-17476"></a><span class='hs-comment'>-- If two elements are equal, the lower-index element appears first.</span>
+<a name="line-17477"></a><span class='hs-comment'>-- </span>
+<a name="line-17478"></a><span class='hs-comment'>-- If `k` varies dynamically, use `TopKV2` below.</span>
+<a name="line-17479"></a><span class='hs-definition'>topK</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-17480"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-17481"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-17482"></a>                                             <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-17483"></a>        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __k__: Number of top elements to look for along the last dimension (along each</span>
+<a name="line-17484"></a>                       <span class='hs-comment'>-- row for matrices).</span>
+<a name="line-17485"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: 1-D or higher with last dimension at least `k`.</span>
+<a name="line-17486"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-17487"></a>        <span class='hs-comment'>-- ^ (__values__, __indices__)</span>
+<a name="line-17488"></a>        <span class='hs-comment'>--</span>
+<a name="line-17489"></a>        <span class='hs-comment'>-- * __values__: The `k` largest elements along each last dimensional slice.</span>
+<a name="line-17490"></a>        <span class='hs-comment'>--</span>
+<a name="line-17491"></a>        <span class='hs-comment'>-- * __indices__: The indices of `values` within the last dimension of `input`.</span>
+<a name="line-17492"></a><span class='hs-definition'>topK</span> <span class='hs-varid'>k</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17493"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TopK"</span>
+<a name="line-17494"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-17495"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"k"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>k</span><span class='hs-layout'>)</span>
+<a name="line-17496"></a>        <span class='hs-varid'>input</span>
+<a name="line-17497"></a><span class='hs-comment'>{-
+<a name="line-17498"></a>attr {
+<a name="line-17499"></a>  description: "Number of top elements to look for along the last dimension (along each\nrow for matrices)."
+<a name="line-17500"></a>  has_minimum: true
+<a name="line-17501"></a>  name: "k"
+<a name="line-17502"></a>  type: "int"
+<a name="line-17503"></a>}
+<a name="line-17504"></a>attr {
+<a name="line-17505"></a>  default_value { b: true }
+<a name="line-17506"></a>  description: "If true the resulting `k` elements will be sorted by the values in\ndescending order."
+<a name="line-17507"></a>  name: "sorted"
+<a name="line-17508"></a>  type: "bool"
+<a name="line-17509"></a>}
+<a name="line-17510"></a>attr {
+<a name="line-17511"></a>  allowed_values {
+<a name="line-17512"></a>    list {
+<a name="line-17513"></a>      type: DT_FLOAT
+<a name="line-17514"></a>      type: DT_DOUBLE
+<a name="line-17515"></a>      type: DT_INT32
+<a name="line-17516"></a>      type: DT_INT64
+<a name="line-17517"></a>      type: DT_UINT8
+<a name="line-17518"></a>      type: DT_INT16
+<a name="line-17519"></a>      type: DT_INT8
+<a name="line-17520"></a>      type: DT_UINT16
+<a name="line-17521"></a>      type: DT_HALF
+<a name="line-17522"></a>    }
+<a name="line-17523"></a>  }
+<a name="line-17524"></a>  name: "T"
+<a name="line-17525"></a>  type: "type"
+<a name="line-17526"></a>}
+<a name="line-17527"></a>input_arg {
+<a name="line-17528"></a>  description: "1-D or higher with last dimension at least `k`."
+<a name="line-17529"></a>  name: "input"
+<a name="line-17530"></a>  type_attr: "T"
+<a name="line-17531"></a>}
+<a name="line-17532"></a>output_arg {
+<a name="line-17533"></a>  description: "The `k` largest elements along each last dimensional slice."
+<a name="line-17534"></a>  name: "values"
+<a name="line-17535"></a>  type_attr: "T"
+<a name="line-17536"></a>}
+<a name="line-17537"></a>output_arg {
+<a name="line-17538"></a>  description: "The indices of `values` within the last dimension of `input`."
+<a name="line-17539"></a>  name: "indices"
+<a name="line-17540"></a>  type: DT_INT32
+<a name="line-17541"></a>}
+<a name="line-17542"></a>-}</span>
+<a name="line-17543"></a>
+<a name="line-17544"></a><a name="topKV2"></a><span class='hs-comment'>-- | Finds values and indices of the `k` largest elements for the last dimension.</span>
+<a name="line-17545"></a><span class='hs-comment'>--</span>
+<a name="line-17546"></a><span class='hs-comment'>-- If the input is a vector (rank-1), finds the `k` largest entries in the vector</span>
+<a name="line-17547"></a><span class='hs-comment'>-- and outputs their values and indices as vectors.  Thus `values[j]` is the</span>
+<a name="line-17548"></a><span class='hs-comment'>-- `j`-th largest entry in `input`, and its index is `indices[j]`.</span>
+<a name="line-17549"></a><span class='hs-comment'>-- </span>
+<a name="line-17550"></a><span class='hs-comment'>-- For matrices (resp. higher rank input), computes the top `k` entries in each</span>
+<a name="line-17551"></a><span class='hs-comment'>-- row (resp. vector along the last dimension).  Thus,</span>
+<a name="line-17552"></a><span class='hs-comment'>-- </span>
+<a name="line-17553"></a><span class='hs-comment'>--     values.shape = indices.shape = input.shape[:-1] + [k]</span>
+<a name="line-17554"></a><span class='hs-comment'>-- </span>
+<a name="line-17555"></a><span class='hs-comment'>-- If two elements are equal, the lower-index element appears first.</span>
+<a name="line-17556"></a><span class='hs-comment'>-- </span>
+<a name="line-17557"></a><span class='hs-comment'>-- This is the same as `TopK`, but takes `k` as in input rather than an attr.</span>
+<a name="line-17558"></a><span class='hs-definition'>topKV2</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-17559"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-17560"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-17561"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-17562"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-17563"></a>                                                  <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-17564"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: 1-D or higher with last dimension at least `k`.</span>
+<a name="line-17565"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __k__: 0-D.  Number of top elements to look for along the last dimension (along each</span>
+<a name="line-17566"></a>                                      <span class='hs-comment'>-- row for matrices).</span>
+<a name="line-17567"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-17568"></a>          <span class='hs-comment'>-- ^ (__values__, __indices__)</span>
+<a name="line-17569"></a>          <span class='hs-comment'>--</span>
+<a name="line-17570"></a>          <span class='hs-comment'>-- * __values__: The `k` largest elements along each last dimensional slice.</span>
+<a name="line-17571"></a>          <span class='hs-comment'>--</span>
+<a name="line-17572"></a>          <span class='hs-comment'>-- * __indices__: The indices of `values` within the last dimension of `input`.</span>
+<a name="line-17573"></a><span class='hs-definition'>topKV2</span> <span class='hs-varid'>input</span> <span class='hs-varid'>k</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17574"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TopKV2"</span>
+<a name="line-17575"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-17576"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>k</span>
+<a name="line-17577"></a><span class='hs-comment'>{-
+<a name="line-17578"></a>attr {
+<a name="line-17579"></a>  default_value { b: true }
+<a name="line-17580"></a>  description: "If true the resulting `k` elements will be sorted by the values in\ndescending order."
+<a name="line-17581"></a>  name: "sorted"
+<a name="line-17582"></a>  type: "bool"
+<a name="line-17583"></a>}
+<a name="line-17584"></a>attr {
+<a name="line-17585"></a>  allowed_values {
+<a name="line-17586"></a>    list {
+<a name="line-17587"></a>      type: DT_FLOAT
+<a name="line-17588"></a>      type: DT_DOUBLE
+<a name="line-17589"></a>      type: DT_INT32
+<a name="line-17590"></a>      type: DT_INT64
+<a name="line-17591"></a>      type: DT_UINT8
+<a name="line-17592"></a>      type: DT_INT16
+<a name="line-17593"></a>      type: DT_INT8
+<a name="line-17594"></a>      type: DT_UINT16
+<a name="line-17595"></a>      type: DT_HALF
+<a name="line-17596"></a>    }
+<a name="line-17597"></a>  }
+<a name="line-17598"></a>  name: "T"
+<a name="line-17599"></a>  type: "type"
+<a name="line-17600"></a>}
+<a name="line-17601"></a>input_arg {
+<a name="line-17602"></a>  description: "1-D or higher with last dimension at least `k`."
+<a name="line-17603"></a>  name: "input"
+<a name="line-17604"></a>  type_attr: "T"
+<a name="line-17605"></a>}
+<a name="line-17606"></a>input_arg {
+<a name="line-17607"></a>  description: "0-D.  Number of top elements to look for along the last dimension (along each\nrow for matrices)."
+<a name="line-17608"></a>  name: "k"
+<a name="line-17609"></a>  type: DT_INT32
+<a name="line-17610"></a>}
+<a name="line-17611"></a>output_arg {
+<a name="line-17612"></a>  description: "The `k` largest elements along each last dimensional slice."
+<a name="line-17613"></a>  name: "values"
+<a name="line-17614"></a>  type_attr: "T"
+<a name="line-17615"></a>}
+<a name="line-17616"></a>output_arg {
+<a name="line-17617"></a>  description: "The indices of `values` within the last dimension of `input`."
+<a name="line-17618"></a>  name: "indices"
+<a name="line-17619"></a>  type: DT_INT32
+<a name="line-17620"></a>}
+<a name="line-17621"></a>-}</span>
+<a name="line-17622"></a>
+<a name="line-17623"></a><a name="fractionalMaxPool"></a><span class='hs-comment'>-- | Performs fractional max pooling on the input.</span>
+<a name="line-17624"></a><span class='hs-comment'>--</span>
+<a name="line-17625"></a><span class='hs-comment'>-- Fractional max pooling is slightly different than regular max pooling.  In</span>
+<a name="line-17626"></a><span class='hs-comment'>-- regular max pooling, you downsize an input set by taking the maximum value of</span>
+<a name="line-17627"></a><span class='hs-comment'>-- smaller N x N subsections of the set (often 2x2), and try to reduce the set by</span>
+<a name="line-17628"></a><span class='hs-comment'>-- a factor of N, where N is an integer.  Fractional max pooling, as you might</span>
+<a name="line-17629"></a><span class='hs-comment'>-- expect from the word "fractional", means that the overall reduction ratio N</span>
+<a name="line-17630"></a><span class='hs-comment'>-- does not have to be an integer.</span>
+<a name="line-17631"></a><span class='hs-comment'>-- </span>
+<a name="line-17632"></a><span class='hs-comment'>-- The sizes of the pooling regions are generated randomly but are fairly uniform.</span>
+<a name="line-17633"></a><span class='hs-comment'>-- For example, let's look at the height dimension, and the constraints on the</span>
+<a name="line-17634"></a><span class='hs-comment'>-- list of rows that will be pool boundaries.</span>
+<a name="line-17635"></a><span class='hs-comment'>-- </span>
+<a name="line-17636"></a><span class='hs-comment'>-- First we define the following:</span>
+<a name="line-17637"></a><span class='hs-comment'>-- </span>
+<a name="line-17638"></a><span class='hs-comment'>-- 1.  input_row_length : the number of rows from the input set</span>
+<a name="line-17639"></a><span class='hs-comment'>-- 2.  output_row_length : which will be smaller than the input</span>
+<a name="line-17640"></a><span class='hs-comment'>-- 3.  alpha = input_row_length / output_row_length : our reduction ratio</span>
+<a name="line-17641"></a><span class='hs-comment'>-- 4.  K = floor(alpha)</span>
+<a name="line-17642"></a><span class='hs-comment'>-- 5.  row_pooling_sequence : this is the result list of pool boundary rows</span>
+<a name="line-17643"></a><span class='hs-comment'>-- </span>
+<a name="line-17644"></a><span class='hs-comment'>-- Then, row_pooling_sequence should satisfy:</span>
+<a name="line-17645"></a><span class='hs-comment'>-- </span>
+<a name="line-17646"></a><span class='hs-comment'>-- 1.  a[0] = 0 : the first value of the sequence is 0</span>
+<a name="line-17647"></a><span class='hs-comment'>-- 2.  a[end] = input_row_length : the last value of the sequence is the size</span>
+<a name="line-17648"></a><span class='hs-comment'>-- 3.  K &lt;= (a[i+1] - a[i]) &lt;= K+1 : all intervals are K or K+1 size</span>
+<a name="line-17649"></a><span class='hs-comment'>-- 4.  length(row_pooling_sequence) = output_row_length+1</span>
+<a name="line-17650"></a><span class='hs-comment'>-- </span>
+<a name="line-17651"></a><span class='hs-comment'>-- For more details on fractional max pooling, see this paper:</span>
+<a name="line-17652"></a><span class='hs-comment'>-- [Benjamin Graham, Fractional Max-Pooling]</span>
+<a name="line-17653"></a><span class='hs-comment'>-- (<a href="http://arxiv.org/abs/1412.6071)">http://arxiv.org/abs/1412.6071)</a></span>
+<a name="line-17654"></a><span class='hs-definition'>fractionalMaxPool</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-17655"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-17656"></a>                                                          <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-17657"></a>                     <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: 4-D with shape `[batch, height, width, channels]`.</span>
+<a name="line-17658"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-17659"></a>                         <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-17660"></a>                     <span class='hs-comment'>-- ^ (__output__, __row_pooling_sequence__, __col_pooling_sequence__)</span>
+<a name="line-17661"></a>                     <span class='hs-comment'>--</span>
+<a name="line-17662"></a>                     <span class='hs-comment'>-- * __output__: output tensor after fractional max pooling.</span>
+<a name="line-17663"></a>                     <span class='hs-comment'>--</span>
+<a name="line-17664"></a>                     <span class='hs-comment'>-- * __row_pooling_sequence__: row pooling sequence, needed to calculate gradient.</span>
+<a name="line-17665"></a>                     <span class='hs-comment'>--</span>
+<a name="line-17666"></a>                     <span class='hs-comment'>-- * __col_pooling_sequence__: column pooling sequence, needed to calculate gradient.</span>
+<a name="line-17667"></a><span class='hs-definition'>fractionalMaxPool</span> <span class='hs-varid'>value</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17668"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"FractionalMaxPool"</span>
+<a name="line-17669"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-17670"></a>        <span class='hs-varid'>value</span>
+<a name="line-17671"></a><span class='hs-comment'>{-
+<a name="line-17672"></a>attr {
+<a name="line-17673"></a>  description: "Pooling ratio for each dimension of `value`, currently only\nsupports row and col dimension and should be &gt;= 1.0. For example, a valid\npooling ratio looks like [1.0, 1.44, 1.73, 1.0]. The first and last elements\nmust be 1.0 because we don\'t allow pooling on batch and channels\ndimensions. 1.44 and 1.73 are pooling ratio on height and width dimensions\nrespectively."
+<a name="line-17674"></a>  has_minimum: true
+<a name="line-17675"></a>  minimum: 4
+<a name="line-17676"></a>  name: "pooling_ratio"
+<a name="line-17677"></a>  type: "list(float)"
+<a name="line-17678"></a>}
+<a name="line-17679"></a>attr {
+<a name="line-17680"></a>  default_value { b: false }
+<a name="line-17681"></a>  description: "When set to True, generates the pooling sequence in a\npseudorandom fashion, otherwise, in a random fashion. Check paper [Benjamin\nGraham, Fractional Max-Pooling] (<a href="http://arxiv.org/abs/1412.6071)">http://arxiv.org/abs/1412.6071)</a> for\ndifference between pseudorandom and random."
+<a name="line-17682"></a>  name: "pseudo_random"
+<a name="line-17683"></a>  type: "bool"
+<a name="line-17684"></a>}
+<a name="line-17685"></a>attr {
+<a name="line-17686"></a>  default_value { b: false }
+<a name="line-17687"></a>  description: "When set to True, it means when pooling, the values at the boundary\nof adjacent pooling cells are used by both cells. For example:\n\n`index  0  1  2  3  4`\n\n`value  20 5  16 3  7`\n\nIf the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.\nThe result would be [20, 16] for fractional max pooling."
+<a name="line-17688"></a>  name: "overlapping"
+<a name="line-17689"></a>  type: "bool"
+<a name="line-17690"></a>}
+<a name="line-17691"></a>attr {
+<a name="line-17692"></a>  default_value { b: false }
+<a name="line-17693"></a>  description: "When set to True, a fixed pooling region will be used when\niterating over a FractionalMaxPool node in the computation graph. Mainly used\nin unit test to make FractionalMaxPool deterministic."
+<a name="line-17694"></a>  name: "deterministic"
+<a name="line-17695"></a>  type: "bool"
+<a name="line-17696"></a>}
+<a name="line-17697"></a>attr {
+<a name="line-17698"></a>  default_value { i: 0 }
+<a name="line-17699"></a>  description: "If either seed or seed2 are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-17700"></a>  name: "seed"
+<a name="line-17701"></a>  type: "int"
+<a name="line-17702"></a>}
+<a name="line-17703"></a>attr {
+<a name="line-17704"></a>  default_value { i: 0 }
+<a name="line-17705"></a>  description: "An second seed to avoid seed collision."
+<a name="line-17706"></a>  name: "seed2"
+<a name="line-17707"></a>  type: "int"
+<a name="line-17708"></a>}
+<a name="line-17709"></a>attr {
+<a name="line-17710"></a>  allowed_values {
+<a name="line-17711"></a>    list {
+<a name="line-17712"></a>      type: DT_FLOAT type: DT_DOUBLE type: DT_INT32 type: DT_INT64
+<a name="line-17713"></a>    }
+<a name="line-17714"></a>  }
+<a name="line-17715"></a>  name: "T"
+<a name="line-17716"></a>  type: "type"
+<a name="line-17717"></a>}
+<a name="line-17718"></a>input_arg {
+<a name="line-17719"></a>  description: "4-D with shape `[batch, height, width, channels]`."
+<a name="line-17720"></a>  name: "value"
+<a name="line-17721"></a>  type_attr: "T"
+<a name="line-17722"></a>}
+<a name="line-17723"></a>output_arg {
+<a name="line-17724"></a>  description: "output tensor after fractional max pooling."
+<a name="line-17725"></a>  name: "output"
+<a name="line-17726"></a>  type_attr: "T"
+<a name="line-17727"></a>}
+<a name="line-17728"></a>output_arg {
+<a name="line-17729"></a>  description: "row pooling sequence, needed to calculate gradient."
+<a name="line-17730"></a>  name: "row_pooling_sequence"
+<a name="line-17731"></a>  type: DT_INT64
+<a name="line-17732"></a>}
+<a name="line-17733"></a>output_arg {
+<a name="line-17734"></a>  description: "column pooling sequence, needed to calculate gradient."
+<a name="line-17735"></a>  name: "col_pooling_sequence"
+<a name="line-17736"></a>  type: DT_INT64
+<a name="line-17737"></a>}
+<a name="line-17738"></a>-}</span>
+<a name="line-17739"></a>
+<a name="line-17740"></a><a name="matrixBandPart"></a><span class='hs-comment'>-- | Copy a tensor setting everything outside a central band in each innermost matrix</span>
+<a name="line-17741"></a><span class='hs-comment'>--</span>
+<a name="line-17742"></a><span class='hs-comment'>-- to zero.</span>
+<a name="line-17743"></a><span class='hs-comment'>-- </span>
+<a name="line-17744"></a><span class='hs-comment'>-- The `band` part is computed as follows:</span>
+<a name="line-17745"></a><span class='hs-comment'>-- Assume `input` has `k` dimensions `[I, J, K, ..., M, N]`, then the output is a</span>
+<a name="line-17746"></a><span class='hs-comment'>-- tensor with the same shape where</span>
+<a name="line-17747"></a><span class='hs-comment'>-- </span>
+<a name="line-17748"></a><span class='hs-comment'>-- `band[i, j, k, ..., m, n] = in_band(m, n) * input[i, j, k, ..., m, n]`.</span>
+<a name="line-17749"></a><span class='hs-comment'>-- </span>
+<a name="line-17750"></a><span class='hs-comment'>-- The indicator function 'in_band(m, n)` is one if</span>
+<a name="line-17751"></a><span class='hs-comment'>-- `(num_lower &lt; 0 || (m-n) &lt;= num_lower)) &amp;&amp;</span>
+<a name="line-17752"></a><span class='hs-comment'>-- (num_upper &lt; 0 || (n-m) &lt;= num_upper)`, and zero otherwise.</span>
+<a name="line-17753"></a><span class='hs-comment'>-- </span>
+<a name="line-17754"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-17755"></a><span class='hs-comment'>-- </span>
+<a name="line-17756"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-17757"></a><span class='hs-comment'>-- # if 'input' is [[ 0,  1,  2, 3]</span>
+<a name="line-17758"></a><span class='hs-comment'>--                  [-1,  0,  1, 2]</span>
+<a name="line-17759"></a><span class='hs-comment'>--                  [-2, -1,  0, 1]</span>
+<a name="line-17760"></a><span class='hs-comment'>--                  [-3, -2, -1, 0]],</span>
+<a name="line-17761"></a><span class='hs-comment'>-- </span>
+<a name="line-17762"></a><span class='hs-comment'>-- tf.matrix_band_part(input, 1, -1) ==&gt; [[ 0,  1,  2, 3]</span>
+<a name="line-17763"></a><span class='hs-comment'>--                                              [-1,  0,  1, 2]</span>
+<a name="line-17764"></a><span class='hs-comment'>--                                              [ 0, -1,  0, 1]</span>
+<a name="line-17765"></a><span class='hs-comment'>--                                              [ 0,  0, -1, 0]],</span>
+<a name="line-17766"></a><span class='hs-comment'>-- </span>
+<a name="line-17767"></a><span class='hs-comment'>-- tf.matrix_band_part(input, 2, 1) ==&gt; [[ 0,  1,  0, 0]</span>
+<a name="line-17768"></a><span class='hs-comment'>--                                             [-1,  0,  1, 0]</span>
+<a name="line-17769"></a><span class='hs-comment'>--                                             [-2, -1,  0, 1]</span>
+<a name="line-17770"></a><span class='hs-comment'>--                                             [ 0, -2, -1, 0]]</span>
+<a name="line-17771"></a><span class='hs-comment'>-- ```</span>
+<a name="line-17772"></a><span class='hs-comment'>-- </span>
+<a name="line-17773"></a><span class='hs-comment'>-- Useful special cases:</span>
+<a name="line-17774"></a><span class='hs-comment'>-- </span>
+<a name="line-17775"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-17776"></a><span class='hs-comment'>--  tf.matrix_band_part(input, 0, -1) ==&gt; Upper triangular part.</span>
+<a name="line-17777"></a><span class='hs-comment'>--  tf.matrix_band_part(input, -1, 0) ==&gt; Lower triangular part.</span>
+<a name="line-17778"></a><span class='hs-comment'>--  tf.matrix_band_part(input, 0, 0) ==&gt; Diagonal.</span>
+<a name="line-17779"></a><span class='hs-comment'>-- ```</span>
+<a name="line-17780"></a><span class='hs-definition'>matrixBandPart</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-17781"></a>                  <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Rank `k` tensor.</span>
+<a name="line-17782"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_lower__: 0-D tensor. Number of subdiagonals to keep. If negative, keep entire</span>
+<a name="line-17783"></a>                                              <span class='hs-comment'>-- lower triangle.</span>
+<a name="line-17784"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_upper__: 0-D tensor. Number of superdiagonals to keep. If negative, keep</span>
+<a name="line-17785"></a>                                              <span class='hs-comment'>-- entire upper triangle.</span>
+<a name="line-17786"></a>                  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __band__: Rank `k` tensor of the same shape as input. The extracted banded tensor.</span>
+<a name="line-17787"></a><span class='hs-definition'>matrixBandPart</span> <span class='hs-varid'>input</span> <span class='hs-varid'>num_lower</span> <span class='hs-varid'>num_upper</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17788"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MatrixBandPart"</span>
+<a name="line-17789"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-17790"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>num_lower</span> <span class='hs-varid'>num_upper</span>
+<a name="line-17791"></a><span class='hs-comment'>{-
+<a name="line-17792"></a>attr { name: "T" type: "type" }
+<a name="line-17793"></a>input_arg {
+<a name="line-17794"></a>  description: "Rank `k` tensor." name: "input" type_attr: "T"
+<a name="line-17795"></a>}
+<a name="line-17796"></a>input_arg {
+<a name="line-17797"></a>  description: "0-D tensor. Number of subdiagonals to keep. If negative, keep entire\nlower triangle."
+<a name="line-17798"></a>  name: "num_lower"
+<a name="line-17799"></a>  type: DT_INT64
+<a name="line-17800"></a>}
+<a name="line-17801"></a>input_arg {
+<a name="line-17802"></a>  description: "0-D tensor. Number of superdiagonals to keep. If negative, keep\nentire upper triangle."
+<a name="line-17803"></a>  name: "num_upper"
+<a name="line-17804"></a>  type: DT_INT64
+<a name="line-17805"></a>}
+<a name="line-17806"></a>output_arg {
+<a name="line-17807"></a>  description: "Rank `k` tensor of the same shape as input. The extracted banded tensor."
+<a name="line-17808"></a>  name: "band"
+<a name="line-17809"></a>  type_attr: "T"
+<a name="line-17810"></a>}
+<a name="line-17811"></a>-}</span>
+<a name="line-17812"></a>
+<a name="line-17813"></a><span class='hs-comment'>-- | Reinterpret the bytes of a string as a vector of numbers.</span>
+<a name="line-17814"></a>
+<a name="line-17815"></a><a name="decodeRaw"></a><span class='hs-definition'>decodeRaw</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>out_type</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-17816"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-17817"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-17818"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-17819"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-17820"></a>                                                                <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-17821"></a>                                                                <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-17822"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __bytes__: All the elements must have the same length.</span>
+<a name="line-17823"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>out_type</span> <span class='hs-comment'>-- ^ __output__: A Tensor with one more dimension than the input `bytes`.  The</span>
+<a name="line-17824"></a>             <span class='hs-comment'>-- added dimension will have size equal to the length of the elements</span>
+<a name="line-17825"></a>             <span class='hs-comment'>-- of `bytes` divided by the number of bytes to represent `out_type`.</span>
+<a name="line-17826"></a><span class='hs-definition'>decodeRaw</span> <span class='hs-varid'>bytes</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17827"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DecodeRaw"</span>
+<a name="line-17828"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"out_type"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>out_type</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-17829"></a>        <span class='hs-varid'>bytes</span>
+<a name="line-17830"></a><span class='hs-comment'>{-
+<a name="line-17831"></a>attr {
+<a name="line-17832"></a>  allowed_values {
+<a name="line-17833"></a>    list {
+<a name="line-17834"></a>      type: DT_FLOAT
+<a name="line-17835"></a>      type: DT_DOUBLE
+<a name="line-17836"></a>      type: DT_INT32
+<a name="line-17837"></a>      type: DT_UINT8
+<a name="line-17838"></a>      type: DT_INT16
+<a name="line-17839"></a>      type: DT_INT8
+<a name="line-17840"></a>      type: DT_INT64
+<a name="line-17841"></a>    }
+<a name="line-17842"></a>  }
+<a name="line-17843"></a>  name: "out_type"
+<a name="line-17844"></a>  type: "type"
+<a name="line-17845"></a>}
+<a name="line-17846"></a>attr {
+<a name="line-17847"></a>  default_value { b: true }
+<a name="line-17848"></a>  description: "Whether the input `bytes` are in little-endian order.\nIgnored for `out_type` values that are stored in a single byte like\n`uint8`."
+<a name="line-17849"></a>  name: "little_endian"
+<a name="line-17850"></a>  type: "bool"
+<a name="line-17851"></a>}
+<a name="line-17852"></a>input_arg {
+<a name="line-17853"></a>  description: "All the elements must have the same length."
+<a name="line-17854"></a>  name: "bytes"
+<a name="line-17855"></a>  type: DT_STRING
+<a name="line-17856"></a>}
+<a name="line-17857"></a>output_arg {
+<a name="line-17858"></a>  description: "A Tensor with one more dimension than the input `bytes`.  The\nadded dimension will have size equal to the length of the elements\nof `bytes` divided by the number of bytes to represent `out_type`."
+<a name="line-17859"></a>  name: "output"
+<a name="line-17860"></a>  type_attr: "out_type"
+<a name="line-17861"></a>}
+<a name="line-17862"></a>-}</span>
+<a name="line-17863"></a>
+<a name="line-17864"></a><a name="decodeJSONExample"></a><span class='hs-comment'>-- | Convert JSON-encoded Example records to binary protocol buffer strings.</span>
+<a name="line-17865"></a><span class='hs-comment'>--</span>
+<a name="line-17866"></a><span class='hs-comment'>-- This op translates a tensor containing Example records, encoded using</span>
+<a name="line-17867"></a><span class='hs-comment'>-- the [standard JSON</span>
+<a name="line-17868"></a><span class='hs-comment'>-- mapping](https://developers.google.com/protocol-buffers/docs/proto3#json),</span>
+<a name="line-17869"></a><span class='hs-comment'>-- into a tensor containing the same records encoded as binary protocol</span>
+<a name="line-17870"></a><span class='hs-comment'>-- buffers. The resulting tensor can then be fed to any of the other</span>
+<a name="line-17871"></a><span class='hs-comment'>-- Example-parsing ops.</span>
+<a name="line-17872"></a><span class='hs-definition'>decodeJSONExample</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __json_examples__: Each string is a JSON object serialized according to the JSON</span>
+<a name="line-17873"></a>                                                          <span class='hs-comment'>-- mapping of the Example proto.</span>
+<a name="line-17874"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __binary_examples__: Each string is a binary Example protocol buffer corresponding</span>
+<a name="line-17875"></a>                     <span class='hs-comment'>-- to the respective element of `json_examples`.</span>
+<a name="line-17876"></a><span class='hs-definition'>decodeJSONExample</span> <span class='hs-varid'>json_examples</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17877"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DecodeJSONExample"</span><span class='hs-layout'>)</span>
+<a name="line-17878"></a>        <span class='hs-varid'>json_examples</span>
+<a name="line-17879"></a><span class='hs-comment'>{-
+<a name="line-17880"></a>input_arg {
+<a name="line-17881"></a>  description: "Each string is a JSON object serialized according to the JSON\nmapping of the Example proto."
+<a name="line-17882"></a>  name: "json_examples"
+<a name="line-17883"></a>  type: DT_STRING
+<a name="line-17884"></a>}
+<a name="line-17885"></a>output_arg {
+<a name="line-17886"></a>  description: "Each string is a binary Example protocol buffer corresponding\nto the respective element of `json_examples`."
+<a name="line-17887"></a>  name: "binary_examples"
+<a name="line-17888"></a>  type: DT_STRING
+<a name="line-17889"></a>}
+<a name="line-17890"></a>-}</span>
+<a name="line-17891"></a>
+<a name="line-17892"></a><a name="truncatedNormal"></a><span class='hs-comment'>-- | Outputs random values from a truncated normal distribution.</span>
+<a name="line-17893"></a><span class='hs-comment'>--</span>
+<a name="line-17894"></a><span class='hs-comment'>-- The generated values follow a normal distribution with mean 0 and standard</span>
+<a name="line-17895"></a><span class='hs-comment'>-- deviation 1, except that values whose magnitude is more than 2 standard</span>
+<a name="line-17896"></a><span class='hs-comment'>-- deviations from the mean are dropped and re-picked.</span>
+<a name="line-17897"></a><span class='hs-definition'>truncatedNormal</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-17898"></a>                                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-17899"></a>                                        <span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>,</span>
+<a name="line-17900"></a>                                        <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-17901"></a>                                                <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-17902"></a>                   <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __shape__: The shape of the output tensor.</span>
+<a name="line-17903"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __output__: A tensor of the specified shape filled with random truncated normal</span>
+<a name="line-17904"></a>                   <span class='hs-comment'>-- values.</span>
+<a name="line-17905"></a><span class='hs-definition'>truncatedNormal</span> <span class='hs-varid'>shape</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17906"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TruncatedNormal"</span>
+<a name="line-17907"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-17908"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-17909"></a>        <span class='hs-varid'>shape</span>
+<a name="line-17910"></a><span class='hs-comment'>{-
+<a name="line-17911"></a>attr {
+<a name="line-17912"></a>  default_value { i: 0 }
+<a name="line-17913"></a>  description: "If either `seed` or `seed2` are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-17914"></a>  name: "seed"
+<a name="line-17915"></a>  type: "int"
+<a name="line-17916"></a>}
+<a name="line-17917"></a>attr {
+<a name="line-17918"></a>  default_value { i: 0 }
+<a name="line-17919"></a>  description: "A second seed to avoid seed collision."
+<a name="line-17920"></a>  name: "seed2"
+<a name="line-17921"></a>  type: "int"
+<a name="line-17922"></a>}
+<a name="line-17923"></a>attr {
+<a name="line-17924"></a>  allowed_values {
+<a name="line-17925"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-17926"></a>  }
+<a name="line-17927"></a>  description: "The type of the output."
+<a name="line-17928"></a>  name: "dtype"
+<a name="line-17929"></a>  type: "type"
+<a name="line-17930"></a>}
+<a name="line-17931"></a>attr {
+<a name="line-17932"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-17933"></a>  name: "T"
+<a name="line-17934"></a>  type: "type"
+<a name="line-17935"></a>}
+<a name="line-17936"></a>input_arg {
+<a name="line-17937"></a>  description: "The shape of the output tensor."
+<a name="line-17938"></a>  name: "shape"
+<a name="line-17939"></a>  type_attr: "T"
+<a name="line-17940"></a>}
+<a name="line-17941"></a>output_arg {
+<a name="line-17942"></a>  description: "A tensor of the specified shape filled with random truncated normal\nvalues."
+<a name="line-17943"></a>  name: "output"
+<a name="line-17944"></a>  type_attr: "dtype"
+<a name="line-17945"></a>}
+<a name="line-17946"></a>-}</span>
+<a name="line-17947"></a>
+<a name="line-17948"></a><a name="randomShuffle"></a><span class='hs-comment'>-- | Randomly shuffles a tensor along its first dimension.</span>
+<a name="line-17949"></a><span class='hs-comment'>--</span>
+<a name="line-17950"></a><span class='hs-comment'>--   The tensor is shuffled along dimension 0, such that each `value[j]` is mapped</span>
+<a name="line-17951"></a><span class='hs-comment'>--   to one and only one `output[i]`. For example, a mapping that might occur for a</span>
+<a name="line-17952"></a><span class='hs-comment'>--   3x2 tensor is:</span>
+<a name="line-17953"></a><span class='hs-comment'>-- </span>
+<a name="line-17954"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-17955"></a><span class='hs-comment'>-- [[1, 2],       [[5, 6],</span>
+<a name="line-17956"></a><span class='hs-comment'>--  [3, 4],  ==&gt;   [1, 2],</span>
+<a name="line-17957"></a><span class='hs-comment'>--  [5, 6]]        [3, 4]]</span>
+<a name="line-17958"></a><span class='hs-comment'>-- ```</span>
+<a name="line-17959"></a><span class='hs-definition'>randomShuffle</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-17960"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: The tensor to be shuffled.</span>
+<a name="line-17961"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: A tensor of same shape and type as `value`, shuffled along its first</span>
+<a name="line-17962"></a>                 <span class='hs-comment'>-- dimension.</span>
+<a name="line-17963"></a><span class='hs-definition'>randomShuffle</span> <span class='hs-varid'>value</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-17964"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RandomShuffle"</span>
+<a name="line-17965"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-17966"></a>        <span class='hs-varid'>value</span>
+<a name="line-17967"></a><span class='hs-comment'>{-
+<a name="line-17968"></a>attr {
+<a name="line-17969"></a>  default_value { i: 0 }
+<a name="line-17970"></a>  description: "If either `seed` or `seed2` are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-17971"></a>  name: "seed"
+<a name="line-17972"></a>  type: "int"
+<a name="line-17973"></a>}
+<a name="line-17974"></a>attr {
+<a name="line-17975"></a>  default_value { i: 0 }
+<a name="line-17976"></a>  description: "A second seed to avoid seed collision."
+<a name="line-17977"></a>  name: "seed2"
+<a name="line-17978"></a>  type: "int"
+<a name="line-17979"></a>}
+<a name="line-17980"></a>attr { name: "T" type: "type" }
+<a name="line-17981"></a>input_arg {
+<a name="line-17982"></a>  description: "The tensor to be shuffled."
+<a name="line-17983"></a>  name: "value"
+<a name="line-17984"></a>  type_attr: "T"
+<a name="line-17985"></a>}
+<a name="line-17986"></a>output_arg {
+<a name="line-17987"></a>  description: "A tensor of same shape and type as `value`, shuffled along its first\ndimension."
+<a name="line-17988"></a>  name: "output"
+<a name="line-17989"></a>  type_attr: "T"
+<a name="line-17990"></a>}
+<a name="line-17991"></a>-}</span>
+<a name="line-17992"></a>
+<a name="line-17993"></a><span class='hs-comment'>-- | Draws samples from a multinomial distribution.</span>
+<a name="line-17994"></a>
+<a name="line-17995"></a><a name="multinomial"></a><span class='hs-definition'>multinomial</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-17996"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-17997"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-17998"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-17999"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-18000"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-18001"></a>                                                       <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18002"></a>               <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __logits__: 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`</span>
+<a name="line-18003"></a>                           <span class='hs-comment'>-- represents the unnormalized log probabilities for all classes.</span>
+<a name="line-18004"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __num_samples__: 0-D.  Number of independent samples to draw for each row slice.</span>
+<a name="line-18005"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __output__: 2-D Tensor with shape `[batch_size, num_samples]`.  Each slice `[i, :]`</span>
+<a name="line-18006"></a>               <span class='hs-comment'>-- contains the drawn class labels with range `[0, num_classes)`.</span>
+<a name="line-18007"></a><span class='hs-definition'>multinomial</span> <span class='hs-varid'>logits</span> <span class='hs-varid'>num_samples</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18008"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Multinomial"</span>
+<a name="line-18009"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18010"></a>        <span class='hs-varid'>logits</span> <span class='hs-varid'>num_samples</span>
+<a name="line-18011"></a><span class='hs-comment'>{-
+<a name="line-18012"></a>attr {
+<a name="line-18013"></a>  default_value { i: 0 }
+<a name="line-18014"></a>  description: "If either seed or seed2 is set to be non-zero, the internal random number\ngenerator is seeded by the given seed.  Otherwise, a random seed is used."
+<a name="line-18015"></a>  name: "seed"
+<a name="line-18016"></a>  type: "int"
+<a name="line-18017"></a>}
+<a name="line-18018"></a>attr {
+<a name="line-18019"></a>  default_value { i: 0 }
+<a name="line-18020"></a>  description: "A second seed to avoid seed collision."
+<a name="line-18021"></a>  name: "seed2"
+<a name="line-18022"></a>  type: "int"
+<a name="line-18023"></a>}
+<a name="line-18024"></a>attr {
+<a name="line-18025"></a>  allowed_values {
+<a name="line-18026"></a>    list {
+<a name="line-18027"></a>      type: DT_FLOAT
+<a name="line-18028"></a>      type: DT_DOUBLE
+<a name="line-18029"></a>      type: DT_INT32
+<a name="line-18030"></a>      type: DT_INT64
+<a name="line-18031"></a>      type: DT_UINT8
+<a name="line-18032"></a>      type: DT_INT16
+<a name="line-18033"></a>      type: DT_INT8
+<a name="line-18034"></a>      type: DT_UINT16
+<a name="line-18035"></a>      type: DT_HALF
+<a name="line-18036"></a>    }
+<a name="line-18037"></a>  }
+<a name="line-18038"></a>  name: "T"
+<a name="line-18039"></a>  type: "type"
+<a name="line-18040"></a>}
+<a name="line-18041"></a>input_arg {
+<a name="line-18042"></a>  description: "2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`\nrepresents the unnormalized log probabilities for all classes."
+<a name="line-18043"></a>  name: "logits"
+<a name="line-18044"></a>  type_attr: "T"
+<a name="line-18045"></a>}
+<a name="line-18046"></a>input_arg {
+<a name="line-18047"></a>  description: "0-D.  Number of independent samples to draw for each row slice."
+<a name="line-18048"></a>  name: "num_samples"
+<a name="line-18049"></a>  type: DT_INT32
+<a name="line-18050"></a>}
+<a name="line-18051"></a>output_arg {
+<a name="line-18052"></a>  description: "2-D Tensor with shape `[batch_size, num_samples]`.  Each slice `[i, :]`\ncontains the drawn class labels with range `[0, num_classes)`."
+<a name="line-18053"></a>  name: "output"
+<a name="line-18054"></a>  type: DT_INT64
+<a name="line-18055"></a>}
+<a name="line-18056"></a>-}</span>
+<a name="line-18057"></a>
+<a name="line-18058"></a><a name="randomGamma"></a><span class='hs-comment'>-- | Outputs random values from the Gamma distribution(s) described by alpha.</span>
+<a name="line-18059"></a><span class='hs-comment'>--</span>
+<a name="line-18060"></a><span class='hs-comment'>-- This op uses the algorithm by Marsaglia et al. to acquire samples via</span>
+<a name="line-18061"></a><span class='hs-comment'>-- transformation-rejection from pairs of uniform and normal random variables.</span>
+<a name="line-18062"></a><span class='hs-comment'>-- See <a href="http://dl.acm.org/citation.cfm?id=358414">http://dl.acm.org/citation.cfm?id=358414</a></span>
+<a name="line-18063"></a><span class='hs-definition'>randomGamma</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>s</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>s</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-18064"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>s</span><span class='hs-layout'>,</span>
+<a name="line-18065"></a>                                   <span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-18066"></a>                                                         <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18067"></a>               <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>s</span> <span class='hs-comment'>-- ^ __shape__: 1-D integer tensor. Shape of independent samples to draw from each</span>
+<a name="line-18068"></a>                           <span class='hs-comment'>-- distribution described by the shape parameters given in alpha.</span>
+<a name="line-18069"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __alpha__: A tensor in which each scalar is a "shape" parameter describing the</span>
+<a name="line-18070"></a>                              <span class='hs-comment'>-- associated gamma distribution.</span>
+<a name="line-18071"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: A tensor with shape `shape + shape(alpha)`. Each slice</span>
+<a name="line-18072"></a>               <span class='hs-comment'>-- `[:, ..., :, i0, i1, ...iN]` contains the samples drawn for</span>
+<a name="line-18073"></a>               <span class='hs-comment'>-- `alpha[i0, i1, ...iN]`. The dtype of the output matches the dtype of alpha.</span>
+<a name="line-18074"></a><span class='hs-definition'>randomGamma</span> <span class='hs-varid'>shape</span> <span class='hs-varid'>alpha</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18075"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"RandomGamma"</span>
+<a name="line-18076"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"S"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>s</span><span class='hs-layout'>)</span>
+<a name="line-18077"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18078"></a>        <span class='hs-varid'>shape</span> <span class='hs-varid'>alpha</span>
+<a name="line-18079"></a><span class='hs-comment'>{-
+<a name="line-18080"></a>attr {
+<a name="line-18081"></a>  default_value { i: 0 }
+<a name="line-18082"></a>  description: "If either `seed` or `seed2` are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-18083"></a>  name: "seed"
+<a name="line-18084"></a>  type: "int"
+<a name="line-18085"></a>}
+<a name="line-18086"></a>attr {
+<a name="line-18087"></a>  default_value { i: 0 }
+<a name="line-18088"></a>  description: "A second seed to avoid seed collision."
+<a name="line-18089"></a>  name: "seed2"
+<a name="line-18090"></a>  type: "int"
+<a name="line-18091"></a>}
+<a name="line-18092"></a>attr {
+<a name="line-18093"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-18094"></a>  name: "S"
+<a name="line-18095"></a>  type: "type"
+<a name="line-18096"></a>}
+<a name="line-18097"></a>attr {
+<a name="line-18098"></a>  allowed_values {
+<a name="line-18099"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-18100"></a>  }
+<a name="line-18101"></a>  name: "T"
+<a name="line-18102"></a>  type: "type"
+<a name="line-18103"></a>}
+<a name="line-18104"></a>input_arg {
+<a name="line-18105"></a>  description: "1-D integer tensor. Shape of independent samples to draw from each\ndistribution described by the shape parameters given in alpha."
+<a name="line-18106"></a>  name: "shape"
+<a name="line-18107"></a>  type_attr: "S"
+<a name="line-18108"></a>}
+<a name="line-18109"></a>input_arg {
+<a name="line-18110"></a>  description: "A tensor in which each scalar is a \"shape\" parameter describing the\nassociated gamma distribution."
+<a name="line-18111"></a>  name: "alpha"
+<a name="line-18112"></a>  type_attr: "T"
+<a name="line-18113"></a>}
+<a name="line-18114"></a>output_arg {
+<a name="line-18115"></a>  description: "A tensor with shape `shape + shape(alpha)`. Each slice\n`[:, ..., :, i0, i1, ...iN]` contains the samples drawn for\n`alpha[i0, i1, ...iN]`. The dtype of the output matches the dtype of alpha."
+<a name="line-18116"></a>  name: "output"
+<a name="line-18117"></a>  type_attr: "T"
+<a name="line-18118"></a>}
+<a name="line-18119"></a>-}</span>
+<a name="line-18120"></a>
+<a name="line-18121"></a><span class='hs-comment'>-- | Add all input tensors element wise.</span>
+<a name="line-18122"></a>
+<a name="line-18123"></a><a name="addN"></a><span class='hs-definition'>addN</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18124"></a>                                             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18125"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-18126"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-18127"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-18128"></a>                                             <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18129"></a>        <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __inputs__: Must all be the same size and shape.</span>
+<a name="line-18130"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __sum__</span>
+<a name="line-18131"></a><span class='hs-definition'>addN</span> <span class='hs-varid'>inputs</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"N"</span><span class='hs-layout'>,</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"inputs"</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>inputs</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18132"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"AddN"</span>
+<a name="line-18133"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-18134"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"N"</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>inputs</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18135"></a>        <span class='hs-varid'>inputs</span>
+<a name="line-18136"></a><span class='hs-comment'>{-
+<a name="line-18137"></a>attr { has_minimum: true minimum: 1 name: "N" type: "int" }
+<a name="line-18138"></a>attr {
+<a name="line-18139"></a>  allowed_values {
+<a name="line-18140"></a>    list {
+<a name="line-18141"></a>      type: DT_FLOAT
+<a name="line-18142"></a>      type: DT_DOUBLE
+<a name="line-18143"></a>      type: DT_INT64
+<a name="line-18144"></a>      type: DT_INT32
+<a name="line-18145"></a>      type: DT_UINT8
+<a name="line-18146"></a>      type: DT_UINT16
+<a name="line-18147"></a>      type: DT_INT16
+<a name="line-18148"></a>      type: DT_INT8
+<a name="line-18149"></a>      type: DT_COMPLEX64
+<a name="line-18150"></a>      type: DT_COMPLEX128
+<a name="line-18151"></a>      type: DT_QINT8
+<a name="line-18152"></a>      type: DT_QUINT8
+<a name="line-18153"></a>      type: DT_QINT32
+<a name="line-18154"></a>      type: DT_HALF
+<a name="line-18155"></a>    }
+<a name="line-18156"></a>  }
+<a name="line-18157"></a>  name: "T"
+<a name="line-18158"></a>  type: "type"
+<a name="line-18159"></a>}
+<a name="line-18160"></a>input_arg {
+<a name="line-18161"></a>  description: "Must all be the same size and shape."
+<a name="line-18162"></a>  name: "inputs"
+<a name="line-18163"></a>  number_attr: "N"
+<a name="line-18164"></a>  type_attr: "T"
+<a name="line-18165"></a>}
+<a name="line-18166"></a>output_arg { name: "sum" type_attr: "T" }
+<a name="line-18167"></a>-}</span>
+<a name="line-18168"></a>
+<a name="line-18169"></a><a name="max"></a><span class='hs-comment'>-- | Computes the maximum of elements across dimensions of a tensor.</span>
+<a name="line-18170"></a><span class='hs-comment'>--</span>
+<a name="line-18171"></a><span class='hs-comment'>-- Reduces `input` along the dimensions given in `reduction_indices`. Unless</span>
+<a name="line-18172"></a><span class='hs-comment'>-- `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in</span>
+<a name="line-18173"></a><span class='hs-comment'>-- `reduction_indices`. If `keep_dims` is true, the reduced dimensions are</span>
+<a name="line-18174"></a><span class='hs-comment'>-- retained with length 1.</span>
+<a name="line-18175"></a><span class='hs-definition'>max</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-18176"></a>                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18177"></a>                                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18178"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-18179"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-18180"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-18181"></a>                                      <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span>
+<a name="line-18182"></a>                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18183"></a>       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: The tensor to reduce.</span>
+<a name="line-18184"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __reduction_indices__: The dimensions to reduce.</span>
+<a name="line-18185"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The reduced tensor.</span>
+<a name="line-18186"></a><span class='hs-definition'>max</span> <span class='hs-varid'>input</span> <span class='hs-varid'>reduction_indices</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18187"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Max"</span>
+<a name="line-18188"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-18189"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18190"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>reduction_indices</span>
+<a name="line-18191"></a><span class='hs-comment'>{-
+<a name="line-18192"></a>attr {
+<a name="line-18193"></a>  default_value { b: false }
+<a name="line-18194"></a>  description: "If true, retain reduced dimensions with length 1."
+<a name="line-18195"></a>  name: "keep_dims"
+<a name="line-18196"></a>  type: "bool"
+<a name="line-18197"></a>}
+<a name="line-18198"></a>attr {
+<a name="line-18199"></a>  allowed_values {
+<a name="line-18200"></a>    list {
+<a name="line-18201"></a>      type: DT_FLOAT
+<a name="line-18202"></a>      type: DT_DOUBLE
+<a name="line-18203"></a>      type: DT_INT64
+<a name="line-18204"></a>      type: DT_INT32
+<a name="line-18205"></a>      type: DT_UINT8
+<a name="line-18206"></a>      type: DT_UINT16
+<a name="line-18207"></a>      type: DT_INT16
+<a name="line-18208"></a>      type: DT_INT8
+<a name="line-18209"></a>      type: DT_COMPLEX64
+<a name="line-18210"></a>      type: DT_COMPLEX128
+<a name="line-18211"></a>      type: DT_QINT8
+<a name="line-18212"></a>      type: DT_QUINT8
+<a name="line-18213"></a>      type: DT_QINT32
+<a name="line-18214"></a>      type: DT_HALF
+<a name="line-18215"></a>    }
+<a name="line-18216"></a>  }
+<a name="line-18217"></a>  name: "T"
+<a name="line-18218"></a>  type: "type"
+<a name="line-18219"></a>}
+<a name="line-18220"></a>attr {
+<a name="line-18221"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-18222"></a>  default_value { type: DT_INT32 }
+<a name="line-18223"></a>  name: "Tidx"
+<a name="line-18224"></a>  type: "type"
+<a name="line-18225"></a>}
+<a name="line-18226"></a>input_arg {
+<a name="line-18227"></a>  description: "The tensor to reduce." name: "input" type_attr: "T"
+<a name="line-18228"></a>}
+<a name="line-18229"></a>input_arg {
+<a name="line-18230"></a>  description: "The dimensions to reduce."
+<a name="line-18231"></a>  name: "reduction_indices"
+<a name="line-18232"></a>  type_attr: "Tidx"
+<a name="line-18233"></a>}
+<a name="line-18234"></a>output_arg {
+<a name="line-18235"></a>  description: "The reduced tensor." name: "output" type_attr: "T"
+<a name="line-18236"></a>}
+<a name="line-18237"></a>-}</span>
+<a name="line-18238"></a>
+<a name="line-18239"></a><span class='hs-comment'>-- | A graph node which represents a return value of a function.</span>
+<a name="line-18240"></a>
+<a name="line-18241"></a><span class='hs-sel'>_Retval</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18242"></a>           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __index__: This return value is the index-th return value of the function.</span>
+<a name="line-18243"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: The return value.</span>
+<a name="line-18244"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-18245"></a><span class='hs-sel'>_Retval</span> <span class='hs-varid'>index</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18246"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"_Retval"</span>
+<a name="line-18247"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-18248"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"index"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>index</span><span class='hs-layout'>)</span>
+<a name="line-18249"></a>        <span class='hs-varid'>input</span>
+<a name="line-18250"></a><span class='hs-comment'>{-
+<a name="line-18251"></a>attr { name: "T" type: "type" }
+<a name="line-18252"></a>attr {
+<a name="line-18253"></a>  description: "This return value is the index-th return value of the function."
+<a name="line-18254"></a>  has_minimum: true
+<a name="line-18255"></a>  name: "index"
+<a name="line-18256"></a>  type: "int"
+<a name="line-18257"></a>}
+<a name="line-18258"></a>input_arg {
+<a name="line-18259"></a>  description: "The return value." name: "input" type_attr: "T"
+<a name="line-18260"></a>}
+<a name="line-18261"></a>-}</span>
+<a name="line-18262"></a>
+<a name="line-18263"></a><a name="destroyTemporaryVariable"></a><span class='hs-comment'>-- | Destroys the temporary variable and returns its final value.</span>
+<a name="line-18264"></a><span class='hs-comment'>--</span>
+<a name="line-18265"></a><span class='hs-comment'>-- Sets output to the value of the Tensor pointed to by 'ref', then destroys</span>
+<a name="line-18266"></a><span class='hs-comment'>-- the temporary variable called 'var_name'.</span>
+<a name="line-18267"></a><span class='hs-comment'>-- All other uses of 'ref' *must* have executed before this op.</span>
+<a name="line-18268"></a><span class='hs-comment'>-- This is typically achieved by chaining the ref through each assign op, or by</span>
+<a name="line-18269"></a><span class='hs-comment'>-- using control dependencies.</span>
+<a name="line-18270"></a><span class='hs-comment'>-- </span>
+<a name="line-18271"></a><span class='hs-comment'>-- Outputs the final value of the tensor pointed to by 'ref'.</span>
+<a name="line-18272"></a><span class='hs-definition'>destroyTemporaryVariable</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18273"></a>                            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __ref__: A reference to the temporary variable tensor.</span>
+<a name="line-18274"></a>                            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__</span>
+<a name="line-18275"></a><span class='hs-definition'>destroyTemporaryVariable</span> <span class='hs-varid'>ref</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18276"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DestroyTemporaryVariable"</span>
+<a name="line-18277"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18278"></a>        <span class='hs-varid'>ref</span>
+<a name="line-18279"></a><span class='hs-comment'>{-
+<a name="line-18280"></a>attr { name: "T" type: "type" }
+<a name="line-18281"></a>attr {
+<a name="line-18282"></a>  description: "Name of the temporary variable, usually the name of the matching\n\'TemporaryVariable\' op."
+<a name="line-18283"></a>  name: "var_name"
+<a name="line-18284"></a>  type: "string"
+<a name="line-18285"></a>}
+<a name="line-18286"></a>input_arg {
+<a name="line-18287"></a>  description: "A reference to the temporary variable tensor."
+<a name="line-18288"></a>  is_ref: true
+<a name="line-18289"></a>  name: "ref"
+<a name="line-18290"></a>  type_attr: "T"
+<a name="line-18291"></a>}
+<a name="line-18292"></a>output_arg { name: "value" type_attr: "T" }
+<a name="line-18293"></a>-}</span>
+<a name="line-18294"></a>
+<a name="line-18295"></a><span class='hs-comment'>-- | Cast x of type SrcT to y of DstT.</span>
+<a name="line-18296"></a>
+<a name="line-18297"></a><a name="cast"></a><span class='hs-definition'>cast</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>dstT</span> <span class='hs-varid'>srcT</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dstT</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>srcT</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18298"></a>        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>srcT</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-18299"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dstT</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-18300"></a><span class='hs-definition'>cast</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18301"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Cast"</span>
+<a name="line-18302"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"DstT"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dstT</span><span class='hs-layout'>)</span>
+<a name="line-18303"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"SrcT"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>srcT</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18304"></a>        <span class='hs-varid'>x</span>
+<a name="line-18305"></a><span class='hs-comment'>{-
+<a name="line-18306"></a>attr { name: "SrcT" type: "type" }
+<a name="line-18307"></a>attr { name: "DstT" type: "type" }
+<a name="line-18308"></a>input_arg { name: "x" type_attr: "SrcT" }
+<a name="line-18309"></a>output_arg { name: "y" type_attr: "DstT" }
+<a name="line-18310"></a>-}</span>
+<a name="line-18311"></a>
+<a name="line-18312"></a><a name="countUpTo"></a><span class='hs-comment'>-- | Increments 'ref' until it reaches 'limit'.</span>
+<a name="line-18313"></a><span class='hs-comment'>--</span>
+<a name="line-18314"></a><span class='hs-comment'>-- This operation outputs "ref" after the update is done.  This makes it</span>
+<a name="line-18315"></a><span class='hs-comment'>-- easier to chain operations that need to use the updated value.</span>
+<a name="line-18316"></a><span class='hs-definition'>countUpTo</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-18317"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18318"></a>             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __limit__: If incrementing ref would bring it above limit, instead generates an</span>
+<a name="line-18319"></a>                            <span class='hs-comment'>-- 'OutOfRange' error.</span>
+<a name="line-18320"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __ref__: Should be from a scalar `Variable` node.</span>
+<a name="line-18321"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: A copy of the input before increment. If nothing else modifies the</span>
+<a name="line-18322"></a>             <span class='hs-comment'>-- input, the values produced will all be distinct.</span>
+<a name="line-18323"></a><span class='hs-definition'>countUpTo</span> <span class='hs-varid'>limit</span> <span class='hs-varid'>ref</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18324"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"CountUpTo"</span>
+<a name="line-18325"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-18326"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"limit"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>limit</span><span class='hs-layout'>)</span>
+<a name="line-18327"></a>        <span class='hs-varid'>ref</span>
+<a name="line-18328"></a><span class='hs-comment'>{-
+<a name="line-18329"></a>attr {
+<a name="line-18330"></a>  description: "If incrementing ref would bring it above limit, instead generates an\n\'OutOfRange\' error."
+<a name="line-18331"></a>  name: "limit"
+<a name="line-18332"></a>  type: "int"
+<a name="line-18333"></a>}
+<a name="line-18334"></a>attr {
+<a name="line-18335"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-18336"></a>  name: "T"
+<a name="line-18337"></a>  type: "type"
+<a name="line-18338"></a>}
+<a name="line-18339"></a>input_arg {
+<a name="line-18340"></a>  description: "Should be from a scalar `Variable` node."
+<a name="line-18341"></a>  is_ref: true
+<a name="line-18342"></a>  name: "ref"
+<a name="line-18343"></a>  type_attr: "T"
+<a name="line-18344"></a>}
+<a name="line-18345"></a>output_arg {
+<a name="line-18346"></a>  description: "A copy of the input before increment. If nothing else modifies the\ninput, the values produced will all be distinct."
+<a name="line-18347"></a>  name: "output"
+<a name="line-18348"></a>  type_attr: "T"
+<a name="line-18349"></a>}
+<a name="line-18350"></a>-}</span>
+<a name="line-18351"></a>
+<a name="line-18352"></a><a name="abs"></a><span class='hs-comment'>-- | Computes the absolute value of a tensor.</span>
+<a name="line-18353"></a><span class='hs-comment'>--</span>
+<a name="line-18354"></a><span class='hs-comment'>-- Given a tensor `x`, this operation returns a tensor containing the absolute</span>
+<a name="line-18355"></a><span class='hs-comment'>-- value of each element in `x`. For example, if x is an input element and y is</span>
+<a name="line-18356"></a><span class='hs-comment'>-- an output element, this operation computes \\(y = |x|\\).</span>
+<a name="line-18357"></a><span class='hs-definition'>abs</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-18358"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-18359"></a>                                            <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-18360"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-18361"></a><span class='hs-definition'>abs</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18362"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Abs"</span>
+<a name="line-18363"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18364"></a>        <span class='hs-varid'>x</span>
+<a name="line-18365"></a><span class='hs-comment'>{-
+<a name="line-18366"></a>attr {
+<a name="line-18367"></a>  allowed_values {
+<a name="line-18368"></a>    list {
+<a name="line-18369"></a>      type: DT_HALF
+<a name="line-18370"></a>      type: DT_FLOAT
+<a name="line-18371"></a>      type: DT_DOUBLE
+<a name="line-18372"></a>      type: DT_INT32
+<a name="line-18373"></a>      type: DT_INT64
+<a name="line-18374"></a>    }
+<a name="line-18375"></a>  }
+<a name="line-18376"></a>  name: "T"
+<a name="line-18377"></a>  type: "type"
+<a name="line-18378"></a>}
+<a name="line-18379"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-18380"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-18381"></a>-}</span>
+<a name="line-18382"></a>
+<a name="line-18383"></a><a name="neg"></a><span class='hs-comment'>-- | Computes numerical negative value element-wise.</span>
+<a name="line-18384"></a><span class='hs-comment'>--</span>
+<a name="line-18385"></a><span class='hs-comment'>-- I.e., \\(y = -x\\).</span>
+<a name="line-18386"></a><span class='hs-definition'>neg</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18387"></a>                                            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18388"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-18389"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-18390"></a>                                            <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-18391"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-18392"></a><span class='hs-definition'>neg</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18393"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Neg"</span>
+<a name="line-18394"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18395"></a>        <span class='hs-varid'>x</span>
+<a name="line-18396"></a><span class='hs-comment'>{-
+<a name="line-18397"></a>attr {
+<a name="line-18398"></a>  allowed_values {
+<a name="line-18399"></a>    list {
+<a name="line-18400"></a>      type: DT_HALF
+<a name="line-18401"></a>      type: DT_FLOAT
+<a name="line-18402"></a>      type: DT_DOUBLE
+<a name="line-18403"></a>      type: DT_INT32
+<a name="line-18404"></a>      type: DT_INT64
+<a name="line-18405"></a>      type: DT_COMPLEX64
+<a name="line-18406"></a>      type: DT_COMPLEX128
+<a name="line-18407"></a>    }
+<a name="line-18408"></a>  }
+<a name="line-18409"></a>  name: "T"
+<a name="line-18410"></a>  type: "type"
+<a name="line-18411"></a>}
+<a name="line-18412"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-18413"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-18414"></a>-}</span>
+<a name="line-18415"></a>
+<a name="line-18416"></a><a name="sparseSparseMaximum"></a><span class='hs-comment'>-- | Returns the element-wise max of two SparseTensors.</span>
+<a name="line-18417"></a><span class='hs-comment'>--</span>
+<a name="line-18418"></a><span class='hs-comment'>-- Assumes the two SparseTensors have the same shape, i.e., no broadcasting.</span>
+<a name="line-18419"></a><span class='hs-definition'>sparseSparseMaximum</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-18420"></a>                                                     <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-18421"></a>                                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-18422"></a>                                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-18423"></a>                                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-18424"></a>                                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-18425"></a>                                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-18426"></a>                                                             <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-18427"></a>                                                             <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18428"></a>                       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __a_indices__: 2-D.  `N x R` matrix with the indices of non-empty values in a</span>
+<a name="line-18429"></a>                                                <span class='hs-comment'>-- SparseTensor, in the canonical lexicographic ordering.</span>
+<a name="line-18430"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __a_values__: 1-D.  `N` non-empty values corresponding to `a_indices`.</span>
+<a name="line-18431"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __a_shape__: 1-D.  Shape of the input SparseTensor.</span>
+<a name="line-18432"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __b_indices__: counterpart to `a_indices` for the other operand.</span>
+<a name="line-18433"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __b_values__: counterpart to `a_values` for the other operand; must be of the same dtype.</span>
+<a name="line-18434"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __b_shape__: counterpart to `a_shape` for the other operand; the two shapes must be equal.</span>
+<a name="line-18435"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-18436"></a>                       <span class='hs-comment'>-- ^ (__output_indices__, __output_values__)</span>
+<a name="line-18437"></a>                       <span class='hs-comment'>--</span>
+<a name="line-18438"></a>                       <span class='hs-comment'>-- * __output_indices__: 2-D.  The indices of the output SparseTensor.</span>
+<a name="line-18439"></a>                       <span class='hs-comment'>--</span>
+<a name="line-18440"></a>                       <span class='hs-comment'>-- * __output_values__: 1-D.  The values of the output SparseTensor.</span>
+<a name="line-18441"></a><span class='hs-definition'>sparseSparseMaximum</span> <span class='hs-varid'>a_indices</span> <span class='hs-varid'>a_values</span> <span class='hs-varid'>a_shape</span> <span class='hs-varid'>b_indices</span> <span class='hs-varid'>b_values</span>
+<a name="line-18442"></a>                    <span class='hs-varid'>b_shape</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18443"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseSparseMaximum"</span>
+<a name="line-18444"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18445"></a>        <span class='hs-varid'>a_indices</span> <span class='hs-varid'>a_values</span> <span class='hs-varid'>a_shape</span> <span class='hs-varid'>b_indices</span> <span class='hs-varid'>b_values</span> <span class='hs-varid'>b_shape</span>
+<a name="line-18446"></a><span class='hs-comment'>{-
+<a name="line-18447"></a>attr {
+<a name="line-18448"></a>  allowed_values {
+<a name="line-18449"></a>    list {
+<a name="line-18450"></a>      type: DT_FLOAT
+<a name="line-18451"></a>      type: DT_DOUBLE
+<a name="line-18452"></a>      type: DT_INT32
+<a name="line-18453"></a>      type: DT_INT64
+<a name="line-18454"></a>      type: DT_UINT8
+<a name="line-18455"></a>      type: DT_INT16
+<a name="line-18456"></a>      type: DT_INT8
+<a name="line-18457"></a>      type: DT_UINT16
+<a name="line-18458"></a>      type: DT_HALF
+<a name="line-18459"></a>    }
+<a name="line-18460"></a>  }
+<a name="line-18461"></a>  name: "T"
+<a name="line-18462"></a>  type: "type"
+<a name="line-18463"></a>}
+<a name="line-18464"></a>input_arg {
+<a name="line-18465"></a>  description: "2-D.  `N x R` matrix with the indices of non-empty values in a\nSparseTensor, in the canonical lexicographic ordering."
+<a name="line-18466"></a>  name: "a_indices"
+<a name="line-18467"></a>  type: DT_INT64
+<a name="line-18468"></a>}
+<a name="line-18469"></a>input_arg {
+<a name="line-18470"></a>  description: "1-D.  `N` non-empty values corresponding to `a_indices`."
+<a name="line-18471"></a>  name: "a_values"
+<a name="line-18472"></a>  type_attr: "T"
+<a name="line-18473"></a>}
+<a name="line-18474"></a>input_arg {
+<a name="line-18475"></a>  description: "1-D.  Shape of the input SparseTensor."
+<a name="line-18476"></a>  name: "a_shape"
+<a name="line-18477"></a>  type: DT_INT64
+<a name="line-18478"></a>}
+<a name="line-18479"></a>input_arg {
+<a name="line-18480"></a>  description: "counterpart to `a_indices` for the other operand."
+<a name="line-18481"></a>  name: "b_indices"
+<a name="line-18482"></a>  type: DT_INT64
+<a name="line-18483"></a>}
+<a name="line-18484"></a>input_arg {
+<a name="line-18485"></a>  description: "counterpart to `a_values` for the other operand; must be of the same dtype."
+<a name="line-18486"></a>  name: "b_values"
+<a name="line-18487"></a>  type_attr: "T"
+<a name="line-18488"></a>}
+<a name="line-18489"></a>input_arg {
+<a name="line-18490"></a>  description: "counterpart to `a_shape` for the other operand; the two shapes must be equal."
+<a name="line-18491"></a>  name: "b_shape"
+<a name="line-18492"></a>  type: DT_INT64
+<a name="line-18493"></a>}
+<a name="line-18494"></a>output_arg {
+<a name="line-18495"></a>  description: "2-D.  The indices of the output SparseTensor."
+<a name="line-18496"></a>  name: "output_indices"
+<a name="line-18497"></a>  type: DT_INT64
+<a name="line-18498"></a>}
+<a name="line-18499"></a>output_arg {
+<a name="line-18500"></a>  description: "1-D.  The values of the output SparseTensor."
+<a name="line-18501"></a>  name: "output_values"
+<a name="line-18502"></a>  type_attr: "T"
+<a name="line-18503"></a>}
+<a name="line-18504"></a>-}</span>
+<a name="line-18505"></a>
+<a name="line-18506"></a><a name="invGrad"></a><span class='hs-comment'>-- | Computes the gradient for the inverse of `x` wrt its input.</span>
+<a name="line-18507"></a><span class='hs-comment'>--</span>
+<a name="line-18508"></a><span class='hs-comment'>-- Specifically, `grad = -dy * y*y`, where `y = 1/x`, and `dy`</span>
+<a name="line-18509"></a><span class='hs-comment'>-- is the corresponding input gradient.</span>
+<a name="line-18510"></a><span class='hs-definition'>invGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-18511"></a>                             <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18512"></a>                                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18513"></a>                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18514"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-18515"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-18516"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-18517"></a><span class='hs-definition'>invGrad</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18518"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"InvGrad"</span>
+<a name="line-18519"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18520"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-18521"></a><span class='hs-comment'>{-
+<a name="line-18522"></a>attr {
+<a name="line-18523"></a>  allowed_values {
+<a name="line-18524"></a>    list {
+<a name="line-18525"></a>      type: DT_HALF
+<a name="line-18526"></a>      type: DT_FLOAT
+<a name="line-18527"></a>      type: DT_DOUBLE
+<a name="line-18528"></a>      type: DT_COMPLEX64
+<a name="line-18529"></a>      type: DT_COMPLEX128
+<a name="line-18530"></a>    }
+<a name="line-18531"></a>  }
+<a name="line-18532"></a>  name: "T"
+<a name="line-18533"></a>  type: "type"
+<a name="line-18534"></a>}
+<a name="line-18535"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-18536"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-18537"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-18538"></a>-}</span>
+<a name="line-18539"></a>
+<a name="line-18540"></a><a name="sqrt"></a><span class='hs-comment'>-- | Computes square root of x element-wise.</span>
+<a name="line-18541"></a><span class='hs-comment'>--</span>
+<a name="line-18542"></a><span class='hs-comment'>-- I.e., \\(y = \sqrt{x} = x^{1/2}\\).</span>
+<a name="line-18543"></a><span class='hs-definition'>sqrt</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18544"></a>                                             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18545"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-18546"></a>                                             <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-18547"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-18548"></a><span class='hs-definition'>sqrt</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18549"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Sqrt"</span>
+<a name="line-18550"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18551"></a>        <span class='hs-varid'>x</span>
+<a name="line-18552"></a><span class='hs-comment'>{-
+<a name="line-18553"></a>attr {
+<a name="line-18554"></a>  allowed_values {
+<a name="line-18555"></a>    list {
+<a name="line-18556"></a>      type: DT_HALF
+<a name="line-18557"></a>      type: DT_FLOAT
+<a name="line-18558"></a>      type: DT_DOUBLE
+<a name="line-18559"></a>      type: DT_COMPLEX64
+<a name="line-18560"></a>      type: DT_COMPLEX128
+<a name="line-18561"></a>    }
+<a name="line-18562"></a>  }
+<a name="line-18563"></a>  name: "T"
+<a name="line-18564"></a>  type: "type"
+<a name="line-18565"></a>}
+<a name="line-18566"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-18567"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-18568"></a>-}</span>
+<a name="line-18569"></a>
+<a name="line-18570"></a><a name="matrixInverse"></a><span class='hs-comment'>-- | Computes the inverse of one or more square invertible matrices or their</span>
+<a name="line-18571"></a><span class='hs-comment'>--</span>
+<a name="line-18572"></a><span class='hs-comment'>-- adjoints (conjugate transposes).</span>
+<a name="line-18573"></a><span class='hs-comment'>-- </span>
+<a name="line-18574"></a><span class='hs-comment'>-- The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions</span>
+<a name="line-18575"></a><span class='hs-comment'>-- form square matrices. The output is a tensor of the same shape as the input</span>
+<a name="line-18576"></a><span class='hs-comment'>-- containing the inverse for all input submatrices `[..., :, :]`.</span>
+<a name="line-18577"></a><span class='hs-comment'>-- </span>
+<a name="line-18578"></a><span class='hs-comment'>-- The op uses LU decomposition with partial pivoting to compute the inverses.</span>
+<a name="line-18579"></a><span class='hs-comment'>-- </span>
+<a name="line-18580"></a><span class='hs-comment'>-- If a matrix is not invertible there is no guarantee what the op does. It</span>
+<a name="line-18581"></a><span class='hs-comment'>-- may detect the condition and raise an exception or it may simply return a</span>
+<a name="line-18582"></a><span class='hs-comment'>-- garbage result.</span>
+<a name="line-18583"></a><span class='hs-definition'>matrixInverse</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18584"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Shape is `[..., M, M]`.</span>
+<a name="line-18585"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Shape is `[..., M, M]`.</span>
+<a name="line-18586"></a><span class='hs-definition'>matrixInverse</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18587"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MatrixInverse"</span>
+<a name="line-18588"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18589"></a>        <span class='hs-varid'>input</span>
+<a name="line-18590"></a><span class='hs-comment'>{-
+<a name="line-18591"></a>attr { default_value { b: false } name: "adjoint" type: "bool" }
+<a name="line-18592"></a>attr {
+<a name="line-18593"></a>  allowed_values { list { type: DT_DOUBLE type: DT_FLOAT } }
+<a name="line-18594"></a>  name: "T"
+<a name="line-18595"></a>  type: "type"
+<a name="line-18596"></a>}
+<a name="line-18597"></a>input_arg {
+<a name="line-18598"></a>  description: "Shape is `[..., M, M]`." name: "input" type_attr: "T"
+<a name="line-18599"></a>}
+<a name="line-18600"></a>output_arg {
+<a name="line-18601"></a>  description: "Shape is `[..., M, M]`."
+<a name="line-18602"></a>  name: "output"
+<a name="line-18603"></a>  type_attr: "T"
+<a name="line-18604"></a>}
+<a name="line-18605"></a>-}</span>
+<a name="line-18606"></a>
+<a name="line-18607"></a><a name="sqrtGrad"></a><span class='hs-comment'>-- | Computes the gradient for the sqrt of `x` wrt its input.</span>
+<a name="line-18608"></a><span class='hs-comment'>--</span>
+<a name="line-18609"></a><span class='hs-comment'>-- Specifically, `grad = dy * 0.5 / y`, where `y = sqrt(x)`, and `dy`</span>
+<a name="line-18610"></a><span class='hs-comment'>-- is the corresponding input gradient.</span>
+<a name="line-18611"></a><span class='hs-definition'>sqrtGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-18612"></a>                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18613"></a>                                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18614"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18615"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-18616"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-18617"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-18618"></a><span class='hs-definition'>sqrtGrad</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18619"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SqrtGrad"</span>
+<a name="line-18620"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18621"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-18622"></a><span class='hs-comment'>{-
+<a name="line-18623"></a>attr {
+<a name="line-18624"></a>  allowed_values {
+<a name="line-18625"></a>    list {
+<a name="line-18626"></a>      type: DT_HALF
+<a name="line-18627"></a>      type: DT_FLOAT
+<a name="line-18628"></a>      type: DT_DOUBLE
+<a name="line-18629"></a>      type: DT_COMPLEX64
+<a name="line-18630"></a>      type: DT_COMPLEX128
+<a name="line-18631"></a>    }
+<a name="line-18632"></a>  }
+<a name="line-18633"></a>  name: "T"
+<a name="line-18634"></a>  type: "type"
+<a name="line-18635"></a>}
+<a name="line-18636"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-18637"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-18638"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-18639"></a>-}</span>
+<a name="line-18640"></a>
+<a name="line-18641"></a><a name="expandDims"></a><span class='hs-comment'>-- | Inserts a dimension of 1 into a tensor's shape.</span>
+<a name="line-18642"></a><span class='hs-comment'>--</span>
+<a name="line-18643"></a><span class='hs-comment'>-- Given a tensor `input`, this operation inserts a dimension of 1 at the</span>
+<a name="line-18644"></a><span class='hs-comment'>-- dimension index `dim` of `input`'s shape. The dimension index `dim` starts at</span>
+<a name="line-18645"></a><span class='hs-comment'>-- zero; if you specify a negative number for `dim` it is counted backward from</span>
+<a name="line-18646"></a><span class='hs-comment'>-- the end.</span>
+<a name="line-18647"></a><span class='hs-comment'>-- </span>
+<a name="line-18648"></a><span class='hs-comment'>-- This operation is useful if you want to add a batch dimension to a single</span>
+<a name="line-18649"></a><span class='hs-comment'>-- element. For example, if you have a single image of shape `[height, width,</span>
+<a name="line-18650"></a><span class='hs-comment'>-- channels]`, you can make it a batch of 1 image with `expand_dims(image, 0)`,</span>
+<a name="line-18651"></a><span class='hs-comment'>-- which will make the shape `[1, height, width, channels]`.</span>
+<a name="line-18652"></a><span class='hs-comment'>-- </span>
+<a name="line-18653"></a><span class='hs-comment'>-- Other examples:</span>
+<a name="line-18654"></a><span class='hs-comment'>-- </span>
+<a name="line-18655"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-18656"></a><span class='hs-comment'>-- # 't' is a tensor of shape [2]</span>
+<a name="line-18657"></a><span class='hs-comment'>-- shape(expand_dims(t, 0)) ==&gt; [1, 2]</span>
+<a name="line-18658"></a><span class='hs-comment'>-- shape(expand_dims(t, 1)) ==&gt; [2, 1]</span>
+<a name="line-18659"></a><span class='hs-comment'>-- shape(expand_dims(t, -1)) ==&gt; [2, 1]</span>
+<a name="line-18660"></a><span class='hs-comment'>-- </span>
+<a name="line-18661"></a><span class='hs-comment'>-- # 't2' is a tensor of shape [2, 3, 5]</span>
+<a name="line-18662"></a><span class='hs-comment'>-- shape(expand_dims(t2, 0)) ==&gt; [1, 2, 3, 5]</span>
+<a name="line-18663"></a><span class='hs-comment'>-- shape(expand_dims(t2, 2)) ==&gt; [2, 3, 1, 5]</span>
+<a name="line-18664"></a><span class='hs-comment'>-- shape(expand_dims(t2, 3)) ==&gt; [2, 3, 5, 1]</span>
+<a name="line-18665"></a><span class='hs-comment'>-- ```</span>
+<a name="line-18666"></a><span class='hs-comment'>-- </span>
+<a name="line-18667"></a><span class='hs-comment'>-- This operation requires that:</span>
+<a name="line-18668"></a><span class='hs-comment'>-- </span>
+<a name="line-18669"></a><span class='hs-comment'>-- `-1-input.dims() &lt;= dim &lt;= input.dims()`</span>
+<a name="line-18670"></a><span class='hs-comment'>-- </span>
+<a name="line-18671"></a><span class='hs-comment'>-- This operation is related to `squeeze()`, which removes dimensions of</span>
+<a name="line-18672"></a><span class='hs-comment'>-- size 1.</span>
+<a name="line-18673"></a><span class='hs-definition'>expandDims</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tdim</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tdim</span><span class='hs-layout'>,</span>
+<a name="line-18674"></a>                                     <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-18675"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tdim</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18676"></a>              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-18677"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tdim</span> <span class='hs-comment'>-- ^ __dim__: 0-D (scalar). Specifies the dimension index at which to</span>
+<a name="line-18678"></a>                                <span class='hs-comment'>-- expand the shape of `input`.</span>
+<a name="line-18679"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Contains the same data as `input`, but its shape has an additional</span>
+<a name="line-18680"></a>              <span class='hs-comment'>-- dimension of size 1 added.</span>
+<a name="line-18681"></a><span class='hs-definition'>expandDims</span> <span class='hs-varid'>input</span> <span class='hs-varid'>dim</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18682"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ExpandDims"</span>
+<a name="line-18683"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-18684"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tdim"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tdim</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18685"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>dim</span>
+<a name="line-18686"></a><span class='hs-comment'>{-
+<a name="line-18687"></a>attr { name: "T" type: "type" }
+<a name="line-18688"></a>attr {
+<a name="line-18689"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-18690"></a>  default_value { type: DT_INT32 }
+<a name="line-18691"></a>  name: "Tdim"
+<a name="line-18692"></a>  type: "type"
+<a name="line-18693"></a>}
+<a name="line-18694"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-18695"></a>input_arg {
+<a name="line-18696"></a>  description: "0-D (scalar). Specifies the dimension index at which to\nexpand the shape of `input`."
+<a name="line-18697"></a>  name: "dim"
+<a name="line-18698"></a>  type_attr: "Tdim"
+<a name="line-18699"></a>}
+<a name="line-18700"></a>output_arg {
+<a name="line-18701"></a>  description: "Contains the same data as `input`, but its shape has an additional\ndimension of size 1 added."
+<a name="line-18702"></a>  name: "output"
+<a name="line-18703"></a>  type_attr: "T"
+<a name="line-18704"></a>}
+<a name="line-18705"></a>-}</span>
+<a name="line-18706"></a>
+<a name="line-18707"></a><a name="all"></a><span class='hs-comment'>-- | Computes the "logical and" of elements across dimensions of a tensor.</span>
+<a name="line-18708"></a><span class='hs-comment'>--</span>
+<a name="line-18709"></a><span class='hs-comment'>-- Reduces `input` along the dimensions given in `reduction_indices`. Unless</span>
+<a name="line-18710"></a><span class='hs-comment'>-- `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in</span>
+<a name="line-18711"></a><span class='hs-comment'>-- `reduction_indices`. If `keep_dims` is true, the reduced dimensions are</span>
+<a name="line-18712"></a><span class='hs-comment'>-- retained with length 1.</span>
+<a name="line-18713"></a><span class='hs-definition'>all</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-18714"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18715"></a>       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __input__: The tensor to reduce.</span>
+<a name="line-18716"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __reduction_indices__: The dimensions to reduce.</span>
+<a name="line-18717"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __output__: The reduced tensor.</span>
+<a name="line-18718"></a><span class='hs-definition'>all</span> <span class='hs-varid'>input</span> <span class='hs-varid'>reduction_indices</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18719"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"All"</span>
+<a name="line-18720"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18721"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>reduction_indices</span>
+<a name="line-18722"></a><span class='hs-comment'>{-
+<a name="line-18723"></a>attr {
+<a name="line-18724"></a>  default_value { b: false }
+<a name="line-18725"></a>  description: "If true, retain reduced dimensions with length 1."
+<a name="line-18726"></a>  name: "keep_dims"
+<a name="line-18727"></a>  type: "bool"
+<a name="line-18728"></a>}
+<a name="line-18729"></a>attr {
+<a name="line-18730"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-18731"></a>  default_value { type: DT_INT32 }
+<a name="line-18732"></a>  name: "Tidx"
+<a name="line-18733"></a>  type: "type"
+<a name="line-18734"></a>}
+<a name="line-18735"></a>input_arg {
+<a name="line-18736"></a>  description: "The tensor to reduce." name: "input" type: DT_BOOL
+<a name="line-18737"></a>}
+<a name="line-18738"></a>input_arg {
+<a name="line-18739"></a>  description: "The dimensions to reduce."
+<a name="line-18740"></a>  name: "reduction_indices"
+<a name="line-18741"></a>  type_attr: "Tidx"
+<a name="line-18742"></a>}
+<a name="line-18743"></a>output_arg {
+<a name="line-18744"></a>  description: "The reduced tensor." name: "output" type: DT_BOOL
+<a name="line-18745"></a>}
+<a name="line-18746"></a>-}</span>
+<a name="line-18747"></a>
+<a name="line-18748"></a><a name="cTCBeamSearchDecoder"></a><span class='hs-comment'>-- | Performs beam search decoding on the logits given in input.</span>
+<a name="line-18749"></a><span class='hs-comment'>--</span>
+<a name="line-18750"></a><span class='hs-comment'>-- A note about the attribute merge_repeated: For the beam search decoder,</span>
+<a name="line-18751"></a><span class='hs-comment'>-- this means that if consecutive entries in a beam are the same, only</span>
+<a name="line-18752"></a><span class='hs-comment'>-- the first of these is emitted.  That is, when the top path is "A B B B B",</span>
+<a name="line-18753"></a><span class='hs-comment'>-- "A B" is returned if merge_repeated = True but "A B B B B" is</span>
+<a name="line-18754"></a><span class='hs-comment'>-- returned if merge_repeated = False.</span>
+<a name="line-18755"></a><span class='hs-definition'>cTCBeamSearchDecoder</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __beam_width__: A scalar &gt;= 0 (beam search beam width).</span>
+<a name="line-18756"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __top_paths__: A scalar &gt;= 0, &lt;= beam_width (controls output size).</span>
+<a name="line-18757"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __inputs__: 3-D, shape: `(max_time x batch_size x num_classes)`, the logits.</span>
+<a name="line-18758"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __sequence_length__: A vector containing sequence lengths, size `(batch)`.</span>
+<a name="line-18759"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-18760"></a>                            <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-18761"></a>                            <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span>
+<a name="line-18762"></a>                        <span class='hs-comment'>-- ^ (__decoded_indices__, __decoded_values__, __decoded_shape__, __log_probability__)</span>
+<a name="line-18763"></a>                        <span class='hs-comment'>--</span>
+<a name="line-18764"></a>                        <span class='hs-comment'>-- * __decoded_indices__: A list (length: top_paths) of indices matrices.  Matrix j,</span>
+<a name="line-18765"></a>                        <span class='hs-comment'>-- size `(total_decoded_outputs[j] x 2)`, has indices of a</span>
+<a name="line-18766"></a>                        <span class='hs-comment'>-- `SparseTensor&lt;int64, 2&gt;`.  The rows store: [batch, time].</span>
+<a name="line-18767"></a>                        <span class='hs-comment'>--</span>
+<a name="line-18768"></a>                        <span class='hs-comment'>-- * __decoded_values__: A list (length: top_paths) of values vectors.  Vector j,</span>
+<a name="line-18769"></a>                        <span class='hs-comment'>-- size `(length total_decoded_outputs[j])`, has the values of a</span>
+<a name="line-18770"></a>                        <span class='hs-comment'>-- `SparseTensor&lt;int64, 2&gt;`.  The vector stores the decoded classes for beam j.</span>
+<a name="line-18771"></a>                        <span class='hs-comment'>--</span>
+<a name="line-18772"></a>                        <span class='hs-comment'>-- * __decoded_shape__: A list (length: top_paths) of shape vector.  Vector j,</span>
+<a name="line-18773"></a>                        <span class='hs-comment'>-- size `(2)`, stores the shape of the decoded `SparseTensor[j]`.</span>
+<a name="line-18774"></a>                        <span class='hs-comment'>-- Its values are: `[batch_size, max_decoded_length[j]]`.</span>
+<a name="line-18775"></a>                        <span class='hs-comment'>--</span>
+<a name="line-18776"></a>                        <span class='hs-comment'>-- * __log_probability__: A matrix, shaped: `(batch_size x top_paths)`.  The</span>
+<a name="line-18777"></a>                        <span class='hs-comment'>-- sequence log-probabilities.</span>
+<a name="line-18778"></a><span class='hs-definition'>cTCBeamSearchDecoder</span> <span class='hs-varid'>beam_width</span> <span class='hs-varid'>top_paths</span> <span class='hs-varid'>inputs</span>
+<a name="line-18779"></a>                     <span class='hs-varid'>sequence_length</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18780"></a>    <span class='hs-varid'>buildListOp</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>top_paths</span><span class='hs-layout'>,</span> <span class='hs-varid'>top_paths</span><span class='hs-layout'>,</span> <span class='hs-varid'>top_paths</span><span class='hs-keyglyph'>]</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"CTCBeamSearchDecoder"</span>
+<a name="line-18781"></a>                                                   <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"beam_width"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>beam_width</span>
+<a name="line-18782"></a>                                                   <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"top_paths"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>top_paths</span><span class='hs-layout'>)</span>
+<a name="line-18783"></a>        <span class='hs-varid'>inputs</span> <span class='hs-varid'>sequence_length</span>
+<a name="line-18784"></a><span class='hs-comment'>{-
+<a name="line-18785"></a>attr {
+<a name="line-18786"></a>  description: "A scalar &gt;= 0 (beam search beam width)."
+<a name="line-18787"></a>  has_minimum: true
+<a name="line-18788"></a>  minimum: 1
+<a name="line-18789"></a>  name: "beam_width"
+<a name="line-18790"></a>  type: "int"
+<a name="line-18791"></a>}
+<a name="line-18792"></a>attr {
+<a name="line-18793"></a>  description: "A scalar &gt;= 0, &lt;= beam_width (controls output size)."
+<a name="line-18794"></a>  has_minimum: true
+<a name="line-18795"></a>  minimum: 1
+<a name="line-18796"></a>  name: "top_paths"
+<a name="line-18797"></a>  type: "int"
+<a name="line-18798"></a>}
+<a name="line-18799"></a>attr {
+<a name="line-18800"></a>  default_value { b: true }
+<a name="line-18801"></a>  description: "If true, merge repeated classes in output."
+<a name="line-18802"></a>  name: "merge_repeated"
+<a name="line-18803"></a>  type: "bool"
+<a name="line-18804"></a>}
+<a name="line-18805"></a>input_arg {
+<a name="line-18806"></a>  description: "3-D, shape: `(max_time x batch_size x num_classes)`, the logits."
+<a name="line-18807"></a>  name: "inputs"
+<a name="line-18808"></a>  type: DT_FLOAT
+<a name="line-18809"></a>}
+<a name="line-18810"></a>input_arg {
+<a name="line-18811"></a>  description: "A vector containing sequence lengths, size `(batch)`."
+<a name="line-18812"></a>  name: "sequence_length"
+<a name="line-18813"></a>  type: DT_INT32
+<a name="line-18814"></a>}
+<a name="line-18815"></a>output_arg {
+<a name="line-18816"></a>  description: "A list (length: top_paths) of indices matrices.  Matrix j,\nsize `(total_decoded_outputs[j] x 2)`, has indices of a\n`SparseTensor&lt;int64, 2&gt;`.  The rows store: [batch, time]."
+<a name="line-18817"></a>  name: "decoded_indices"
+<a name="line-18818"></a>  number_attr: "top_paths"
+<a name="line-18819"></a>  type: DT_INT64
+<a name="line-18820"></a>}
+<a name="line-18821"></a>output_arg {
+<a name="line-18822"></a>  description: "A list (length: top_paths) of values vectors.  Vector j,\nsize `(length total_decoded_outputs[j])`, has the values of a\n`SparseTensor&lt;int64, 2&gt;`.  The vector stores the decoded classes for beam j."
+<a name="line-18823"></a>  name: "decoded_values"
+<a name="line-18824"></a>  number_attr: "top_paths"
+<a name="line-18825"></a>  type: DT_INT64
+<a name="line-18826"></a>}
+<a name="line-18827"></a>output_arg {
+<a name="line-18828"></a>  description: "A list (length: top_paths) of shape vector.  Vector j,\nsize `(2)`, stores the shape of the decoded `SparseTensor[j]`.\nIts values are: `[batch_size, max_decoded_length[j]]`."
+<a name="line-18829"></a>  name: "decoded_shape"
+<a name="line-18830"></a>  number_attr: "top_paths"
+<a name="line-18831"></a>  type: DT_INT64
+<a name="line-18832"></a>}
+<a name="line-18833"></a>output_arg {
+<a name="line-18834"></a>  description: "A matrix, shaped: `(batch_size x top_paths)`.  The\nsequence log-probabilities."
+<a name="line-18835"></a>  name: "log_probability"
+<a name="line-18836"></a>  type: DT_FLOAT
+<a name="line-18837"></a>}
+<a name="line-18838"></a>-}</span>
+<a name="line-18839"></a>
+<a name="line-18840"></a><a name="rsqrt"></a><span class='hs-comment'>-- | Computes reciprocal of square root of x element-wise.</span>
+<a name="line-18841"></a><span class='hs-comment'>--</span>
+<a name="line-18842"></a><span class='hs-comment'>-- I.e., \\(y = 1 / \sqrt{x}\\).</span>
+<a name="line-18843"></a><span class='hs-definition'>rsqrt</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18844"></a>                                              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18845"></a>                                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-18846"></a>                                              <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18847"></a>         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-18848"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-18849"></a><span class='hs-definition'>rsqrt</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18850"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Rsqrt"</span>
+<a name="line-18851"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18852"></a>        <span class='hs-varid'>x</span>
+<a name="line-18853"></a><span class='hs-comment'>{-
+<a name="line-18854"></a>attr {
+<a name="line-18855"></a>  allowed_values {
+<a name="line-18856"></a>    list {
+<a name="line-18857"></a>      type: DT_HALF
+<a name="line-18858"></a>      type: DT_FLOAT
+<a name="line-18859"></a>      type: DT_DOUBLE
+<a name="line-18860"></a>      type: DT_COMPLEX64
+<a name="line-18861"></a>      type: DT_COMPLEX128
+<a name="line-18862"></a>    }
+<a name="line-18863"></a>  }
+<a name="line-18864"></a>  name: "T"
+<a name="line-18865"></a>  type: "type"
+<a name="line-18866"></a>}
+<a name="line-18867"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-18868"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-18869"></a>-}</span>
+<a name="line-18870"></a>
+<a name="line-18871"></a><a name="tanhGrad"></a><span class='hs-comment'>-- | Computes the gradient for the tanh of `x` wrt its input.</span>
+<a name="line-18872"></a><span class='hs-comment'>--</span>
+<a name="line-18873"></a><span class='hs-comment'>-- Specifically, `grad = dy * (1 - y*y)`, where `y = tanh(x)`, and `dy`</span>
+<a name="line-18874"></a><span class='hs-comment'>-- is the corresponding input gradient.</span>
+<a name="line-18875"></a><span class='hs-definition'>tanhGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-18876"></a>                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18877"></a>                                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18878"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18879"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-18880"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-18881"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-18882"></a><span class='hs-definition'>tanhGrad</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18883"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TanhGrad"</span>
+<a name="line-18884"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18885"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-18886"></a><span class='hs-comment'>{-
+<a name="line-18887"></a>attr {
+<a name="line-18888"></a>  allowed_values {
+<a name="line-18889"></a>    list {
+<a name="line-18890"></a>      type: DT_HALF
+<a name="line-18891"></a>      type: DT_FLOAT
+<a name="line-18892"></a>      type: DT_DOUBLE
+<a name="line-18893"></a>      type: DT_COMPLEX64
+<a name="line-18894"></a>      type: DT_COMPLEX128
+<a name="line-18895"></a>    }
+<a name="line-18896"></a>  }
+<a name="line-18897"></a>  name: "T"
+<a name="line-18898"></a>  type: "type"
+<a name="line-18899"></a>}
+<a name="line-18900"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-18901"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-18902"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-18903"></a>-}</span>
+<a name="line-18904"></a>
+<a name="line-18905"></a><span class='hs-comment'>-- | Computes sin of x element-wise.</span>
+<a name="line-18906"></a>
+<a name="line-18907"></a><a name="sin"></a><span class='hs-definition'>sin</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18908"></a>                                            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18909"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-18910"></a>                                            <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-18911"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-18912"></a><span class='hs-definition'>sin</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18913"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Sin"</span>
+<a name="line-18914"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18915"></a>        <span class='hs-varid'>x</span>
+<a name="line-18916"></a><span class='hs-comment'>{-
+<a name="line-18917"></a>attr {
+<a name="line-18918"></a>  allowed_values {
+<a name="line-18919"></a>    list {
+<a name="line-18920"></a>      type: DT_HALF
+<a name="line-18921"></a>      type: DT_FLOAT
+<a name="line-18922"></a>      type: DT_DOUBLE
+<a name="line-18923"></a>      type: DT_COMPLEX64
+<a name="line-18924"></a>      type: DT_COMPLEX128
+<a name="line-18925"></a>    }
+<a name="line-18926"></a>  }
+<a name="line-18927"></a>  name: "T"
+<a name="line-18928"></a>  type: "type"
+<a name="line-18929"></a>}
+<a name="line-18930"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-18931"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-18932"></a>-}</span>
+<a name="line-18933"></a>
+<a name="line-18934"></a><a name="matrixDeterminant"></a><span class='hs-comment'>-- | Computes the determinant of one ore more square matrices.</span>
+<a name="line-18935"></a><span class='hs-comment'>--</span>
+<a name="line-18936"></a><span class='hs-comment'>-- The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions</span>
+<a name="line-18937"></a><span class='hs-comment'>-- form square matrices. The output is a tensor containing the determinants</span>
+<a name="line-18938"></a><span class='hs-comment'>-- for all input submatrices `[..., :, :]`.</span>
+<a name="line-18939"></a><span class='hs-definition'>matrixDeterminant</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-18940"></a>                     <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Shape is `[..., M, M]`.</span>
+<a name="line-18941"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Shape is `[...]`.</span>
+<a name="line-18942"></a><span class='hs-definition'>matrixDeterminant</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18943"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MatrixDeterminant"</span>
+<a name="line-18944"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18945"></a>        <span class='hs-varid'>input</span>
+<a name="line-18946"></a><span class='hs-comment'>{-
+<a name="line-18947"></a>attr {
+<a name="line-18948"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-18949"></a>  name: "T"
+<a name="line-18950"></a>  type: "type"
+<a name="line-18951"></a>}
+<a name="line-18952"></a>input_arg {
+<a name="line-18953"></a>  description: "Shape is `[..., M, M]`." name: "input" type_attr: "T"
+<a name="line-18954"></a>}
+<a name="line-18955"></a>output_arg {
+<a name="line-18956"></a>  description: "Shape is `[...]`." name: "output" type_attr: "T"
+<a name="line-18957"></a>}
+<a name="line-18958"></a>-}</span>
+<a name="line-18959"></a>
+<a name="line-18960"></a><span class='hs-comment'>-- | Computes cos of x element-wise.</span>
+<a name="line-18961"></a>
+<a name="line-18962"></a><a name="cos"></a><span class='hs-definition'>cos</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18963"></a>                                            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-18964"></a>                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-18965"></a>                                            <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-18966"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-18967"></a><span class='hs-definition'>cos</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-18968"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Cos"</span>
+<a name="line-18969"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-18970"></a>        <span class='hs-varid'>x</span>
+<a name="line-18971"></a><span class='hs-comment'>{-
+<a name="line-18972"></a>attr {
+<a name="line-18973"></a>  allowed_values {
+<a name="line-18974"></a>    list {
+<a name="line-18975"></a>      type: DT_HALF
+<a name="line-18976"></a>      type: DT_FLOAT
+<a name="line-18977"></a>      type: DT_DOUBLE
+<a name="line-18978"></a>      type: DT_COMPLEX64
+<a name="line-18979"></a>      type: DT_COMPLEX128
+<a name="line-18980"></a>    }
+<a name="line-18981"></a>  }
+<a name="line-18982"></a>  name: "T"
+<a name="line-18983"></a>  type: "type"
+<a name="line-18984"></a>}
+<a name="line-18985"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-18986"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-18987"></a>-}</span>
+<a name="line-18988"></a>
+<a name="line-18989"></a><a name="batchToSpace"></a><span class='hs-comment'>-- | BatchToSpace for 4-D tensors of type T.</span>
+<a name="line-18990"></a><span class='hs-comment'>--</span>
+<a name="line-18991"></a><span class='hs-comment'>-- This is a legacy version of the more general BatchToSpaceND.</span>
+<a name="line-18992"></a><span class='hs-comment'>-- </span>
+<a name="line-18993"></a><span class='hs-comment'>-- Rearranges (permutes) data from batch into blocks of spatial data, followed by</span>
+<a name="line-18994"></a><span class='hs-comment'>-- cropping. This is the reverse transformation of SpaceToBatch. More specifically,</span>
+<a name="line-18995"></a><span class='hs-comment'>-- this op outputs a copy of the input tensor where values from the `batch`</span>
+<a name="line-18996"></a><span class='hs-comment'>-- dimension are moved in spatial blocks to the `height` and `width` dimensions,</span>
+<a name="line-18997"></a><span class='hs-comment'>-- followed by cropping along the `height` and `width` dimensions.</span>
+<a name="line-18998"></a><span class='hs-definition'>batchToSpace</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span>
+<a name="line-18999"></a>                                       <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-19000"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-19001"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __block_size__</span>
+<a name="line-19002"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: 4-D tensor with shape</span>
+<a name="line-19003"></a>                               <span class='hs-comment'>-- `[batch*block_size*block_size, height_pad/block_size, width_pad/block_size,</span>
+<a name="line-19004"></a>                               <span class='hs-comment'>--   depth]`. Note that the batch size of the input tensor must be divisible by</span>
+<a name="line-19005"></a>                               <span class='hs-comment'>-- `block_size * block_size`.</span>
+<a name="line-19006"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __crops__: 2-D tensor of non-negative integers with shape `[2, 2]`. It specifies</span>
+<a name="line-19007"></a>                                  <span class='hs-comment'>-- how many elements to crop from the intermediate result across the spatial</span>
+<a name="line-19008"></a>                                  <span class='hs-comment'>-- dimensions as follows:</span>
+<a name="line-19009"></a>                                  <span class='hs-comment'>-- </span>
+<a name="line-19010"></a>                                  <span class='hs-comment'>--     crops = [[crop_top, crop_bottom], [crop_left, crop_right]]</span>
+<a name="line-19011"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 4-D with shape `[batch, height, width, depth]`, where:</span>
+<a name="line-19012"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19013"></a>                <span class='hs-comment'>--       height = height_pad - crop_top - crop_bottom</span>
+<a name="line-19014"></a>                <span class='hs-comment'>--       width = width_pad - crop_left - crop_right</span>
+<a name="line-19015"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19016"></a>                <span class='hs-comment'>-- The attr `block_size` must be greater than one. It indicates the block size.</span>
+<a name="line-19017"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19018"></a>                <span class='hs-comment'>-- Some examples:</span>
+<a name="line-19019"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19020"></a>                <span class='hs-comment'>-- (1) For the following input of shape `[4, 1, 1, 1]` and block_size of 2:</span>
+<a name="line-19021"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19022"></a>                <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-19023"></a>                <span class='hs-comment'>-- [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]</span>
+<a name="line-19024"></a>                <span class='hs-comment'>-- ```</span>
+<a name="line-19025"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19026"></a>                <span class='hs-comment'>-- The output tensor has shape `[1, 2, 2, 1]` and value:</span>
+<a name="line-19027"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19028"></a>                <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-19029"></a>                <span class='hs-comment'>-- x = [[[[1], [2]], [[3], [4]]]]</span>
+<a name="line-19030"></a>                <span class='hs-comment'>-- ```</span>
+<a name="line-19031"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19032"></a>                <span class='hs-comment'>-- (2) For the following input of shape `[4, 1, 1, 3]` and block_size of 2:</span>
+<a name="line-19033"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19034"></a>                <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-19035"></a>                <span class='hs-comment'>-- [[[1, 2, 3]], [[4, 5, 6]], [[7, 8, 9]], [[10, 11, 12]]]</span>
+<a name="line-19036"></a>                <span class='hs-comment'>-- ```</span>
+<a name="line-19037"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19038"></a>                <span class='hs-comment'>-- The output tensor has shape `[1, 2, 2, 3]` and value:</span>
+<a name="line-19039"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19040"></a>                <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-19041"></a>                <span class='hs-comment'>-- x = [[[[1, 2, 3], [4, 5, 6]],</span>
+<a name="line-19042"></a>                <span class='hs-comment'>--       [[7, 8, 9], [10, 11, 12]]]]</span>
+<a name="line-19043"></a>                <span class='hs-comment'>-- ```</span>
+<a name="line-19044"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19045"></a>                <span class='hs-comment'>-- (3) For the following input of shape `[4, 2, 2, 1]` and block_size of 2:</span>
+<a name="line-19046"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19047"></a>                <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-19048"></a>                <span class='hs-comment'>-- x = [[[[1], [3]], [[5], [7]]],</span>
+<a name="line-19049"></a>                <span class='hs-comment'>--      [[[2], [4]], [[10], [12]]],</span>
+<a name="line-19050"></a>                <span class='hs-comment'>--      [[[5], [7]], [[13], [15]]],</span>
+<a name="line-19051"></a>                <span class='hs-comment'>--      [[[6], [8]], [[14], [16]]]]</span>
+<a name="line-19052"></a>                <span class='hs-comment'>-- ```</span>
+<a name="line-19053"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19054"></a>                <span class='hs-comment'>-- The output tensor has shape `[1, 4, 4, 1]` and value:</span>
+<a name="line-19055"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19056"></a>                <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-19057"></a>                <span class='hs-comment'>-- x = [[[1],   [2],  [3],  [4]],</span>
+<a name="line-19058"></a>                <span class='hs-comment'>--      [[5],   [6],  [7],  [8]],</span>
+<a name="line-19059"></a>                <span class='hs-comment'>--      [[9],  [10], [11],  [12]],</span>
+<a name="line-19060"></a>                <span class='hs-comment'>--      [[13], [14], [15],  [16]]]</span>
+<a name="line-19061"></a>                <span class='hs-comment'>-- ```</span>
+<a name="line-19062"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19063"></a>                <span class='hs-comment'>-- (4) For the following input of shape `[8, 1, 2, 1]` and block_size of 2:</span>
+<a name="line-19064"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19065"></a>                <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-19066"></a>                <span class='hs-comment'>-- x = [[[[1], [3]]], [[[9], [11]]], [[[2], [4]]], [[[10], [12]]],</span>
+<a name="line-19067"></a>                <span class='hs-comment'>--      [[[5], [7]]], [[[13], [15]]], [[[6], [8]]], [[[14], [16]]]]</span>
+<a name="line-19068"></a>                <span class='hs-comment'>-- ```</span>
+<a name="line-19069"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19070"></a>                <span class='hs-comment'>-- The output tensor has shape `[2, 2, 4, 1]` and value:</span>
+<a name="line-19071"></a>                <span class='hs-comment'>-- </span>
+<a name="line-19072"></a>                <span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-19073"></a>                <span class='hs-comment'>-- x = [[[[1], [3]], [[5], [7]]],</span>
+<a name="line-19074"></a>                <span class='hs-comment'>--      [[[2], [4]], [[10], [12]]],</span>
+<a name="line-19075"></a>                <span class='hs-comment'>--      [[[5], [7]], [[13], [15]]],</span>
+<a name="line-19076"></a>                <span class='hs-comment'>--      [[[6], [8]], [[14], [16]]]]</span>
+<a name="line-19077"></a>                <span class='hs-comment'>-- ```</span>
+<a name="line-19078"></a><span class='hs-definition'>batchToSpace</span> <span class='hs-varid'>block_size</span> <span class='hs-varid'>input</span> <span class='hs-varid'>crops</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19079"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchToSpace"</span>
+<a name="line-19080"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-19081"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span>
+<a name="line-19082"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"block_size"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>block_size</span><span class='hs-layout'>)</span>
+<a name="line-19083"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>crops</span>
+<a name="line-19084"></a><span class='hs-comment'>{-
+<a name="line-19085"></a>attr { name: "T" type: "type" }
+<a name="line-19086"></a>attr {
+<a name="line-19087"></a>  has_minimum: true minimum: 2 name: "block_size" type: "int"
+<a name="line-19088"></a>}
+<a name="line-19089"></a>attr {
+<a name="line-19090"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-19091"></a>  default_value { type: DT_INT32 }
+<a name="line-19092"></a>  name: "Tidx"
+<a name="line-19093"></a>  type: "type"
+<a name="line-19094"></a>}
+<a name="line-19095"></a>input_arg {
+<a name="line-19096"></a>  description: "4-D tensor with shape\n`[batch*block_size*block_size, height_pad/block_size, width_pad/block_size,\n  depth]`. Note that the batch size of the input tensor must be divisible by\n`block_size * block_size`."
+<a name="line-19097"></a>  name: "input"
+<a name="line-19098"></a>  type_attr: "T"
+<a name="line-19099"></a>}
+<a name="line-19100"></a>input_arg {
+<a name="line-19101"></a>  description: "2-D tensor of non-negative integers with shape `[2, 2]`. It specifies\nhow many elements to crop from the intermediate result across the spatial\ndimensions as follows:\n\n    crops = [[crop_top, crop_bottom], [crop_left, crop_right]]"
+<a name="line-19102"></a>  name: "crops"
+<a name="line-19103"></a>  type_attr: "Tidx"
+<a name="line-19104"></a>}
+<a name="line-19105"></a>output_arg {
+<a name="line-19106"></a>  description: "4-D with shape `[batch, height, width, depth]`, where:\n\n      height = height_pad - crop_top - crop_bottom\n      width = width_pad - crop_left - crop_right\n\nThe attr `block_size` must be greater than one. It indicates the block size.\n\nSome examples:\n\n(1) For the following input of shape `[4, 1, 1, 1]` and block_size of 2:\n\n```prettyprint\n[[[[1]]], [[[2]]], [[[3]]], [[[4]]]]\n```\n\nThe output tensor has shape `[1, 2, 2, 1]` and value:\n\n```prettyprint\nx = [[[[1], [2]], [[3], [4]]]]\n```\n\n(2) For the following input of shape `[4, 1, 1, 3]` and block_size of 2:\n\n```prettyprint\n[[[1, 2, 3]], [[4, 5, 6]], [[7, 8, 9]], [[10, 11, 12]]]\n```\n\nThe output tensor has shape `[1, 2, 2, 3]` and value:\n\n```prettyprint\nx = [[[[1, 2, 3], [4, 5, 6]],\n      [[7, 8, 9], [10, 11, 12]]]]\n```\n\n(3) For the following input of shape `[4, 2, 2, 1]` and block_size of 2:\n\n```prettyprint\nx = [[[[1], [3]], [[5], [7]]],\n     [[[2], [4]], [[10], [12]]],\n     [[[5], [7]], [[13], [15]]],\n     [[[6], [8]], [[14], [16]]]]\n```\n\nThe output tensor has shape `[1, 4, 4, 1]` and value:\n\n```prettyprint\nx = [[[1],   [2],  [3],  [4]],\n     [[5],   [6],  [7],  [8]],\n     [[9],  [10], [11],  [12]],\n     [[13], [14], [15],  [16]]]\n```\n\n(4) For the following input of shape `[8, 1, 2, 1]` and block_size of 2:\n\n```prettyprint\nx = [[[[1], [3]]], [[[9], [11]]], [[[2], [4]]], [[[10], [12]]],\n     [[[5], [7]]], [[[13], [15]]], [[[6], [8]]], [[[14], [16]]]]\n```\n\nThe output tensor has shape `[2, 2, 4, 1]` and value:\n\n```prettyprint\nx = [[[[1], [3]], [[5], [7]]],\n     [[[2], [4]], [[10], [12]]],\n     [[[5], [7]], [[13], [15]]],\n     [[[6], [8]], [[14], [16]]]]\n```"
+<a name="line-19107"></a>  name: "output"
+<a name="line-19108"></a>  type_attr: "T"
+<a name="line-19109"></a>}
+<a name="line-19110"></a>-}</span>
+<a name="line-19111"></a>
+<a name="line-19112"></a><a name="sparseToDense"></a><span class='hs-comment'>-- | Converts a sparse representation into a dense tensor.</span>
+<a name="line-19113"></a><span class='hs-comment'>--</span>
+<a name="line-19114"></a><span class='hs-comment'>-- Builds an array `dense` with shape `output_shape` such that</span>
+<a name="line-19115"></a><span class='hs-comment'>-- </span>
+<a name="line-19116"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-19117"></a><span class='hs-comment'>-- # If sparse_indices is scalar</span>
+<a name="line-19118"></a><span class='hs-comment'>-- dense[i] = (i == sparse_indices ? sparse_values : default_value)</span>
+<a name="line-19119"></a><span class='hs-comment'>-- </span>
+<a name="line-19120"></a><span class='hs-comment'>-- # If sparse_indices is a vector, then for each i</span>
+<a name="line-19121"></a><span class='hs-comment'>-- dense[sparse_indices[i]] = sparse_values[i]</span>
+<a name="line-19122"></a><span class='hs-comment'>-- </span>
+<a name="line-19123"></a><span class='hs-comment'>-- # If sparse_indices is an n by d matrix, then for each i in [0, n)</span>
+<a name="line-19124"></a><span class='hs-comment'>-- dense[sparse_indices[i][0], ..., sparse_indices[i][d-1]] = sparse_values[i]</span>
+<a name="line-19125"></a><span class='hs-comment'>-- ```</span>
+<a name="line-19126"></a><span class='hs-comment'>-- </span>
+<a name="line-19127"></a><span class='hs-comment'>-- All other values in `dense` are set to `default_value`.  If `sparse_values` is a</span>
+<a name="line-19128"></a><span class='hs-comment'>-- scalar, all sparse indices are set to this single value.</span>
+<a name="line-19129"></a><span class='hs-comment'>-- </span>
+<a name="line-19130"></a><span class='hs-comment'>-- Indices should be sorted in lexicographic order, and indices must not</span>
+<a name="line-19131"></a><span class='hs-comment'>-- contain any repeats. If `validate_indices` is true, these properties</span>
+<a name="line-19132"></a><span class='hs-comment'>-- are checked during execution.</span>
+<a name="line-19133"></a><span class='hs-definition'>sparseToDense</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-19134"></a>                                                  <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-19135"></a>                                                  <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-19136"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-19137"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __sparse_indices__: 0-D, 1-D, or 2-D.  `sparse_indices[i]` contains the complete</span>
+<a name="line-19138"></a>                                    <span class='hs-comment'>-- index where `sparse_values[i]` will be placed.</span>
+<a name="line-19139"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __output_shape__: 1-D.  Shape of the dense output tensor.</span>
+<a name="line-19140"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __sparse_values__: 1-D.  Values corresponding to each row of `sparse_indices`,</span>
+<a name="line-19141"></a>                                <span class='hs-comment'>-- or a scalar value to be used for all sparse indices.</span>
+<a name="line-19142"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __default_value__: Scalar value to set for indices not specified in</span>
+<a name="line-19143"></a>                                <span class='hs-comment'>-- `sparse_indices`.</span>
+<a name="line-19144"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __dense__: Dense output tensor of shape `output_shape`.</span>
+<a name="line-19145"></a><span class='hs-definition'>sparseToDense</span> <span class='hs-varid'>sparse_indices</span> <span class='hs-varid'>output_shape</span> <span class='hs-varid'>sparse_values</span>
+<a name="line-19146"></a>              <span class='hs-varid'>default_value</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19147"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseToDense"</span>
+<a name="line-19148"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-19149"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-19150"></a>        <span class='hs-varid'>sparse_indices</span> <span class='hs-varid'>output_shape</span> <span class='hs-varid'>sparse_values</span> <span class='hs-varid'>default_value</span>
+<a name="line-19151"></a><span class='hs-comment'>{-
+<a name="line-19152"></a>attr {
+<a name="line-19153"></a>  default_value { b: true }
+<a name="line-19154"></a>  description: "If true, indices are checked to make sure they are sorted in\nlexicographic order and that there are no repeats."
+<a name="line-19155"></a>  name: "validate_indices"
+<a name="line-19156"></a>  type: "bool"
+<a name="line-19157"></a>}
+<a name="line-19158"></a>attr { name: "T" type: "type" }
+<a name="line-19159"></a>attr {
+<a name="line-19160"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-19161"></a>  name: "Tindices"
+<a name="line-19162"></a>  type: "type"
+<a name="line-19163"></a>}
+<a name="line-19164"></a>input_arg {
+<a name="line-19165"></a>  description: "0-D, 1-D, or 2-D.  `sparse_indices[i]` contains the complete\nindex where `sparse_values[i]` will be placed."
+<a name="line-19166"></a>  name: "sparse_indices"
+<a name="line-19167"></a>  type_attr: "Tindices"
+<a name="line-19168"></a>}
+<a name="line-19169"></a>input_arg {
+<a name="line-19170"></a>  description: "1-D.  Shape of the dense output tensor."
+<a name="line-19171"></a>  name: "output_shape"
+<a name="line-19172"></a>  type_attr: "Tindices"
+<a name="line-19173"></a>}
+<a name="line-19174"></a>input_arg {
+<a name="line-19175"></a>  description: "1-D.  Values corresponding to each row of `sparse_indices`,\nor a scalar value to be used for all sparse indices."
+<a name="line-19176"></a>  name: "sparse_values"
+<a name="line-19177"></a>  type_attr: "T"
+<a name="line-19178"></a>}
+<a name="line-19179"></a>input_arg {
+<a name="line-19180"></a>  description: "Scalar value to set for indices not specified in\n`sparse_indices`."
+<a name="line-19181"></a>  name: "default_value"
+<a name="line-19182"></a>  type_attr: "T"
+<a name="line-19183"></a>}
+<a name="line-19184"></a>output_arg {
+<a name="line-19185"></a>  description: "Dense output tensor of shape `output_shape`."
+<a name="line-19186"></a>  name: "dense"
+<a name="line-19187"></a>  type_attr: "T"
+<a name="line-19188"></a>}
+<a name="line-19189"></a>-}</span>
+<a name="line-19190"></a>
+<a name="line-19191"></a><span class='hs-comment'>-- | Computes asin of x element-wise.</span>
+<a name="line-19192"></a>
+<a name="line-19193"></a><a name="asin"></a><span class='hs-definition'>asin</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19194"></a>                                             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19195"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-19196"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-19197"></a>                                             <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-19198"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-19199"></a><span class='hs-definition'>asin</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19200"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Asin"</span>
+<a name="line-19201"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-19202"></a>        <span class='hs-varid'>x</span>
+<a name="line-19203"></a><span class='hs-comment'>{-
+<a name="line-19204"></a>attr {
+<a name="line-19205"></a>  allowed_values {
+<a name="line-19206"></a>    list {
+<a name="line-19207"></a>      type: DT_HALF
+<a name="line-19208"></a>      type: DT_FLOAT
+<a name="line-19209"></a>      type: DT_DOUBLE
+<a name="line-19210"></a>      type: DT_INT32
+<a name="line-19211"></a>      type: DT_INT64
+<a name="line-19212"></a>      type: DT_COMPLEX64
+<a name="line-19213"></a>      type: DT_COMPLEX128
+<a name="line-19214"></a>    }
+<a name="line-19215"></a>  }
+<a name="line-19216"></a>  name: "T"
+<a name="line-19217"></a>  type: "type"
+<a name="line-19218"></a>}
+<a name="line-19219"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-19220"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-19221"></a>-}</span>
+<a name="line-19222"></a>
+<a name="line-19223"></a><span class='hs-comment'>-- | Returns the index with the smallest value across dimensions of a tensor.</span>
+<a name="line-19224"></a>
+<a name="line-19225"></a><a name="argMin"></a><span class='hs-definition'>argMin</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-19226"></a>                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19227"></a>                                         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19228"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-19229"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-19230"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-19231"></a>                                         <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span>
+<a name="line-19232"></a>                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-19233"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-19234"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-19235"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __dimension__: int32, 0 &lt;= dimension &lt; rank(input).  Describes which dimension</span>
+<a name="line-19236"></a>                            <span class='hs-comment'>-- of the input Tensor to reduce across. For vectors, use dimension = 0.</span>
+<a name="line-19237"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-19238"></a><span class='hs-definition'>argMin</span> <span class='hs-varid'>input</span> <span class='hs-varid'>dimension</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19239"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ArgMin"</span>
+<a name="line-19240"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-19241"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-19242"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>dimension</span>
+<a name="line-19243"></a><span class='hs-comment'>{-
+<a name="line-19244"></a>attr {
+<a name="line-19245"></a>  allowed_values {
+<a name="line-19246"></a>    list {
+<a name="line-19247"></a>      type: DT_FLOAT
+<a name="line-19248"></a>      type: DT_DOUBLE
+<a name="line-19249"></a>      type: DT_INT64
+<a name="line-19250"></a>      type: DT_INT32
+<a name="line-19251"></a>      type: DT_UINT8
+<a name="line-19252"></a>      type: DT_UINT16
+<a name="line-19253"></a>      type: DT_INT16
+<a name="line-19254"></a>      type: DT_INT8
+<a name="line-19255"></a>      type: DT_COMPLEX64
+<a name="line-19256"></a>      type: DT_COMPLEX128
+<a name="line-19257"></a>      type: DT_QINT8
+<a name="line-19258"></a>      type: DT_QUINT8
+<a name="line-19259"></a>      type: DT_QINT32
+<a name="line-19260"></a>      type: DT_HALF
+<a name="line-19261"></a>    }
+<a name="line-19262"></a>  }
+<a name="line-19263"></a>  name: "T"
+<a name="line-19264"></a>  type: "type"
+<a name="line-19265"></a>}
+<a name="line-19266"></a>attr {
+<a name="line-19267"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-19268"></a>  default_value { type: DT_INT32 }
+<a name="line-19269"></a>  name: "Tidx"
+<a name="line-19270"></a>  type: "type"
+<a name="line-19271"></a>}
+<a name="line-19272"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-19273"></a>input_arg {
+<a name="line-19274"></a>  description: "int32, 0 &lt;= dimension &lt; rank(input).  Describes which dimension\nof the input Tensor to reduce across. For vectors, use dimension = 0."
+<a name="line-19275"></a>  name: "dimension"
+<a name="line-19276"></a>  type_attr: "Tidx"
+<a name="line-19277"></a>}
+<a name="line-19278"></a>output_arg { name: "output" type: DT_INT64 }
+<a name="line-19279"></a>-}</span>
+<a name="line-19280"></a>
+<a name="line-19281"></a><span class='hs-comment'>-- | Returns which elements of x are Inf.</span>
+<a name="line-19282"></a>
+<a name="line-19283"></a><a name="isInf"></a><span class='hs-definition'>isInf</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-19284"></a>                                              <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-19285"></a>         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-19286"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-19287"></a><span class='hs-definition'>isInf</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19288"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"IsInf"</span>
+<a name="line-19289"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-19290"></a>        <span class='hs-varid'>x</span>
+<a name="line-19291"></a><span class='hs-comment'>{-
+<a name="line-19292"></a>attr {
+<a name="line-19293"></a>  allowed_values {
+<a name="line-19294"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-19295"></a>  }
+<a name="line-19296"></a>  name: "T"
+<a name="line-19297"></a>  type: "type"
+<a name="line-19298"></a>}
+<a name="line-19299"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-19300"></a>output_arg { name: "y" type: DT_BOOL }
+<a name="line-19301"></a>-}</span>
+<a name="line-19302"></a>
+<a name="line-19303"></a><a name="sign"></a><span class='hs-comment'>-- | Returns an element-wise indication of the sign of a number.</span>
+<a name="line-19304"></a><span class='hs-comment'>--</span>
+<a name="line-19305"></a><span class='hs-comment'>-- `y = sign(x) = -1` if `x &lt; 0`; 0 if `x == 0`; 1 if `x &gt; 0`.</span>
+<a name="line-19306"></a><span class='hs-comment'>-- </span>
+<a name="line-19307"></a><span class='hs-comment'>-- For complex numbers, `y = sign(x) = x / |x|` if `x != 0`, otherwise `y = 0`.</span>
+<a name="line-19308"></a><span class='hs-definition'>sign</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19309"></a>                                             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19310"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-19311"></a>                                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-19312"></a>                                             <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-19313"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-19314"></a><span class='hs-definition'>sign</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19315"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Sign"</span>
+<a name="line-19316"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-19317"></a>        <span class='hs-varid'>x</span>
+<a name="line-19318"></a><span class='hs-comment'>{-
+<a name="line-19319"></a>attr {
+<a name="line-19320"></a>  allowed_values {
+<a name="line-19321"></a>    list {
+<a name="line-19322"></a>      type: DT_HALF
+<a name="line-19323"></a>      type: DT_FLOAT
+<a name="line-19324"></a>      type: DT_DOUBLE
+<a name="line-19325"></a>      type: DT_INT32
+<a name="line-19326"></a>      type: DT_INT64
+<a name="line-19327"></a>      type: DT_COMPLEX64
+<a name="line-19328"></a>      type: DT_COMPLEX128
+<a name="line-19329"></a>    }
+<a name="line-19330"></a>  }
+<a name="line-19331"></a>  name: "T"
+<a name="line-19332"></a>  type: "type"
+<a name="line-19333"></a>}
+<a name="line-19334"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-19335"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-19336"></a>-}</span>
+<a name="line-19337"></a>
+<a name="line-19338"></a><a name="add"></a><span class='hs-comment'>-- | Returns x + y element-wise.</span>
+<a name="line-19339"></a><span class='hs-comment'>--</span>
+<a name="line-19340"></a><span class='hs-comment'>-- *NOTE*: `Add` supports broadcasting. `AddN` does not. More about broadcasting</span>
+<a name="line-19341"></a><span class='hs-comment'>-- [here](<a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)">http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)</a></span>
+<a name="line-19342"></a><span class='hs-definition'>add</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19343"></a>                                               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19344"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>,</span>
+<a name="line-19345"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-19346"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-19347"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-19348"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-19349"></a>                                               <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-19350"></a>       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-19351"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-19352"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-19353"></a><span class='hs-definition'>add</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19354"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Add"</span>
+<a name="line-19355"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-19356"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-19357"></a><span class='hs-comment'>{-
+<a name="line-19358"></a>attr {
+<a name="line-19359"></a>  allowed_values {
+<a name="line-19360"></a>    list {
+<a name="line-19361"></a>      type: DT_HALF
+<a name="line-19362"></a>      type: DT_FLOAT
+<a name="line-19363"></a>      type: DT_DOUBLE
+<a name="line-19364"></a>      type: DT_UINT8
+<a name="line-19365"></a>      type: DT_INT8
+<a name="line-19366"></a>      type: DT_INT16
+<a name="line-19367"></a>      type: DT_INT32
+<a name="line-19368"></a>      type: DT_INT64
+<a name="line-19369"></a>      type: DT_COMPLEX64
+<a name="line-19370"></a>      type: DT_COMPLEX128
+<a name="line-19371"></a>      type: DT_STRING
+<a name="line-19372"></a>    }
+<a name="line-19373"></a>  }
+<a name="line-19374"></a>  name: "T"
+<a name="line-19375"></a>  type: "type"
+<a name="line-19376"></a>}
+<a name="line-19377"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-19378"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-19379"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-19380"></a>-}</span>
+<a name="line-19381"></a>
+<a name="line-19382"></a><a name="sparseApplyFtrl"></a><span class='hs-comment'>-- | Update relevant entries in '*var' according to the Ftrl-proximal scheme.</span>
+<a name="line-19383"></a><span class='hs-comment'>--</span>
+<a name="line-19384"></a><span class='hs-comment'>-- That is for rows we have grad for, we update var, accum and linear as follows:</span>
+<a name="line-19385"></a><span class='hs-comment'>-- accum_new = accum + grad * grad</span>
+<a name="line-19386"></a><span class='hs-comment'>-- linear += grad + (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var</span>
+<a name="line-19387"></a><span class='hs-comment'>-- quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2</span>
+<a name="line-19388"></a><span class='hs-comment'>-- var = (sign(linear) * l1 - linear) / quadratic if |linear| &gt; l1 else 0.0</span>
+<a name="line-19389"></a><span class='hs-comment'>-- accum = accum_new</span>
+<a name="line-19390"></a><span class='hs-definition'>sparseApplyFtrl</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>v8</span> <span class='hs-varid'>v9</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-19391"></a>                                                                   <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19392"></a>                                                                           <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19393"></a>                                                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-19394"></a>                                                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-19395"></a>                                                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-19396"></a>                                                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-19397"></a>                                                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-19398"></a>                                                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-19399"></a>                                                                           <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-19400"></a>                                                                           <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-19401"></a>                                                                   <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-19402"></a>                                                                   <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-19403"></a>                                                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-19404"></a>                   <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __var__: Should be from a Variable().</span>
+<a name="line-19405"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __accum__: Should be from a Variable().</span>
+<a name="line-19406"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __linear__: Should be from a Variable().</span>
+<a name="line-19407"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: The gradient.</span>
+<a name="line-19408"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __indices__: A vector of indices into the first dimension of var and accum.</span>
+<a name="line-19409"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr__: Scaling factor. Must be a scalar.</span>
+<a name="line-19410"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l1__: L1 regularization. Must be a scalar.</span>
+<a name="line-19411"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v8</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __l2__: L2 regularization. Must be a scalar.</span>
+<a name="line-19412"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v9</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __lr_power__: Scaling factor. Must be a scalar.</span>
+<a name="line-19413"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__: Same as "var".</span>
+<a name="line-19414"></a><span class='hs-definition'>sparseApplyFtrl</span> <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>linear</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>l1</span> <span class='hs-varid'>l2</span>
+<a name="line-19415"></a>                <span class='hs-varid'>lr_power</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19416"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseApplyFtrl"</span>
+<a name="line-19417"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-19418"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-19419"></a>        <span class='hs-varid'>var</span> <span class='hs-varid'>accum</span> <span class='hs-varid'>linear</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>lr</span> <span class='hs-varid'>l1</span> <span class='hs-varid'>l2</span> <span class='hs-varid'>lr_power</span>
+<a name="line-19420"></a><span class='hs-comment'>{-
+<a name="line-19421"></a>attr {
+<a name="line-19422"></a>  allowed_values {
+<a name="line-19423"></a>    list {
+<a name="line-19424"></a>      type: DT_FLOAT
+<a name="line-19425"></a>      type: DT_DOUBLE
+<a name="line-19426"></a>      type: DT_INT64
+<a name="line-19427"></a>      type: DT_INT32
+<a name="line-19428"></a>      type: DT_UINT8
+<a name="line-19429"></a>      type: DT_UINT16
+<a name="line-19430"></a>      type: DT_INT16
+<a name="line-19431"></a>      type: DT_INT8
+<a name="line-19432"></a>      type: DT_COMPLEX64
+<a name="line-19433"></a>      type: DT_COMPLEX128
+<a name="line-19434"></a>      type: DT_QINT8
+<a name="line-19435"></a>      type: DT_QUINT8
+<a name="line-19436"></a>      type: DT_QINT32
+<a name="line-19437"></a>      type: DT_HALF
+<a name="line-19438"></a>    }
+<a name="line-19439"></a>  }
+<a name="line-19440"></a>  name: "T"
+<a name="line-19441"></a>  type: "type"
+<a name="line-19442"></a>}
+<a name="line-19443"></a>attr {
+<a name="line-19444"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-19445"></a>  name: "Tindices"
+<a name="line-19446"></a>  type: "type"
+<a name="line-19447"></a>}
+<a name="line-19448"></a>attr {
+<a name="line-19449"></a>  default_value { b: false }
+<a name="line-19450"></a>  description: "If `True`, updating of the var and accum tensors will be protected\nby a lock; otherwise the behavior is undefined, but may exhibit less\ncontention."
+<a name="line-19451"></a>  name: "use_locking"
+<a name="line-19452"></a>  type: "bool"
+<a name="line-19453"></a>}
+<a name="line-19454"></a>input_arg {
+<a name="line-19455"></a>  description: "Should be from a Variable()."
+<a name="line-19456"></a>  is_ref: true
+<a name="line-19457"></a>  name: "var"
+<a name="line-19458"></a>  type_attr: "T"
+<a name="line-19459"></a>}
+<a name="line-19460"></a>input_arg {
+<a name="line-19461"></a>  description: "Should be from a Variable()."
+<a name="line-19462"></a>  is_ref: true
+<a name="line-19463"></a>  name: "accum"
+<a name="line-19464"></a>  type_attr: "T"
+<a name="line-19465"></a>}
+<a name="line-19466"></a>input_arg {
+<a name="line-19467"></a>  description: "Should be from a Variable()."
+<a name="line-19468"></a>  is_ref: true
+<a name="line-19469"></a>  name: "linear"
+<a name="line-19470"></a>  type_attr: "T"
+<a name="line-19471"></a>}
+<a name="line-19472"></a>input_arg {
+<a name="line-19473"></a>  description: "The gradient." name: "grad" type_attr: "T"
+<a name="line-19474"></a>}
+<a name="line-19475"></a>input_arg {
+<a name="line-19476"></a>  description: "A vector of indices into the first dimension of var and accum."
+<a name="line-19477"></a>  name: "indices"
+<a name="line-19478"></a>  type_attr: "Tindices"
+<a name="line-19479"></a>}
+<a name="line-19480"></a>input_arg {
+<a name="line-19481"></a>  description: "Scaling factor. Must be a scalar."
+<a name="line-19482"></a>  name: "lr"
+<a name="line-19483"></a>  type_attr: "T"
+<a name="line-19484"></a>}
+<a name="line-19485"></a>input_arg {
+<a name="line-19486"></a>  description: "L1 regularization. Must be a scalar."
+<a name="line-19487"></a>  name: "l1"
+<a name="line-19488"></a>  type_attr: "T"
+<a name="line-19489"></a>}
+<a name="line-19490"></a>input_arg {
+<a name="line-19491"></a>  description: "L2 regularization. Must be a scalar."
+<a name="line-19492"></a>  name: "l2"
+<a name="line-19493"></a>  type_attr: "T"
+<a name="line-19494"></a>}
+<a name="line-19495"></a>input_arg {
+<a name="line-19496"></a>  description: "Scaling factor. Must be a scalar."
+<a name="line-19497"></a>  name: "lr_power"
+<a name="line-19498"></a>  type_attr: "T"
+<a name="line-19499"></a>}
+<a name="line-19500"></a>output_arg {
+<a name="line-19501"></a>  description: "Same as \"var\"."
+<a name="line-19502"></a>  is_ref: true
+<a name="line-19503"></a>  name: "out"
+<a name="line-19504"></a>  type_attr: "T"
+<a name="line-19505"></a>}
+<a name="line-19506"></a>-}</span>
+<a name="line-19507"></a>
+<a name="line-19508"></a><a name="sub"></a><span class='hs-comment'>-- | Returns x - y element-wise.</span>
+<a name="line-19509"></a><span class='hs-comment'>--</span>
+<a name="line-19510"></a><span class='hs-comment'>-- *NOTE*: `Sub` supports broadcasting. More about broadcasting</span>
+<a name="line-19511"></a><span class='hs-comment'>-- [here](<a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)">http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)</a></span>
+<a name="line-19512"></a><span class='hs-definition'>sub</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19513"></a>                                               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19514"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-19515"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-19516"></a>                                               <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-19517"></a>       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-19518"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-19519"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-19520"></a><span class='hs-definition'>sub</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19521"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Sub"</span>
+<a name="line-19522"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-19523"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-19524"></a><span class='hs-comment'>{-
+<a name="line-19525"></a>attr {
+<a name="line-19526"></a>  allowed_values {
+<a name="line-19527"></a>    list {
+<a name="line-19528"></a>      type: DT_HALF
+<a name="line-19529"></a>      type: DT_FLOAT
+<a name="line-19530"></a>      type: DT_DOUBLE
+<a name="line-19531"></a>      type: DT_INT32
+<a name="line-19532"></a>      type: DT_INT64
+<a name="line-19533"></a>      type: DT_COMPLEX64
+<a name="line-19534"></a>      type: DT_COMPLEX128
+<a name="line-19535"></a>    }
+<a name="line-19536"></a>  }
+<a name="line-19537"></a>  name: "T"
+<a name="line-19538"></a>  type: "type"
+<a name="line-19539"></a>}
+<a name="line-19540"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-19541"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-19542"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-19543"></a>-}</span>
+<a name="line-19544"></a>
+<a name="line-19545"></a><span class='hs-comment'>-- | </span>
+<a name="line-19546"></a>
+<a name="line-19547"></a><a name="batchFFT3D"></a><span class='hs-definition'>batchFFT3D</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-19548"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-19549"></a><span class='hs-definition'>batchFFT3D</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19550"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchFFT3D"</span><span class='hs-layout'>)</span>
+<a name="line-19551"></a>        <span class='hs-varid'>input</span>
+<a name="line-19552"></a><span class='hs-comment'>{-
+<a name="line-19553"></a>input_arg { name: "input" type: DT_COMPLEX64 }
+<a name="line-19554"></a>output_arg { name: "output" type: DT_COMPLEX64 }
+<a name="line-19555"></a>-}</span>
+<a name="line-19556"></a>
+<a name="line-19557"></a><a name="sparseReduceSumSparse"></a><span class='hs-comment'>-- | Computes the sum of elements across dimensions of a SparseTensor.</span>
+<a name="line-19558"></a><span class='hs-comment'>--</span>
+<a name="line-19559"></a><span class='hs-comment'>-- This Op takes a SparseTensor and is the sparse counterpart to</span>
+<a name="line-19560"></a><span class='hs-comment'>-- `tf.reduce_sum()`.  In contrast to SparseReduceSum, this Op returns a</span>
+<a name="line-19561"></a><span class='hs-comment'>-- SparseTensor.</span>
+<a name="line-19562"></a><span class='hs-comment'>-- </span>
+<a name="line-19563"></a><span class='hs-comment'>-- Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless</span>
+<a name="line-19564"></a><span class='hs-comment'>-- `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in</span>
+<a name="line-19565"></a><span class='hs-comment'>-- `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained</span>
+<a name="line-19566"></a><span class='hs-comment'>-- with length 1.</span>
+<a name="line-19567"></a><span class='hs-comment'>-- </span>
+<a name="line-19568"></a><span class='hs-comment'>-- If `reduction_axes` has no entries, all dimensions are reduced, and a tensor</span>
+<a name="line-19569"></a><span class='hs-comment'>-- with a single element is returned.  Additionally, the axes can be negative,</span>
+<a name="line-19570"></a><span class='hs-comment'>-- which are interpreted according to the indexing rules in Python.</span>
+<a name="line-19571"></a><span class='hs-definition'>sparseReduceSumSparse</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-19572"></a>                                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19573"></a>                                                         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19574"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-19575"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-19576"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-19577"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-19578"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-19579"></a>                                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-19580"></a>                                                         <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-19581"></a>                         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __input_indices__: 2-D.  `N x R` matrix with the indices of non-empty values in a</span>
+<a name="line-19582"></a>                                                  <span class='hs-comment'>-- SparseTensor, possibly not in canonical ordering.</span>
+<a name="line-19583"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input_values__: 1-D.  `N` non-empty values corresponding to `input_indices`.</span>
+<a name="line-19584"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __input_shape__: 1-D.  Shape of the input SparseTensor.</span>
+<a name="line-19585"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __reduction_axes__: 1-D.  Length-`K` vector containing the reduction axes.</span>
+<a name="line-19586"></a>                         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-19587"></a>                             <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-19588"></a>                         <span class='hs-comment'>-- ^ (__output_indices__, __output_values__, __output_shape__)</span>
+<a name="line-19589"></a>                         <span class='hs-comment'>--</span>
+<a name="line-19590"></a>                         <span class='hs-comment'>-- * __output_indices__</span>
+<a name="line-19591"></a>                         <span class='hs-comment'>--</span>
+<a name="line-19592"></a>                         <span class='hs-comment'>-- * __output_values__</span>
+<a name="line-19593"></a>                         <span class='hs-comment'>--</span>
+<a name="line-19594"></a>                         <span class='hs-comment'>-- * __output_shape__</span>
+<a name="line-19595"></a><span class='hs-definition'>sparseReduceSumSparse</span> <span class='hs-varid'>input_indices</span> <span class='hs-varid'>input_values</span> <span class='hs-varid'>input_shape</span>
+<a name="line-19596"></a>                      <span class='hs-varid'>reduction_axes</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19597"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseReduceSumSparse"</span>
+<a name="line-19598"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-19599"></a>        <span class='hs-varid'>input_indices</span> <span class='hs-varid'>input_values</span> <span class='hs-varid'>input_shape</span> <span class='hs-varid'>reduction_axes</span>
+<a name="line-19600"></a><span class='hs-comment'>{-
+<a name="line-19601"></a>attr {
+<a name="line-19602"></a>  default_value { b: false }
+<a name="line-19603"></a>  description: "If true, retain reduced dimensions with length 1."
+<a name="line-19604"></a>  name: "keep_dims"
+<a name="line-19605"></a>  type: "bool"
+<a name="line-19606"></a>}
+<a name="line-19607"></a>attr {
+<a name="line-19608"></a>  allowed_values {
+<a name="line-19609"></a>    list {
+<a name="line-19610"></a>      type: DT_FLOAT
+<a name="line-19611"></a>      type: DT_DOUBLE
+<a name="line-19612"></a>      type: DT_INT64
+<a name="line-19613"></a>      type: DT_INT32
+<a name="line-19614"></a>      type: DT_UINT8
+<a name="line-19615"></a>      type: DT_UINT16
+<a name="line-19616"></a>      type: DT_INT16
+<a name="line-19617"></a>      type: DT_INT8
+<a name="line-19618"></a>      type: DT_COMPLEX64
+<a name="line-19619"></a>      type: DT_COMPLEX128
+<a name="line-19620"></a>      type: DT_QINT8
+<a name="line-19621"></a>      type: DT_QUINT8
+<a name="line-19622"></a>      type: DT_QINT32
+<a name="line-19623"></a>      type: DT_HALF
+<a name="line-19624"></a>    }
+<a name="line-19625"></a>  }
+<a name="line-19626"></a>  name: "T"
+<a name="line-19627"></a>  type: "type"
+<a name="line-19628"></a>}
+<a name="line-19629"></a>input_arg {
+<a name="line-19630"></a>  description: "2-D.  `N x R` matrix with the indices of non-empty values in a\nSparseTensor, possibly not in canonical ordering."
+<a name="line-19631"></a>  name: "input_indices"
+<a name="line-19632"></a>  type: DT_INT64
+<a name="line-19633"></a>}
+<a name="line-19634"></a>input_arg {
+<a name="line-19635"></a>  description: "1-D.  `N` non-empty values corresponding to `input_indices`."
+<a name="line-19636"></a>  name: "input_values"
+<a name="line-19637"></a>  type_attr: "T"
+<a name="line-19638"></a>}
+<a name="line-19639"></a>input_arg {
+<a name="line-19640"></a>  description: "1-D.  Shape of the input SparseTensor."
+<a name="line-19641"></a>  name: "input_shape"
+<a name="line-19642"></a>  type: DT_INT64
+<a name="line-19643"></a>}
+<a name="line-19644"></a>input_arg {
+<a name="line-19645"></a>  description: "1-D.  Length-`K` vector containing the reduction axes."
+<a name="line-19646"></a>  name: "reduction_axes"
+<a name="line-19647"></a>  type: DT_INT32
+<a name="line-19648"></a>}
+<a name="line-19649"></a>output_arg { name: "output_indices" type: DT_INT64 }
+<a name="line-19650"></a>output_arg { name: "output_values" type_attr: "T" }
+<a name="line-19651"></a>output_arg { name: "output_shape" type: DT_INT64 }
+<a name="line-19652"></a>-}</span>
+<a name="line-19653"></a>
+<a name="line-19654"></a><a name="biasAdd"></a><span class='hs-comment'>-- | Adds `bias` to `value`.</span>
+<a name="line-19655"></a><span class='hs-comment'>--</span>
+<a name="line-19656"></a><span class='hs-comment'>-- This is a special case of `tf.add` where `bias` is restricted to be 1-D.</span>
+<a name="line-19657"></a><span class='hs-comment'>-- Broadcasting is supported, so `value` may have any number of dimensions.</span>
+<a name="line-19658"></a><span class='hs-definition'>biasAdd</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-19659"></a>                             <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19660"></a>                                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19661"></a>                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-19662"></a>                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-19663"></a>                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-19664"></a>                                     <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-19665"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __value__: Any number of dimensions.</span>
+<a name="line-19666"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __bias__: 1-D with size the last dimension of `value`.</span>
+<a name="line-19667"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Broadcasted sum of `value` and `bias`.</span>
+<a name="line-19668"></a><span class='hs-definition'>biasAdd</span> <span class='hs-varid'>value</span> <span class='hs-varid'>bias</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19669"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BiasAdd"</span>
+<a name="line-19670"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-19671"></a>        <span class='hs-varid'>value</span> <span class='hs-varid'>bias</span>
+<a name="line-19672"></a><span class='hs-comment'>{-
+<a name="line-19673"></a>attr {
+<a name="line-19674"></a>  allowed_values {
+<a name="line-19675"></a>    list {
+<a name="line-19676"></a>      type: DT_FLOAT
+<a name="line-19677"></a>      type: DT_DOUBLE
+<a name="line-19678"></a>      type: DT_INT64
+<a name="line-19679"></a>      type: DT_INT32
+<a name="line-19680"></a>      type: DT_UINT8
+<a name="line-19681"></a>      type: DT_UINT16
+<a name="line-19682"></a>      type: DT_INT16
+<a name="line-19683"></a>      type: DT_INT8
+<a name="line-19684"></a>      type: DT_COMPLEX64
+<a name="line-19685"></a>      type: DT_COMPLEX128
+<a name="line-19686"></a>      type: DT_QINT8
+<a name="line-19687"></a>      type: DT_QUINT8
+<a name="line-19688"></a>      type: DT_QINT32
+<a name="line-19689"></a>      type: DT_HALF
+<a name="line-19690"></a>    }
+<a name="line-19691"></a>  }
+<a name="line-19692"></a>  name: "T"
+<a name="line-19693"></a>  type: "type"
+<a name="line-19694"></a>}
+<a name="line-19695"></a>attr {
+<a name="line-19696"></a>  allowed_values { list { s: "NHWC" s: "NCHW" } }
+<a name="line-19697"></a>  default_value { s: "NHWC" }
+<a name="line-19698"></a>  description: "Specify the data format of the input and output data. With the\ndefault format \"NHWC\", the bias tensor will be added to the last dimension\nof the value tensor.\nAlternatively, the format could be \"NCHW\", the data storage order of:\n    [batch, in_channels, in_height, in_width].\nThe tensor will be added to \"in_channels\", the third-to-the-last\n    dimension."
+<a name="line-19699"></a>  name: "data_format"
+<a name="line-19700"></a>  type: "string"
+<a name="line-19701"></a>}
+<a name="line-19702"></a>input_arg {
+<a name="line-19703"></a>  description: "Any number of dimensions."
+<a name="line-19704"></a>  name: "value"
+<a name="line-19705"></a>  type_attr: "T"
+<a name="line-19706"></a>}
+<a name="line-19707"></a>input_arg {
+<a name="line-19708"></a>  description: "1-D with size the last dimension of `value`."
+<a name="line-19709"></a>  name: "bias"
+<a name="line-19710"></a>  type_attr: "T"
+<a name="line-19711"></a>}
+<a name="line-19712"></a>output_arg {
+<a name="line-19713"></a>  description: "Broadcasted sum of `value` and `bias`."
+<a name="line-19714"></a>  name: "output"
+<a name="line-19715"></a>  type_attr: "T"
+<a name="line-19716"></a>}
+<a name="line-19717"></a>-}</span>
+<a name="line-19718"></a>
+<a name="line-19719"></a><a name="mul"></a><span class='hs-comment'>-- | Returns x * y element-wise.</span>
+<a name="line-19720"></a><span class='hs-comment'>--</span>
+<a name="line-19721"></a><span class='hs-comment'>-- *NOTE*: `Mul` supports broadcasting. More about broadcasting</span>
+<a name="line-19722"></a><span class='hs-comment'>-- [here](<a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)">http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)</a></span>
+<a name="line-19723"></a><span class='hs-definition'>mul</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19724"></a>                                               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19725"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-19726"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-19727"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-19728"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-19729"></a>                                               <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-19730"></a>       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-19731"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-19732"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-19733"></a><span class='hs-definition'>mul</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19734"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Mul"</span>
+<a name="line-19735"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-19736"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-19737"></a><span class='hs-comment'>{-
+<a name="line-19738"></a>attr {
+<a name="line-19739"></a>  allowed_values {
+<a name="line-19740"></a>    list {
+<a name="line-19741"></a>      type: DT_HALF
+<a name="line-19742"></a>      type: DT_FLOAT
+<a name="line-19743"></a>      type: DT_DOUBLE
+<a name="line-19744"></a>      type: DT_UINT8
+<a name="line-19745"></a>      type: DT_INT8
+<a name="line-19746"></a>      type: DT_UINT16
+<a name="line-19747"></a>      type: DT_INT16
+<a name="line-19748"></a>      type: DT_INT32
+<a name="line-19749"></a>      type: DT_INT64
+<a name="line-19750"></a>      type: DT_COMPLEX64
+<a name="line-19751"></a>      type: DT_COMPLEX128
+<a name="line-19752"></a>    }
+<a name="line-19753"></a>  }
+<a name="line-19754"></a>  name: "T"
+<a name="line-19755"></a>  type: "type"
+<a name="line-19756"></a>}
+<a name="line-19757"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-19758"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-19759"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-19760"></a>-}</span>
+<a name="line-19761"></a>
+<a name="line-19762"></a><a name="div"></a><span class='hs-comment'>-- | Returns x / y element-wise.</span>
+<a name="line-19763"></a><span class='hs-comment'>--</span>
+<a name="line-19764"></a><span class='hs-comment'>-- *NOTE*: `Div` supports broadcasting. More about broadcasting</span>
+<a name="line-19765"></a><span class='hs-comment'>-- [here](<a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)">http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)</a></span>
+<a name="line-19766"></a><span class='hs-definition'>div</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19767"></a>                                               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19768"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-19769"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-19770"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-19771"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-19772"></a>                                               <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-19773"></a>       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-19774"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-19775"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-19776"></a><span class='hs-definition'>div</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19777"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Div"</span>
+<a name="line-19778"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-19779"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-19780"></a><span class='hs-comment'>{-
+<a name="line-19781"></a>attr {
+<a name="line-19782"></a>  allowed_values {
+<a name="line-19783"></a>    list {
+<a name="line-19784"></a>      type: DT_HALF
+<a name="line-19785"></a>      type: DT_FLOAT
+<a name="line-19786"></a>      type: DT_DOUBLE
+<a name="line-19787"></a>      type: DT_UINT8
+<a name="line-19788"></a>      type: DT_INT8
+<a name="line-19789"></a>      type: DT_UINT16
+<a name="line-19790"></a>      type: DT_INT16
+<a name="line-19791"></a>      type: DT_INT32
+<a name="line-19792"></a>      type: DT_INT64
+<a name="line-19793"></a>      type: DT_COMPLEX64
+<a name="line-19794"></a>      type: DT_COMPLEX128
+<a name="line-19795"></a>    }
+<a name="line-19796"></a>  }
+<a name="line-19797"></a>  name: "T"
+<a name="line-19798"></a>  type: "type"
+<a name="line-19799"></a>}
+<a name="line-19800"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-19801"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-19802"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-19803"></a>-}</span>
+<a name="line-19804"></a>
+<a name="line-19805"></a><a name="loopCond"></a><span class='hs-comment'>-- | Forwards the input to the output.</span>
+<a name="line-19806"></a><span class='hs-comment'>--</span>
+<a name="line-19807"></a><span class='hs-comment'>-- This operator represents the loop termination condition used by the</span>
+<a name="line-19808"></a><span class='hs-comment'>-- "pivot" switches of a loop.</span>
+<a name="line-19809"></a><span class='hs-definition'>loopCond</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __input__: A boolean scalar, representing the branch predicate of the Switch op.</span>
+<a name="line-19810"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __output__: The same tensor as `input`.</span>
+<a name="line-19811"></a><span class='hs-definition'>loopCond</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19812"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"LoopCond"</span><span class='hs-layout'>)</span>
+<a name="line-19813"></a>        <span class='hs-varid'>input</span>
+<a name="line-19814"></a><span class='hs-comment'>{-
+<a name="line-19815"></a>input_arg {
+<a name="line-19816"></a>  description: "A boolean scalar, representing the branch predicate of the Switch op."
+<a name="line-19817"></a>  name: "input"
+<a name="line-19818"></a>  type: DT_BOOL
+<a name="line-19819"></a>}
+<a name="line-19820"></a>output_arg {
+<a name="line-19821"></a>  description: "The same tensor as `input`."
+<a name="line-19822"></a>  name: "output"
+<a name="line-19823"></a>  type: DT_BOOL
+<a name="line-19824"></a>}
+<a name="line-19825"></a>-}</span>
+<a name="line-19826"></a>
+<a name="line-19827"></a><a name="squaredDifference"></a><span class='hs-comment'>-- | Returns (x - y)(x - y) element-wise.</span>
+<a name="line-19828"></a><span class='hs-comment'>--</span>
+<a name="line-19829"></a><span class='hs-comment'>-- *NOTE*: `SquaredDifference` supports broadcasting. More about broadcasting</span>
+<a name="line-19830"></a><span class='hs-comment'>-- [here](<a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)">http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)</a></span>
+<a name="line-19831"></a><span class='hs-definition'>squaredDifference</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-19832"></a>                                       <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19833"></a>                                               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-19834"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-19835"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-19836"></a>                                               <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-19837"></a>                     <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-19838"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-19839"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-19840"></a><span class='hs-definition'>squaredDifference</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19841"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SquaredDifference"</span>
+<a name="line-19842"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-19843"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-19844"></a><span class='hs-comment'>{-
+<a name="line-19845"></a>attr {
+<a name="line-19846"></a>  allowed_values {
+<a name="line-19847"></a>    list {
+<a name="line-19848"></a>      type: DT_HALF
+<a name="line-19849"></a>      type: DT_FLOAT
+<a name="line-19850"></a>      type: DT_DOUBLE
+<a name="line-19851"></a>      type: DT_INT32
+<a name="line-19852"></a>      type: DT_INT64
+<a name="line-19853"></a>      type: DT_COMPLEX64
+<a name="line-19854"></a>      type: DT_COMPLEX128
+<a name="line-19855"></a>    }
+<a name="line-19856"></a>  }
+<a name="line-19857"></a>  name: "T"
+<a name="line-19858"></a>  type: "type"
+<a name="line-19859"></a>}
+<a name="line-19860"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-19861"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-19862"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-19863"></a>-}</span>
+<a name="line-19864"></a>
+<a name="line-19865"></a><a name="maximum"></a><span class='hs-comment'>-- | Returns the max of x and y (i.e. x &gt; y ? x : y) element-wise.</span>
+<a name="line-19866"></a><span class='hs-comment'>--</span>
+<a name="line-19867"></a><span class='hs-comment'>-- *NOTE*: `Maximum` supports broadcasting. More about broadcasting</span>
+<a name="line-19868"></a><span class='hs-comment'>-- [here](<a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)">http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)</a></span>
+<a name="line-19869"></a><span class='hs-definition'>maximum</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-19870"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-19871"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-19872"></a>                                                   <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-19873"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-19874"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-19875"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-19876"></a><span class='hs-definition'>maximum</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19877"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Maximum"</span>
+<a name="line-19878"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-19879"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-19880"></a><span class='hs-comment'>{-
+<a name="line-19881"></a>attr {
+<a name="line-19882"></a>  allowed_values {
+<a name="line-19883"></a>    list {
+<a name="line-19884"></a>      type: DT_HALF
+<a name="line-19885"></a>      type: DT_FLOAT
+<a name="line-19886"></a>      type: DT_DOUBLE
+<a name="line-19887"></a>      type: DT_INT32
+<a name="line-19888"></a>      type: DT_INT64
+<a name="line-19889"></a>    }
+<a name="line-19890"></a>  }
+<a name="line-19891"></a>  name: "T"
+<a name="line-19892"></a>  type: "type"
+<a name="line-19893"></a>}
+<a name="line-19894"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-19895"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-19896"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-19897"></a>-}</span>
+<a name="line-19898"></a>
+<a name="line-19899"></a><a name="logUniformCandidateSampler"></a><span class='hs-comment'>-- | Generates labels for candidate sampling with a log-uniform distribution.</span>
+<a name="line-19900"></a><span class='hs-comment'>--</span>
+<a name="line-19901"></a><span class='hs-comment'>-- See explanations of candidate sampling and the data formats at</span>
+<a name="line-19902"></a><span class='hs-comment'>-- go/candidate-sampling.</span>
+<a name="line-19903"></a><span class='hs-comment'>-- </span>
+<a name="line-19904"></a><span class='hs-comment'>-- For each batch, this op picks a single set of sampled candidate labels.</span>
+<a name="line-19905"></a><span class='hs-comment'>-- </span>
+<a name="line-19906"></a><span class='hs-comment'>-- The advantages of sampling candidates per-batch are simplicity and the</span>
+<a name="line-19907"></a><span class='hs-comment'>-- possibility of efficient dense matrix multiplication. The disadvantage is that</span>
+<a name="line-19908"></a><span class='hs-comment'>-- the sampled candidates must be chosen independently of the context and of the</span>
+<a name="line-19909"></a><span class='hs-comment'>-- true labels.</span>
+<a name="line-19910"></a><span class='hs-definition'>logUniformCandidateSampler</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_sampled__: Number of candidates to randomly sample per batch.</span>
+<a name="line-19911"></a>                              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_true__: Number of true labels per context.</span>
+<a name="line-19912"></a>                              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __range_max__: The sampler will sample integers from the interval [0, range_max).</span>
+<a name="line-19913"></a>                              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __unique__: If unique is true, we sample with rejection, so that all sampled</span>
+<a name="line-19914"></a>                                      <span class='hs-comment'>-- candidates in a batch are unique. This requires some approximation to</span>
+<a name="line-19915"></a>                                      <span class='hs-comment'>-- estimate the post-rejection sampling probabilities.</span>
+<a name="line-19916"></a>                              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __true_classes__: A batch_size * num_true matrix, in which each row contains the</span>
+<a name="line-19917"></a>                                                          <span class='hs-comment'>-- IDs of the num_true target_classes in the corresponding original label.</span>
+<a name="line-19918"></a>                              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-19919"></a>                                  <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span>
+<a name="line-19920"></a>                              <span class='hs-comment'>-- ^ (__sampled_candidates__, __true_expected_count__, __sampled_expected_count__)</span>
+<a name="line-19921"></a>                              <span class='hs-comment'>--</span>
+<a name="line-19922"></a>                              <span class='hs-comment'>-- * __sampled_candidates__: A vector of length num_sampled, in which each element is</span>
+<a name="line-19923"></a>                              <span class='hs-comment'>-- the ID of a sampled candidate.</span>
+<a name="line-19924"></a>                              <span class='hs-comment'>--</span>
+<a name="line-19925"></a>                              <span class='hs-comment'>-- * __true_expected_count__: A batch_size * num_true matrix, representing</span>
+<a name="line-19926"></a>                              <span class='hs-comment'>-- the number of times each candidate is expected to occur in a batch</span>
+<a name="line-19927"></a>                              <span class='hs-comment'>-- of sampled candidates. If unique=true, then this is a probability.</span>
+<a name="line-19928"></a>                              <span class='hs-comment'>--</span>
+<a name="line-19929"></a>                              <span class='hs-comment'>-- * __sampled_expected_count__: A vector of length num_sampled, for each sampled</span>
+<a name="line-19930"></a>                              <span class='hs-comment'>-- candidate representing the number of times the candidate is expected</span>
+<a name="line-19931"></a>                              <span class='hs-comment'>-- to occur in a batch of sampled candidates.  If unique=true, then this is a</span>
+<a name="line-19932"></a>                              <span class='hs-comment'>-- probability.</span>
+<a name="line-19933"></a><span class='hs-definition'>logUniformCandidateSampler</span> <span class='hs-varid'>num_sampled</span> <span class='hs-varid'>num_true</span> <span class='hs-varid'>range_max</span> <span class='hs-varid'>unique</span>
+<a name="line-19934"></a>                           <span class='hs-varid'>true_classes</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-19935"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"LogUniformCandidateSampler"</span>
+<a name="line-19936"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_sampled"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_sampled</span>
+<a name="line-19937"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_true"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_true</span>
+<a name="line-19938"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"range_max"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>range_max</span>
+<a name="line-19939"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"unique"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>unique</span><span class='hs-layout'>)</span>
+<a name="line-19940"></a>        <span class='hs-varid'>true_classes</span>
+<a name="line-19941"></a><span class='hs-comment'>{-
+<a name="line-19942"></a>attr {
+<a name="line-19943"></a>  description: "Number of true labels per context."
+<a name="line-19944"></a>  has_minimum: true
+<a name="line-19945"></a>  minimum: 1
+<a name="line-19946"></a>  name: "num_true"
+<a name="line-19947"></a>  type: "int"
+<a name="line-19948"></a>}
+<a name="line-19949"></a>attr {
+<a name="line-19950"></a>  description: "Number of candidates to randomly sample per batch."
+<a name="line-19951"></a>  has_minimum: true
+<a name="line-19952"></a>  minimum: 1
+<a name="line-19953"></a>  name: "num_sampled"
+<a name="line-19954"></a>  type: "int"
+<a name="line-19955"></a>}
+<a name="line-19956"></a>attr {
+<a name="line-19957"></a>  description: "If unique is true, we sample with rejection, so that all sampled\ncandidates in a batch are unique. This requires some approximation to\nestimate the post-rejection sampling probabilities."
+<a name="line-19958"></a>  name: "unique"
+<a name="line-19959"></a>  type: "bool"
+<a name="line-19960"></a>}
+<a name="line-19961"></a>attr {
+<a name="line-19962"></a>  description: "The sampler will sample integers from the interval [0, range_max)."
+<a name="line-19963"></a>  has_minimum: true
+<a name="line-19964"></a>  minimum: 1
+<a name="line-19965"></a>  name: "range_max"
+<a name="line-19966"></a>  type: "int"
+<a name="line-19967"></a>}
+<a name="line-19968"></a>attr {
+<a name="line-19969"></a>  default_value { i: 0 }
+<a name="line-19970"></a>  description: "If either seed or seed2 are set to be non-zero, the random number\ngenerator is seeded by the given seed.  Otherwise, it is seeded by a\nrandom seed."
+<a name="line-19971"></a>  name: "seed"
+<a name="line-19972"></a>  type: "int"
+<a name="line-19973"></a>}
+<a name="line-19974"></a>attr {
+<a name="line-19975"></a>  default_value { i: 0 }
+<a name="line-19976"></a>  description: "An second seed to avoid seed collision."
+<a name="line-19977"></a>  name: "seed2"
+<a name="line-19978"></a>  type: "int"
+<a name="line-19979"></a>}
+<a name="line-19980"></a>input_arg {
+<a name="line-19981"></a>  description: "A batch_size * num_true matrix, in which each row contains the\nIDs of the num_true target_classes in the corresponding original label."
+<a name="line-19982"></a>  name: "true_classes"
+<a name="line-19983"></a>  type: DT_INT64
+<a name="line-19984"></a>}
+<a name="line-19985"></a>output_arg {
+<a name="line-19986"></a>  description: "A vector of length num_sampled, in which each element is\nthe ID of a sampled candidate."
+<a name="line-19987"></a>  name: "sampled_candidates"
+<a name="line-19988"></a>  type: DT_INT64
+<a name="line-19989"></a>}
+<a name="line-19990"></a>output_arg {
+<a name="line-19991"></a>  description: "A batch_size * num_true matrix, representing\nthe number of times each candidate is expected to occur in a batch\nof sampled candidates. If unique=true, then this is a probability."
+<a name="line-19992"></a>  name: "true_expected_count"
+<a name="line-19993"></a>  type: DT_FLOAT
+<a name="line-19994"></a>}
+<a name="line-19995"></a>output_arg {
+<a name="line-19996"></a>  description: "A vector of length num_sampled, for each sampled\ncandidate representing the number of times the candidate is expected\nto occur in a batch of sampled candidates.  If unique=true, then this is a\nprobability."
+<a name="line-19997"></a>  name: "sampled_expected_count"
+<a name="line-19998"></a>  type: DT_FLOAT
+<a name="line-19999"></a>}
+<a name="line-20000"></a>-}</span>
+<a name="line-20001"></a>
+<a name="line-20002"></a><a name="less"></a><span class='hs-comment'>-- | Returns the truth value of (x &lt; y) element-wise.</span>
+<a name="line-20003"></a><span class='hs-comment'>--</span>
+<a name="line-20004"></a><span class='hs-comment'>-- *NOTE*: `Less` supports broadcasting. More about broadcasting</span>
+<a name="line-20005"></a><span class='hs-comment'>-- [here](<a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)">http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)</a></span>
+<a name="line-20006"></a><span class='hs-definition'>less</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-20007"></a>                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-20008"></a>                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-20009"></a>                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-20010"></a>                                                <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20011"></a>        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-20012"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-20013"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-20014"></a><span class='hs-definition'>less</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20015"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Less"</span>
+<a name="line-20016"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20017"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-20018"></a><span class='hs-comment'>{-
+<a name="line-20019"></a>attr {
+<a name="line-20020"></a>  allowed_values {
+<a name="line-20021"></a>    list {
+<a name="line-20022"></a>      type: DT_FLOAT
+<a name="line-20023"></a>      type: DT_DOUBLE
+<a name="line-20024"></a>      type: DT_INT32
+<a name="line-20025"></a>      type: DT_INT64
+<a name="line-20026"></a>      type: DT_UINT8
+<a name="line-20027"></a>      type: DT_INT16
+<a name="line-20028"></a>      type: DT_INT8
+<a name="line-20029"></a>      type: DT_UINT16
+<a name="line-20030"></a>      type: DT_HALF
+<a name="line-20031"></a>    }
+<a name="line-20032"></a>  }
+<a name="line-20033"></a>  name: "T"
+<a name="line-20034"></a>  type: "type"
+<a name="line-20035"></a>}
+<a name="line-20036"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-20037"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-20038"></a>output_arg { name: "z" type: DT_BOOL }
+<a name="line-20039"></a>-}</span>
+<a name="line-20040"></a>
+<a name="line-20041"></a><a name="pow"></a><span class='hs-comment'>-- | Computes the power of one value to another.</span>
+<a name="line-20042"></a><span class='hs-comment'>--</span>
+<a name="line-20043"></a><span class='hs-comment'>-- Given a tensor `x` and a tensor `y`, this operation computes \\(x^y\\) for</span>
+<a name="line-20044"></a><span class='hs-comment'>-- corresponding elements in `x` and `y`. For example:</span>
+<a name="line-20045"></a><span class='hs-comment'>-- </span>
+<a name="line-20046"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20047"></a><span class='hs-comment'>-- # tensor 'x' is [[2, 2]], [3, 3]]</span>
+<a name="line-20048"></a><span class='hs-comment'>-- # tensor 'y' is [[8, 16], [2, 3]]</span>
+<a name="line-20049"></a><span class='hs-comment'>-- tf.pow(x, y) ==&gt; [[256, 65536], [9, 27]]</span>
+<a name="line-20050"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20051"></a><span class='hs-definition'>pow</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-20052"></a>                                               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-20053"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-20054"></a>                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-20055"></a>                                               <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20056"></a>       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-20057"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-20058"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-20059"></a><span class='hs-definition'>pow</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20060"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Pow"</span>
+<a name="line-20061"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20062"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-20063"></a><span class='hs-comment'>{-
+<a name="line-20064"></a>attr {
+<a name="line-20065"></a>  allowed_values {
+<a name="line-20066"></a>    list {
+<a name="line-20067"></a>      type: DT_HALF
+<a name="line-20068"></a>      type: DT_FLOAT
+<a name="line-20069"></a>      type: DT_DOUBLE
+<a name="line-20070"></a>      type: DT_INT32
+<a name="line-20071"></a>      type: DT_INT64
+<a name="line-20072"></a>      type: DT_COMPLEX64
+<a name="line-20073"></a>      type: DT_COMPLEX128
+<a name="line-20074"></a>    }
+<a name="line-20075"></a>  }
+<a name="line-20076"></a>  name: "T"
+<a name="line-20077"></a>  type: "type"
+<a name="line-20078"></a>}
+<a name="line-20079"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-20080"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-20081"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-20082"></a>-}</span>
+<a name="line-20083"></a>
+<a name="line-20084"></a><a name="igammac"></a><span class='hs-comment'>-- | Compute the upper regularized incomplete Gamma function `Q(a, x)`.</span>
+<a name="line-20085"></a><span class='hs-comment'>--</span>
+<a name="line-20086"></a><span class='hs-comment'>-- The upper regularized incomplete Gamma function is defined as:</span>
+<a name="line-20087"></a><span class='hs-comment'>-- </span>
+<a name="line-20088"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20089"></a><span class='hs-comment'>-- Q(a, x) = Gamma(a, x) / Gamma(a) = 1 - P(a, x)</span>
+<a name="line-20090"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20091"></a><span class='hs-comment'>-- where</span>
+<a name="line-20092"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20093"></a><span class='hs-comment'>-- Gamma(a, x) = int_{x}^{\infty} t^{a-1} exp(-t) dt</span>
+<a name="line-20094"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20095"></a><span class='hs-comment'>-- is the upper incomplete Gama function.</span>
+<a name="line-20096"></a><span class='hs-comment'>-- </span>
+<a name="line-20097"></a><span class='hs-comment'>-- Note, above `P(a, x)` (`Igamma`) is the lower regularized complete</span>
+<a name="line-20098"></a><span class='hs-comment'>-- Gamma function.</span>
+<a name="line-20099"></a><span class='hs-definition'>igammac</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20100"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __a__</span>
+<a name="line-20101"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-20102"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-20103"></a><span class='hs-definition'>igammac</span> <span class='hs-varid'>a</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20104"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Igammac"</span>
+<a name="line-20105"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20106"></a>        <span class='hs-varid'>a</span> <span class='hs-varid'>x</span>
+<a name="line-20107"></a><span class='hs-comment'>{-
+<a name="line-20108"></a>attr {
+<a name="line-20109"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-20110"></a>  name: "T"
+<a name="line-20111"></a>  type: "type"
+<a name="line-20112"></a>}
+<a name="line-20113"></a>input_arg { name: "a" type_attr: "T" }
+<a name="line-20114"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-20115"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-20116"></a>-}</span>
+<a name="line-20117"></a>
+<a name="line-20118"></a><a name="igamma"></a><span class='hs-comment'>-- | Compute the lower regularized incomplete Gamma function `Q(a, x)`.</span>
+<a name="line-20119"></a><span class='hs-comment'>--</span>
+<a name="line-20120"></a><span class='hs-comment'>-- The lower regularized incomplete Gamma function is defined as:</span>
+<a name="line-20121"></a><span class='hs-comment'>-- </span>
+<a name="line-20122"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20123"></a><span class='hs-comment'>-- P(a, x) = gamma(a, x) / Gamma(a) = 1 - Q(a, x)</span>
+<a name="line-20124"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20125"></a><span class='hs-comment'>-- where</span>
+<a name="line-20126"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20127"></a><span class='hs-comment'>-- gamma(a, x) = int_{0}^{x} t^{a-1} exp(-t) dt</span>
+<a name="line-20128"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20129"></a><span class='hs-comment'>-- is the lower incomplete Gamma function.</span>
+<a name="line-20130"></a><span class='hs-comment'>-- </span>
+<a name="line-20131"></a><span class='hs-comment'>-- Note, above `Q(a, x)` (`Igammac`) is the upper regularized complete</span>
+<a name="line-20132"></a><span class='hs-comment'>-- Gamma function.</span>
+<a name="line-20133"></a><span class='hs-definition'>igamma</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20134"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __a__</span>
+<a name="line-20135"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-20136"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-20137"></a><span class='hs-definition'>igamma</span> <span class='hs-varid'>a</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20138"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Igamma"</span>
+<a name="line-20139"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20140"></a>        <span class='hs-varid'>a</span> <span class='hs-varid'>x</span>
+<a name="line-20141"></a><span class='hs-comment'>{-
+<a name="line-20142"></a>attr {
+<a name="line-20143"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-20144"></a>  name: "T"
+<a name="line-20145"></a>  type: "type"
+<a name="line-20146"></a>}
+<a name="line-20147"></a>input_arg { name: "a" type_attr: "T" }
+<a name="line-20148"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-20149"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-20150"></a>-}</span>
+<a name="line-20151"></a>
+<a name="line-20152"></a><a name="zeta"></a><span class='hs-comment'>-- | Compute the Hurwitz zeta function \\(\zeta(x, q)\\).</span>
+<a name="line-20153"></a><span class='hs-comment'>--</span>
+<a name="line-20154"></a><span class='hs-comment'>-- The Hurwitz zeta function is defined as:</span>
+<a name="line-20155"></a><span class='hs-comment'>-- </span>
+<a name="line-20156"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20157"></a><span class='hs-comment'>-- \zeta(x, q) = \sum_{n=0}^{\infty} (q + n)^{-x}</span>
+<a name="line-20158"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20159"></a><span class='hs-definition'>zeta</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20160"></a>        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-20161"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __q__</span>
+<a name="line-20162"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-20163"></a><span class='hs-definition'>zeta</span> <span class='hs-varid'>x</span> <span class='hs-varid'>q</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20164"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Zeta"</span>
+<a name="line-20165"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20166"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>q</span>
+<a name="line-20167"></a><span class='hs-comment'>{-
+<a name="line-20168"></a>attr {
+<a name="line-20169"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-20170"></a>  name: "T"
+<a name="line-20171"></a>  type: "type"
+<a name="line-20172"></a>}
+<a name="line-20173"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-20174"></a>input_arg { name: "q" type_attr: "T" }
+<a name="line-20175"></a>output_arg { name: "z" type_attr: "T" }
+<a name="line-20176"></a>-}</span>
+<a name="line-20177"></a>
+<a name="line-20178"></a><a name="imag"></a><span class='hs-comment'>-- | Returns the imaginary part of a complex number.</span>
+<a name="line-20179"></a><span class='hs-comment'>--</span>
+<a name="line-20180"></a><span class='hs-comment'>-- Given a tensor `input` of complex numbers, this operation returns a tensor of</span>
+<a name="line-20181"></a><span class='hs-comment'>-- type `float` that is the imaginary part of each element in `input`. All</span>
+<a name="line-20182"></a><span class='hs-comment'>-- elements in `input` must be complex numbers of the form \\(a + bj\\), where *a*</span>
+<a name="line-20183"></a><span class='hs-comment'>-- is the real part and *b* is the imaginary part returned by this operation.</span>
+<a name="line-20184"></a><span class='hs-comment'>-- </span>
+<a name="line-20185"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-20186"></a><span class='hs-comment'>-- </span>
+<a name="line-20187"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20188"></a><span class='hs-comment'>-- # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]</span>
+<a name="line-20189"></a><span class='hs-comment'>-- tf.imag(input) ==&gt; [4.75, 5.75]</span>
+<a name="line-20190"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20191"></a><span class='hs-definition'>imag</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tout</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-20192"></a>                                                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-20193"></a>                            <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tout</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tout</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20194"></a>        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-20195"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>tout</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-20196"></a><span class='hs-definition'>imag</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20197"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Imag"</span>
+<a name="line-20198"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-20199"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tout"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tout</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20200"></a>        <span class='hs-varid'>input</span>
+<a name="line-20201"></a><span class='hs-comment'>{-
+<a name="line-20202"></a>attr {
+<a name="line-20203"></a>  allowed_values { list { type: DT_COMPLEX64 type: DT_COMPLEX128 } }
+<a name="line-20204"></a>  default_value { type: DT_COMPLEX64 }
+<a name="line-20205"></a>  name: "T"
+<a name="line-20206"></a>  type: "type"
+<a name="line-20207"></a>}
+<a name="line-20208"></a>attr {
+<a name="line-20209"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-20210"></a>  default_value { type: DT_FLOAT }
+<a name="line-20211"></a>  name: "Tout"
+<a name="line-20212"></a>  type: "type"
+<a name="line-20213"></a>}
+<a name="line-20214"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-20215"></a>output_arg { name: "output" type_attr: "Tout" }
+<a name="line-20216"></a>-}</span>
+<a name="line-20217"></a>
+<a name="line-20218"></a><a name="complex"></a><span class='hs-comment'>-- | Converts two real numbers to a complex number.</span>
+<a name="line-20219"></a><span class='hs-comment'>--</span>
+<a name="line-20220"></a><span class='hs-comment'>-- Given a tensor `real` representing the real part of a complex number, and a</span>
+<a name="line-20221"></a><span class='hs-comment'>-- tensor `imag` representing the imaginary part of a complex number, this</span>
+<a name="line-20222"></a><span class='hs-comment'>-- operation returns complex numbers elementwise of the form \\(a + bj\\), where</span>
+<a name="line-20223"></a><span class='hs-comment'>-- *a* represents the `real` part and *b* represents the `imag` part.</span>
+<a name="line-20224"></a><span class='hs-comment'>-- </span>
+<a name="line-20225"></a><span class='hs-comment'>-- The input tensors `real` and `imag` must have the same shape.</span>
+<a name="line-20226"></a><span class='hs-comment'>-- </span>
+<a name="line-20227"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-20228"></a><span class='hs-comment'>-- </span>
+<a name="line-20229"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20230"></a><span class='hs-comment'>-- # tensor 'real' is [2.25, 3.25]</span>
+<a name="line-20231"></a><span class='hs-comment'>-- # tensor `imag` is [4.75, 5.75]</span>
+<a name="line-20232"></a><span class='hs-comment'>-- tf.complex(real, imag) ==&gt; [[2.25 + 4.75j], [3.25 + 5.75j]]</span>
+<a name="line-20233"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20234"></a><span class='hs-definition'>complex</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tout</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-20235"></a>                                  <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tout</span><span class='hs-layout'>,</span>
+<a name="line-20236"></a>                                  <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-20237"></a>                                          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tout</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20238"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __real__</span>
+<a name="line-20239"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __imag__</span>
+<a name="line-20240"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>tout</span> <span class='hs-comment'>-- ^ __out__</span>
+<a name="line-20241"></a><span class='hs-definition'>complex</span> <span class='hs-varid'>real</span> <span class='hs-varid'>imag</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20242"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Complex"</span>
+<a name="line-20243"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-20244"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tout"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tout</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20245"></a>        <span class='hs-varid'>real</span> <span class='hs-varid'>imag</span>
+<a name="line-20246"></a><span class='hs-comment'>{-
+<a name="line-20247"></a>attr {
+<a name="line-20248"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-20249"></a>  default_value { type: DT_FLOAT }
+<a name="line-20250"></a>  name: "T"
+<a name="line-20251"></a>  type: "type"
+<a name="line-20252"></a>}
+<a name="line-20253"></a>attr {
+<a name="line-20254"></a>  allowed_values { list { type: DT_COMPLEX64 type: DT_COMPLEX128 } }
+<a name="line-20255"></a>  default_value { type: DT_COMPLEX64 }
+<a name="line-20256"></a>  name: "Tout"
+<a name="line-20257"></a>  type: "type"
+<a name="line-20258"></a>}
+<a name="line-20259"></a>input_arg { name: "real" type_attr: "T" }
+<a name="line-20260"></a>input_arg { name: "imag" type_attr: "T" }
+<a name="line-20261"></a>output_arg { name: "out" type_attr: "Tout" }
+<a name="line-20262"></a>-}</span>
+<a name="line-20263"></a>
+<a name="line-20264"></a><a name="notEqual"></a><span class='hs-comment'>-- | Returns the truth value of (x != y) element-wise.</span>
+<a name="line-20265"></a><span class='hs-comment'>--</span>
+<a name="line-20266"></a><span class='hs-comment'>-- *NOTE*: `NotEqual` supports broadcasting. More about broadcasting</span>
+<a name="line-20267"></a><span class='hs-comment'>-- [here](<a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)">http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)</a></span>
+<a name="line-20268"></a><span class='hs-definition'>notEqual</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-20269"></a>                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-20270"></a>                                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-20271"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>,</span>
+<a name="line-20272"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-20273"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-20274"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-20275"></a>                                      <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-20276"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-20277"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-20278"></a><span class='hs-definition'>notEqual</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20279"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"NotEqual"</span>
+<a name="line-20280"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20281"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-20282"></a><span class='hs-comment'>{-
+<a name="line-20283"></a>attr {
+<a name="line-20284"></a>  allowed_values {
+<a name="line-20285"></a>    list {
+<a name="line-20286"></a>      type: DT_HALF
+<a name="line-20287"></a>      type: DT_FLOAT
+<a name="line-20288"></a>      type: DT_DOUBLE
+<a name="line-20289"></a>      type: DT_UINT8
+<a name="line-20290"></a>      type: DT_INT8
+<a name="line-20291"></a>      type: DT_INT16
+<a name="line-20292"></a>      type: DT_INT32
+<a name="line-20293"></a>      type: DT_INT64
+<a name="line-20294"></a>      type: DT_COMPLEX64
+<a name="line-20295"></a>      type: DT_QUINT8
+<a name="line-20296"></a>      type: DT_QINT8
+<a name="line-20297"></a>      type: DT_QINT32
+<a name="line-20298"></a>      type: DT_STRING
+<a name="line-20299"></a>      type: DT_BOOL
+<a name="line-20300"></a>      type: DT_COMPLEX128
+<a name="line-20301"></a>    }
+<a name="line-20302"></a>  }
+<a name="line-20303"></a>  name: "T"
+<a name="line-20304"></a>  type: "type"
+<a name="line-20305"></a>}
+<a name="line-20306"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-20307"></a>input_arg { name: "y" type_attr: "T" }
+<a name="line-20308"></a>output_arg { name: "z" type: DT_BOOL }
+<a name="line-20309"></a>-}</span>
+<a name="line-20310"></a>
+<a name="line-20311"></a><a name="complexAbs"></a><span class='hs-comment'>-- | Computes the complex absolute value of a tensor.</span>
+<a name="line-20312"></a><span class='hs-comment'>--</span>
+<a name="line-20313"></a><span class='hs-comment'>-- Given a tensor `x` of complex numbers, this operation returns a tensor of type</span>
+<a name="line-20314"></a><span class='hs-comment'>-- `float` or `double` that is the absolute value of each element in `x`. All</span>
+<a name="line-20315"></a><span class='hs-comment'>-- elements in `x` must be complex numbers of the form \\(a + bj\\). The absolute</span>
+<a name="line-20316"></a><span class='hs-comment'>-- value is computed as \\( \sqrt{a^2 + b^2}\\).</span>
+<a name="line-20317"></a><span class='hs-comment'>-- </span>
+<a name="line-20318"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-20319"></a><span class='hs-comment'>-- </span>
+<a name="line-20320"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20321"></a><span class='hs-comment'>-- # tensor 'x' is [[-2.25 + 4.75j], [-3.25 + 5.75j]]</span>
+<a name="line-20322"></a><span class='hs-comment'>-- tf.complex_abs(x) ==&gt; [5.25594902, 6.60492229]</span>
+<a name="line-20323"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20324"></a><span class='hs-definition'>complexAbs</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tout</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-20325"></a>                                  <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-20326"></a>                                          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-20327"></a>                                  <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tout</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-20328"></a>                                                           <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tout</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20329"></a>              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-20330"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>tout</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-20331"></a><span class='hs-definition'>complexAbs</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20332"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ComplexAbs"</span>
+<a name="line-20333"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-20334"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tout"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tout</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20335"></a>        <span class='hs-varid'>x</span>
+<a name="line-20336"></a><span class='hs-comment'>{-
+<a name="line-20337"></a>attr {
+<a name="line-20338"></a>  allowed_values { list { type: DT_COMPLEX64 type: DT_COMPLEX128 } }
+<a name="line-20339"></a>  default_value { type: DT_COMPLEX64 }
+<a name="line-20340"></a>  name: "T"
+<a name="line-20341"></a>  type: "type"
+<a name="line-20342"></a>}
+<a name="line-20343"></a>attr {
+<a name="line-20344"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-20345"></a>  default_value { type: DT_FLOAT }
+<a name="line-20346"></a>  name: "Tout"
+<a name="line-20347"></a>  type: "type"
+<a name="line-20348"></a>}
+<a name="line-20349"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-20350"></a>output_arg { name: "y" type_attr: "Tout" }
+<a name="line-20351"></a>-}</span>
+<a name="line-20352"></a>
+<a name="line-20353"></a><a name="logicalAnd"></a><span class='hs-comment'>-- | Returns the truth value of x AND y element-wise.</span>
+<a name="line-20354"></a><span class='hs-comment'>--</span>
+<a name="line-20355"></a><span class='hs-comment'>-- *NOTE*: `LogicalAnd` supports broadcasting. More about broadcasting</span>
+<a name="line-20356"></a><span class='hs-comment'>-- [here](<a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)">http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)</a></span>
+<a name="line-20357"></a><span class='hs-definition'>logicalAnd</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-20358"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-20359"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __z__</span>
+<a name="line-20360"></a><span class='hs-definition'>logicalAnd</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20361"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"LogicalAnd"</span><span class='hs-layout'>)</span>
+<a name="line-20362"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>y</span>
+<a name="line-20363"></a><span class='hs-comment'>{-
+<a name="line-20364"></a>input_arg { name: "x" type: DT_BOOL }
+<a name="line-20365"></a>input_arg { name: "y" type: DT_BOOL }
+<a name="line-20366"></a>output_arg { name: "z" type: DT_BOOL }
+<a name="line-20367"></a>-}</span>
+<a name="line-20368"></a>
+<a name="line-20369"></a><span class='hs-comment'>-- | </span>
+<a name="line-20370"></a>
+<a name="line-20371"></a><a name="batchFFT"></a><span class='hs-definition'>batchFFT</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-20372"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-20373"></a><span class='hs-definition'>batchFFT</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20374"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchFFT"</span><span class='hs-layout'>)</span>
+<a name="line-20375"></a>        <span class='hs-varid'>input</span>
+<a name="line-20376"></a><span class='hs-comment'>{-
+<a name="line-20377"></a>input_arg { name: "input" type: DT_COMPLEX64 }
+<a name="line-20378"></a>output_arg { name: "output" type: DT_COMPLEX64 }
+<a name="line-20379"></a>-}</span>
+<a name="line-20380"></a>
+<a name="line-20381"></a><a name="select"></a><span class='hs-comment'>-- | Selects elements from `t` or `e`, depending on `condition`.</span>
+<a name="line-20382"></a><span class='hs-comment'>--</span>
+<a name="line-20383"></a><span class='hs-comment'>-- The `t`, and `e` tensors must all have the same shape,</span>
+<a name="line-20384"></a><span class='hs-comment'>-- and the output will also have that shape.  The `condition` tensor</span>
+<a name="line-20385"></a><span class='hs-comment'>-- must be a scalar if `t` and `e` are scalars.  If `t` and `e` are vectors</span>
+<a name="line-20386"></a><span class='hs-comment'>-- or higher rank, then `condition` must be either a vector with size</span>
+<a name="line-20387"></a><span class='hs-comment'>-- matching the first dimension of `t`, or must have the same shape as `t`.</span>
+<a name="line-20388"></a><span class='hs-comment'>-- </span>
+<a name="line-20389"></a><span class='hs-comment'>-- The `condition` tensor acts as a mask that chooses, based on the value at each</span>
+<a name="line-20390"></a><span class='hs-comment'>-- element, whether the corresponding element / row in the output should be</span>
+<a name="line-20391"></a><span class='hs-comment'>-- taken from `t` (if true) or `e` (if false).</span>
+<a name="line-20392"></a><span class='hs-comment'>-- </span>
+<a name="line-20393"></a><span class='hs-comment'>-- If `condition` is a vector and `t` and `e` are higher rank matrices, then</span>
+<a name="line-20394"></a><span class='hs-comment'>-- it chooses which row (outer dimension) to copy from `t` and `e`.</span>
+<a name="line-20395"></a><span class='hs-comment'>-- If `condition` has the same shape as `t` and `e`, then it chooses which</span>
+<a name="line-20396"></a><span class='hs-comment'>-- element to copy from `t` and `e`.</span>
+<a name="line-20397"></a><span class='hs-comment'>-- </span>
+<a name="line-20398"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-20399"></a><span class='hs-comment'>-- </span>
+<a name="line-20400"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-20401"></a><span class='hs-comment'>-- # 'condition' tensor is [[True,  False]</span>
+<a name="line-20402"></a><span class='hs-comment'>-- #                        [False, True]]</span>
+<a name="line-20403"></a><span class='hs-comment'>-- # 't' is [[1, 2],</span>
+<a name="line-20404"></a><span class='hs-comment'>-- #         [3, 4]]</span>
+<a name="line-20405"></a><span class='hs-comment'>-- # 'e' is [[5, 6],</span>
+<a name="line-20406"></a><span class='hs-comment'>-- #         [7, 8]]</span>
+<a name="line-20407"></a><span class='hs-comment'>-- select(condition, t, e) ==&gt; [[1, 6],</span>
+<a name="line-20408"></a><span class='hs-comment'>--                              [7, 4]]</span>
+<a name="line-20409"></a><span class='hs-comment'>-- </span>
+<a name="line-20410"></a><span class='hs-comment'>-- </span>
+<a name="line-20411"></a><span class='hs-comment'>-- # 'condition' tensor is [True, False]</span>
+<a name="line-20412"></a><span class='hs-comment'>-- # 't' is [[1, 2],</span>
+<a name="line-20413"></a><span class='hs-comment'>-- #         [3, 4]]</span>
+<a name="line-20414"></a><span class='hs-comment'>-- # 'e' is [[5, 6],</span>
+<a name="line-20415"></a><span class='hs-comment'>-- #         [7, 8]]</span>
+<a name="line-20416"></a><span class='hs-comment'>-- select(condition, t, e) ==&gt; [[1, 2],</span>
+<a name="line-20417"></a><span class='hs-comment'>--                              [7, 8]]</span>
+<a name="line-20418"></a><span class='hs-comment'>-- </span>
+<a name="line-20419"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20420"></a><span class='hs-definition'>select</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20421"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __condition__</span>
+<a name="line-20422"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __t__: = A `Tensor` which may have the same shape as `condition`.</span>
+<a name="line-20423"></a>                         <span class='hs-comment'>-- If `condition` is rank 1, `t` may have higher rank,</span>
+<a name="line-20424"></a>                         <span class='hs-comment'>-- but its first dimension must match the size of `condition`.</span>
+<a name="line-20425"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __e__: = A `Tensor` with the same type and shape as `t`.</span>
+<a name="line-20426"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: = A `Tensor` with the same type and shape as `t` and `e`.</span>
+<a name="line-20427"></a><span class='hs-definition'>select</span> <span class='hs-varid'>condition</span> <span class='hs-varid'>t</span> <span class='hs-varid'>e</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20428"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Select"</span>
+<a name="line-20429"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20430"></a>        <span class='hs-varid'>condition</span> <span class='hs-varid'>t</span> <span class='hs-varid'>e</span>
+<a name="line-20431"></a><span class='hs-comment'>{-
+<a name="line-20432"></a>attr { name: "T" type: "type" }
+<a name="line-20433"></a>input_arg { name: "condition" type: DT_BOOL }
+<a name="line-20434"></a>input_arg {
+<a name="line-20435"></a>  description: "= A `Tensor` which may have the same shape as `condition`.\nIf `condition` is rank 1, `t` may have higher rank,\nbut its first dimension must match the size of `condition`."
+<a name="line-20436"></a>  name: "t"
+<a name="line-20437"></a>  type_attr: "T"
+<a name="line-20438"></a>}
+<a name="line-20439"></a>input_arg {
+<a name="line-20440"></a>  description: "= A `Tensor` with the same type and shape as `t`."
+<a name="line-20441"></a>  name: "e"
+<a name="line-20442"></a>  type_attr: "T"
+<a name="line-20443"></a>}
+<a name="line-20444"></a>output_arg {
+<a name="line-20445"></a>  description: "= A `Tensor` with the same type and shape as `t` and `e`."
+<a name="line-20446"></a>  name: "output"
+<a name="line-20447"></a>  type_attr: "T"
+<a name="line-20448"></a>}
+<a name="line-20449"></a>-}</span>
+<a name="line-20450"></a>
+<a name="line-20451"></a><a name="matMul"></a><span class='hs-comment'>-- | Multiply the matrix "a" by the matrix "b".</span>
+<a name="line-20452"></a><span class='hs-comment'>--</span>
+<a name="line-20453"></a><span class='hs-comment'>-- The inputs must be two-dimensional matrices and the inner dimension of</span>
+<a name="line-20454"></a><span class='hs-comment'>-- "a" (after being transposed if transpose_a is true) must match the</span>
+<a name="line-20455"></a><span class='hs-comment'>-- outer dimension of "b" (after being transposed if transposed_b is</span>
+<a name="line-20456"></a><span class='hs-comment'>-- true).</span>
+<a name="line-20457"></a><span class='hs-comment'>-- </span>
+<a name="line-20458"></a><span class='hs-comment'>-- *Note*: The default kernel implementation for MatMul on GPUs uses</span>
+<a name="line-20459"></a><span class='hs-comment'>-- cublas.</span>
+<a name="line-20460"></a><span class='hs-definition'>matMul</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-20461"></a>                                                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-20462"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-20463"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-20464"></a>                                                  <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20465"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __a__</span>
+<a name="line-20466"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __b__</span>
+<a name="line-20467"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __product__</span>
+<a name="line-20468"></a><span class='hs-definition'>matMul</span> <span class='hs-varid'>a</span> <span class='hs-varid'>b</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20469"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"MatMul"</span>
+<a name="line-20470"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20471"></a>        <span class='hs-varid'>a</span> <span class='hs-varid'>b</span>
+<a name="line-20472"></a><span class='hs-comment'>{-
+<a name="line-20473"></a>attr {
+<a name="line-20474"></a>  default_value { b: false }
+<a name="line-20475"></a>  description: "If true, \"a\" is transposed before multiplication."
+<a name="line-20476"></a>  name: "transpose_a"
+<a name="line-20477"></a>  type: "bool"
+<a name="line-20478"></a>}
+<a name="line-20479"></a>attr {
+<a name="line-20480"></a>  default_value { b: false }
+<a name="line-20481"></a>  description: "If true, \"b\" is transposed before multiplication."
+<a name="line-20482"></a>  name: "transpose_b"
+<a name="line-20483"></a>  type: "bool"
+<a name="line-20484"></a>}
+<a name="line-20485"></a>attr {
+<a name="line-20486"></a>  allowed_values {
+<a name="line-20487"></a>    list {
+<a name="line-20488"></a>      type: DT_HALF
+<a name="line-20489"></a>      type: DT_FLOAT
+<a name="line-20490"></a>      type: DT_DOUBLE
+<a name="line-20491"></a>      type: DT_INT32
+<a name="line-20492"></a>      type: DT_COMPLEX64
+<a name="line-20493"></a>      type: DT_COMPLEX128
+<a name="line-20494"></a>    }
+<a name="line-20495"></a>  }
+<a name="line-20496"></a>  name: "T"
+<a name="line-20497"></a>  type: "type"
+<a name="line-20498"></a>}
+<a name="line-20499"></a>input_arg { name: "a" type_attr: "T" }
+<a name="line-20500"></a>input_arg { name: "b" type_attr: "T" }
+<a name="line-20501"></a>output_arg { name: "product" type_attr: "T" }
+<a name="line-20502"></a>-}</span>
+<a name="line-20503"></a>
+<a name="line-20504"></a><a name="digamma"></a><span class='hs-comment'>-- | Computes Psi, the derivative of Lgamma (the log of the absolute value of</span>
+<a name="line-20505"></a><span class='hs-comment'>--</span>
+<a name="line-20506"></a><span class='hs-comment'>-- `Gamma(x)`), element-wise.</span>
+<a name="line-20507"></a><span class='hs-definition'>digamma</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-20508"></a>                                                <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20509"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-20510"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-20511"></a><span class='hs-definition'>digamma</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20512"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Digamma"</span>
+<a name="line-20513"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20514"></a>        <span class='hs-varid'>x</span>
+<a name="line-20515"></a><span class='hs-comment'>{-
+<a name="line-20516"></a>attr {
+<a name="line-20517"></a>  allowed_values {
+<a name="line-20518"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-20519"></a>  }
+<a name="line-20520"></a>  name: "T"
+<a name="line-20521"></a>  type: "type"
+<a name="line-20522"></a>}
+<a name="line-20523"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-20524"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-20525"></a>-}</span>
+<a name="line-20526"></a>
+<a name="line-20527"></a><span class='hs-comment'>-- | Computes the gradients of convolution with respect to the filter.</span>
+<a name="line-20528"></a>
+<a name="line-20529"></a><a name="conv2DBackpropFilter"></a><span class='hs-definition'>conv2DBackpropFilter</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-20530"></a>                                             <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-20531"></a>                                                     <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20532"></a>                        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: 4-D with shape `[batch, in_height, in_width, in_channels]`.</span>
+<a name="line-20533"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __filter_sizes__: An integer vector representing the tensor shape of `filter`,</span>
+<a name="line-20534"></a>                                                    <span class='hs-comment'>-- where `filter` is a 4-D</span>
+<a name="line-20535"></a>                                                    <span class='hs-comment'>-- `[filter_height, filter_width, in_channels, out_channels]` tensor.</span>
+<a name="line-20536"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out_backprop__: 4-D with shape `[batch, out_height, out_width, out_channels]`.</span>
+<a name="line-20537"></a>                                       <span class='hs-comment'>-- Gradients w.r.t. the output of the convolution.</span>
+<a name="line-20538"></a>                        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 4-D with shape</span>
+<a name="line-20539"></a>                        <span class='hs-comment'>-- `[filter_height, filter_width, in_channels, out_channels]`.  Gradient w.r.t.</span>
+<a name="line-20540"></a>                        <span class='hs-comment'>-- the `filter` input of the convolution.</span>
+<a name="line-20541"></a><span class='hs-definition'>conv2DBackpropFilter</span> <span class='hs-varid'>input</span> <span class='hs-varid'>filter_sizes</span> <span class='hs-varid'>out_backprop</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20542"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Conv2DBackpropFilter"</span>
+<a name="line-20543"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20544"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>filter_sizes</span> <span class='hs-varid'>out_backprop</span>
+<a name="line-20545"></a><span class='hs-comment'>{-
+<a name="line-20546"></a>attr {
+<a name="line-20547"></a>  allowed_values {
+<a name="line-20548"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-20549"></a>  }
+<a name="line-20550"></a>  name: "T"
+<a name="line-20551"></a>  type: "type"
+<a name="line-20552"></a>}
+<a name="line-20553"></a>attr {
+<a name="line-20554"></a>  description: "The stride of the sliding window for each dimension of the input\nof the convolution. Must be in the same order as the dimension specified with\nformat."
+<a name="line-20555"></a>  name: "strides"
+<a name="line-20556"></a>  type: "list(int)"
+<a name="line-20557"></a>}
+<a name="line-20558"></a>attr {
+<a name="line-20559"></a>  default_value { b: true } name: "use_cudnn_on_gpu" type: "bool"
+<a name="line-20560"></a>}
+<a name="line-20561"></a>attr {
+<a name="line-20562"></a>  allowed_values { list { s: "SAME" s: "VALID" } }
+<a name="line-20563"></a>  description: "The type of padding algorithm to use."
+<a name="line-20564"></a>  name: "padding"
+<a name="line-20565"></a>  type: "string"
+<a name="line-20566"></a>}
+<a name="line-20567"></a>attr {
+<a name="line-20568"></a>  allowed_values { list { s: "NHWC" s: "NCHW" } }
+<a name="line-20569"></a>  default_value { s: "NHWC" }
+<a name="line-20570"></a>  description: "Specify the data format of the input and output data. With the\ndefault format \"NHWC\", the data is stored in the order of:\n    [batch, in_height, in_width, in_channels].\nAlternatively, the format could be \"NCHW\", the data storage order of:\n    [batch, in_channels, in_height, in_width]."
+<a name="line-20571"></a>  name: "data_format"
+<a name="line-20572"></a>  type: "string"
+<a name="line-20573"></a>}
+<a name="line-20574"></a>input_arg {
+<a name="line-20575"></a>  description: "4-D with shape `[batch, in_height, in_width, in_channels]`."
+<a name="line-20576"></a>  name: "input"
+<a name="line-20577"></a>  type_attr: "T"
+<a name="line-20578"></a>}
+<a name="line-20579"></a>input_arg {
+<a name="line-20580"></a>  description: "An integer vector representing the tensor shape of `filter`,\nwhere `filter` is a 4-D\n`[filter_height, filter_width, in_channels, out_channels]` tensor."
+<a name="line-20581"></a>  name: "filter_sizes"
+<a name="line-20582"></a>  type: DT_INT32
+<a name="line-20583"></a>}
+<a name="line-20584"></a>input_arg {
+<a name="line-20585"></a>  description: "4-D with shape `[batch, out_height, out_width, out_channels]`.\nGradients w.r.t. the output of the convolution."
+<a name="line-20586"></a>  name: "out_backprop"
+<a name="line-20587"></a>  type_attr: "T"
+<a name="line-20588"></a>}
+<a name="line-20589"></a>output_arg {
+<a name="line-20590"></a>  description: "4-D with shape\n`[filter_height, filter_width, in_channels, out_channels]`.  Gradient w.r.t.\nthe `filter` input of the convolution."
+<a name="line-20591"></a>  name: "output"
+<a name="line-20592"></a>  type_attr: "T"
+<a name="line-20593"></a>}
+<a name="line-20594"></a>-}</span>
+<a name="line-20595"></a>
+<a name="line-20596"></a><a name="min"></a><span class='hs-comment'>-- | Computes the minimum of elements across dimensions of a tensor.</span>
+<a name="line-20597"></a><span class='hs-comment'>--</span>
+<a name="line-20598"></a><span class='hs-comment'>-- Reduces `input` along the dimensions given in `reduction_indices`. Unless</span>
+<a name="line-20599"></a><span class='hs-comment'>-- `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in</span>
+<a name="line-20600"></a><span class='hs-comment'>-- `reduction_indices`. If `keep_dims` is true, the reduced dimensions are</span>
+<a name="line-20601"></a><span class='hs-comment'>-- retained with length 1.</span>
+<a name="line-20602"></a><span class='hs-definition'>min</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-20603"></a>                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-20604"></a>                                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-20605"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-20606"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-20607"></a>                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-20608"></a>                                      <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span>
+<a name="line-20609"></a>                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20610"></a>       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: The tensor to reduce.</span>
+<a name="line-20611"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __reduction_indices__: The dimensions to reduce.</span>
+<a name="line-20612"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: The reduced tensor.</span>
+<a name="line-20613"></a><span class='hs-definition'>min</span> <span class='hs-varid'>input</span> <span class='hs-varid'>reduction_indices</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20614"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Min"</span>
+<a name="line-20615"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-20616"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20617"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>reduction_indices</span>
+<a name="line-20618"></a><span class='hs-comment'>{-
+<a name="line-20619"></a>attr {
+<a name="line-20620"></a>  default_value { b: false }
+<a name="line-20621"></a>  description: "If true, retain reduced dimensions with length 1."
+<a name="line-20622"></a>  name: "keep_dims"
+<a name="line-20623"></a>  type: "bool"
+<a name="line-20624"></a>}
+<a name="line-20625"></a>attr {
+<a name="line-20626"></a>  allowed_values {
+<a name="line-20627"></a>    list {
+<a name="line-20628"></a>      type: DT_FLOAT
+<a name="line-20629"></a>      type: DT_DOUBLE
+<a name="line-20630"></a>      type: DT_INT64
+<a name="line-20631"></a>      type: DT_INT32
+<a name="line-20632"></a>      type: DT_UINT8
+<a name="line-20633"></a>      type: DT_UINT16
+<a name="line-20634"></a>      type: DT_INT16
+<a name="line-20635"></a>      type: DT_INT8
+<a name="line-20636"></a>      type: DT_COMPLEX64
+<a name="line-20637"></a>      type: DT_COMPLEX128
+<a name="line-20638"></a>      type: DT_QINT8
+<a name="line-20639"></a>      type: DT_QUINT8
+<a name="line-20640"></a>      type: DT_QINT32
+<a name="line-20641"></a>      type: DT_HALF
+<a name="line-20642"></a>    }
+<a name="line-20643"></a>  }
+<a name="line-20644"></a>  name: "T"
+<a name="line-20645"></a>  type: "type"
+<a name="line-20646"></a>}
+<a name="line-20647"></a>attr {
+<a name="line-20648"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-20649"></a>  default_value { type: DT_INT32 }
+<a name="line-20650"></a>  name: "Tidx"
+<a name="line-20651"></a>  type: "type"
+<a name="line-20652"></a>}
+<a name="line-20653"></a>input_arg {
+<a name="line-20654"></a>  description: "The tensor to reduce." name: "input" type_attr: "T"
+<a name="line-20655"></a>}
+<a name="line-20656"></a>input_arg {
+<a name="line-20657"></a>  description: "The dimensions to reduce."
+<a name="line-20658"></a>  name: "reduction_indices"
+<a name="line-20659"></a>  type_attr: "Tidx"
+<a name="line-20660"></a>}
+<a name="line-20661"></a>output_arg {
+<a name="line-20662"></a>  description: "The reduced tensor." name: "output" type_attr: "T"
+<a name="line-20663"></a>}
+<a name="line-20664"></a>-}</span>
+<a name="line-20665"></a>
+<a name="line-20666"></a><span class='hs-comment'>-- | Returns which elements of x are finite.</span>
+<a name="line-20667"></a>
+<a name="line-20668"></a><a name="isFinite"></a><span class='hs-definition'>isFinite</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-20669"></a>                                                 <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20670"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-20671"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-20672"></a><span class='hs-definition'>isFinite</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20673"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"IsFinite"</span>
+<a name="line-20674"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20675"></a>        <span class='hs-varid'>x</span>
+<a name="line-20676"></a><span class='hs-comment'>{-
+<a name="line-20677"></a>attr {
+<a name="line-20678"></a>  allowed_values {
+<a name="line-20679"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-20680"></a>  }
+<a name="line-20681"></a>  name: "T"
+<a name="line-20682"></a>  type: "type"
+<a name="line-20683"></a>}
+<a name="line-20684"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-20685"></a>output_arg { name: "y" type: DT_BOOL }
+<a name="line-20686"></a>-}</span>
+<a name="line-20687"></a>
+<a name="line-20688"></a><span class='hs-comment'>-- | Returns the index with the largest value across dimensions of a tensor.</span>
+<a name="line-20689"></a>
+<a name="line-20690"></a><a name="argMax"></a><span class='hs-definition'>argMax</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-20691"></a>                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-20692"></a>                                         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-20693"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-20694"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-20695"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-20696"></a>                                         <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span>
+<a name="line-20697"></a>                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-20698"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20699"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-20700"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __dimension__: int32, 0 &lt;= dimension &lt; rank(input).  Describes which dimension</span>
+<a name="line-20701"></a>                            <span class='hs-comment'>-- of the input Tensor to reduce across. For vectors, use dimension = 0.</span>
+<a name="line-20702"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-20703"></a><span class='hs-definition'>argMax</span> <span class='hs-varid'>input</span> <span class='hs-varid'>dimension</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20704"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ArgMax"</span>
+<a name="line-20705"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-20706"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20707"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>dimension</span>
+<a name="line-20708"></a><span class='hs-comment'>{-
+<a name="line-20709"></a>attr {
+<a name="line-20710"></a>  allowed_values {
+<a name="line-20711"></a>    list {
+<a name="line-20712"></a>      type: DT_FLOAT
+<a name="line-20713"></a>      type: DT_DOUBLE
+<a name="line-20714"></a>      type: DT_INT64
+<a name="line-20715"></a>      type: DT_INT32
+<a name="line-20716"></a>      type: DT_UINT8
+<a name="line-20717"></a>      type: DT_UINT16
+<a name="line-20718"></a>      type: DT_INT16
+<a name="line-20719"></a>      type: DT_INT8
+<a name="line-20720"></a>      type: DT_COMPLEX64
+<a name="line-20721"></a>      type: DT_COMPLEX128
+<a name="line-20722"></a>      type: DT_QINT8
+<a name="line-20723"></a>      type: DT_QUINT8
+<a name="line-20724"></a>      type: DT_QINT32
+<a name="line-20725"></a>      type: DT_HALF
+<a name="line-20726"></a>    }
+<a name="line-20727"></a>  }
+<a name="line-20728"></a>  name: "T"
+<a name="line-20729"></a>  type: "type"
+<a name="line-20730"></a>}
+<a name="line-20731"></a>attr {
+<a name="line-20732"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-20733"></a>  default_value { type: DT_INT32 }
+<a name="line-20734"></a>  name: "Tidx"
+<a name="line-20735"></a>  type: "type"
+<a name="line-20736"></a>}
+<a name="line-20737"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-20738"></a>input_arg {
+<a name="line-20739"></a>  description: "int32, 0 &lt;= dimension &lt; rank(input).  Describes which dimension\nof the input Tensor to reduce across. For vectors, use dimension = 0."
+<a name="line-20740"></a>  name: "dimension"
+<a name="line-20741"></a>  type_attr: "Tidx"
+<a name="line-20742"></a>}
+<a name="line-20743"></a>output_arg { name: "output" type: DT_INT64 }
+<a name="line-20744"></a>-}</span>
+<a name="line-20745"></a>
+<a name="line-20746"></a><a name="segmentMean"></a><span class='hs-comment'>-- | Computes the mean along segments of a tensor.</span>
+<a name="line-20747"></a><span class='hs-comment'>--</span>
+<a name="line-20748"></a><span class='hs-comment'>-- Read [the section on</span>
+<a name="line-20749"></a><span class='hs-comment'>-- Segmentation](../../api_docs/python/math_ops.md#segmentation) for an explanation</span>
+<a name="line-20750"></a><span class='hs-comment'>-- of segments.</span>
+<a name="line-20751"></a><span class='hs-comment'>-- </span>
+<a name="line-20752"></a><span class='hs-comment'>-- Computes a tensor such that</span>
+<a name="line-20753"></a><span class='hs-comment'>-- \\(output_i = \frac{\sum_j data_j}{N}\\) where `mean` is</span>
+<a name="line-20754"></a><span class='hs-comment'>-- over `j` such that `segment_ids[j] == i` and `N` is the total number of</span>
+<a name="line-20755"></a><span class='hs-comment'>-- values summed.</span>
+<a name="line-20756"></a><span class='hs-comment'>-- </span>
+<a name="line-20757"></a><span class='hs-comment'>-- &lt;div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;"&gt;</span>
+<a name="line-20758"></a><span class='hs-comment'>-- &lt;img style="width:100%" src="../../images/SegmentMean.png" alt&gt;</span>
+<a name="line-20759"></a><span class='hs-comment'>-- &lt;/div&gt;</span>
+<a name="line-20760"></a><span class='hs-definition'>segmentMean</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-20761"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-20762"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-20763"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-20764"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-20765"></a>                                                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-20766"></a>                                                                <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-20767"></a>                                                                <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-20768"></a>                                          <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-20769"></a>                                          <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-20770"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20771"></a>               <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__</span>
+<a name="line-20772"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __segment_ids__: A 1-D tensor whose rank is equal to the rank of `data`'s</span>
+<a name="line-20773"></a>                                     <span class='hs-comment'>-- first dimension.  Values should be sorted and can be repeated.</span>
+<a name="line-20774"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Has same shape as data, except for dimension 0 which</span>
+<a name="line-20775"></a>               <span class='hs-comment'>-- has size `k`, the number of segments.</span>
+<a name="line-20776"></a><span class='hs-definition'>segmentMean</span> <span class='hs-varid'>data'</span> <span class='hs-varid'>segment_ids</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20777"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SegmentMean"</span>
+<a name="line-20778"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-20779"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20780"></a>        <span class='hs-varid'>data'</span> <span class='hs-varid'>segment_ids</span>
+<a name="line-20781"></a><span class='hs-comment'>{-
+<a name="line-20782"></a>attr {
+<a name="line-20783"></a>  allowed_values {
+<a name="line-20784"></a>    list {
+<a name="line-20785"></a>      type: DT_FLOAT
+<a name="line-20786"></a>      type: DT_DOUBLE
+<a name="line-20787"></a>      type: DT_INT32
+<a name="line-20788"></a>      type: DT_INT64
+<a name="line-20789"></a>      type: DT_UINT8
+<a name="line-20790"></a>      type: DT_INT16
+<a name="line-20791"></a>      type: DT_INT8
+<a name="line-20792"></a>      type: DT_UINT16
+<a name="line-20793"></a>      type: DT_HALF
+<a name="line-20794"></a>    }
+<a name="line-20795"></a>  }
+<a name="line-20796"></a>  name: "T"
+<a name="line-20797"></a>  type: "type"
+<a name="line-20798"></a>}
+<a name="line-20799"></a>attr {
+<a name="line-20800"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-20801"></a>  name: "Tindices"
+<a name="line-20802"></a>  type: "type"
+<a name="line-20803"></a>}
+<a name="line-20804"></a>input_arg { name: "data" type_attr: "T" }
+<a name="line-20805"></a>input_arg {
+<a name="line-20806"></a>  description: "A 1-D tensor whose rank is equal to the rank of `data`\'s\nfirst dimension.  Values should be sorted and can be repeated."
+<a name="line-20807"></a>  name: "segment_ids"
+<a name="line-20808"></a>  type_attr: "Tindices"
+<a name="line-20809"></a>}
+<a name="line-20810"></a>output_arg {
+<a name="line-20811"></a>  description: "Has same shape as data, except for dimension 0 which\nhas size `k`, the number of segments."
+<a name="line-20812"></a>  name: "output"
+<a name="line-20813"></a>  type_attr: "T"
+<a name="line-20814"></a>}
+<a name="line-20815"></a>-}</span>
+<a name="line-20816"></a>
+<a name="line-20817"></a><a name="cumprod"></a><span class='hs-comment'>-- | Compute the cumulative product of the tensor `x` along `axis`.</span>
+<a name="line-20818"></a><span class='hs-comment'>--</span>
+<a name="line-20819"></a><span class='hs-comment'>-- By default, this op performs an inclusive cumprod, which means that the first</span>
+<a name="line-20820"></a><span class='hs-comment'>-- element of the input is identical to the first element of the output:</span>
+<a name="line-20821"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-20822"></a><span class='hs-comment'>-- tf.cumprod([a, b, c]) ==&gt; [a, a * b, a * b * c]</span>
+<a name="line-20823"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20824"></a><span class='hs-comment'>-- </span>
+<a name="line-20825"></a><span class='hs-comment'>-- By setting the `exclusive` kwarg to `True`, an exclusive cumprod is</span>
+<a name="line-20826"></a><span class='hs-comment'>-- performed instead:</span>
+<a name="line-20827"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-20828"></a><span class='hs-comment'>-- tf.cumprod([a, b, c], exclusive=True) ==&gt; [0, a, a * b]</span>
+<a name="line-20829"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20830"></a><span class='hs-comment'>-- </span>
+<a name="line-20831"></a><span class='hs-comment'>-- By setting the `reverse` kwarg to `True`, the cumprod is performed in the</span>
+<a name="line-20832"></a><span class='hs-comment'>-- opposite direction:</span>
+<a name="line-20833"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-20834"></a><span class='hs-comment'>-- tf.cumprod([a, b, c], reverse=True) ==&gt; [a * b * c, b * c, c]</span>
+<a name="line-20835"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20836"></a><span class='hs-comment'>-- This is more efficient than using separate `tf.reverse` ops.</span>
+<a name="line-20837"></a><span class='hs-comment'>-- </span>
+<a name="line-20838"></a><span class='hs-comment'>-- The `reverse` and `exclusive` kwargs can also be combined:</span>
+<a name="line-20839"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-20840"></a><span class='hs-comment'>-- tf.cumprod([a, b, c], exclusive=True, reverse=True) ==&gt; [b * c, c, 0]</span>
+<a name="line-20841"></a><span class='hs-comment'>-- ```</span>
+<a name="line-20842"></a><span class='hs-definition'>cumprod</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-20843"></a>                                  <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-20844"></a>                                          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-20845"></a>                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-20846"></a>                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-20847"></a>                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-20848"></a>                                          <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span>
+<a name="line-20849"></a>                                  <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-20850"></a>                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20851"></a>           <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-20852"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __axis__</span>
+<a name="line-20853"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__</span>
+<a name="line-20854"></a><span class='hs-definition'>cumprod</span> <span class='hs-varid'>x</span> <span class='hs-varid'>axis</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20855"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Cumprod"</span>
+<a name="line-20856"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-20857"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20858"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>axis</span>
+<a name="line-20859"></a><span class='hs-comment'>{-
+<a name="line-20860"></a>attr { default_value { b: false } name: "exclusive" type: "bool" }
+<a name="line-20861"></a>attr { default_value { b: false } name: "reverse" type: "bool" }
+<a name="line-20862"></a>attr {
+<a name="line-20863"></a>  allowed_values {
+<a name="line-20864"></a>    list {
+<a name="line-20865"></a>      type: DT_FLOAT
+<a name="line-20866"></a>      type: DT_DOUBLE
+<a name="line-20867"></a>      type: DT_INT64
+<a name="line-20868"></a>      type: DT_INT32
+<a name="line-20869"></a>      type: DT_UINT8
+<a name="line-20870"></a>      type: DT_UINT16
+<a name="line-20871"></a>      type: DT_INT16
+<a name="line-20872"></a>      type: DT_INT8
+<a name="line-20873"></a>      type: DT_COMPLEX64
+<a name="line-20874"></a>      type: DT_COMPLEX128
+<a name="line-20875"></a>      type: DT_QINT8
+<a name="line-20876"></a>      type: DT_QUINT8
+<a name="line-20877"></a>      type: DT_QINT32
+<a name="line-20878"></a>      type: DT_HALF
+<a name="line-20879"></a>    }
+<a name="line-20880"></a>  }
+<a name="line-20881"></a>  name: "T"
+<a name="line-20882"></a>  type: "type"
+<a name="line-20883"></a>}
+<a name="line-20884"></a>attr {
+<a name="line-20885"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-20886"></a>  default_value { type: DT_INT32 }
+<a name="line-20887"></a>  name: "Tidx"
+<a name="line-20888"></a>  type: "type"
+<a name="line-20889"></a>}
+<a name="line-20890"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-20891"></a>input_arg { name: "axis" type_attr: "Tidx" }
+<a name="line-20892"></a>output_arg { name: "out" type_attr: "T" }
+<a name="line-20893"></a>-}</span>
+<a name="line-20894"></a>
+<a name="line-20895"></a><a name="segmentMin"></a><span class='hs-comment'>-- | Computes the minimum along segments of a tensor.</span>
+<a name="line-20896"></a><span class='hs-comment'>--</span>
+<a name="line-20897"></a><span class='hs-comment'>-- Read [the section on</span>
+<a name="line-20898"></a><span class='hs-comment'>-- Segmentation](../../api_docs/python/math_ops.md#segmentation) for an explanation</span>
+<a name="line-20899"></a><span class='hs-comment'>-- of segments.</span>
+<a name="line-20900"></a><span class='hs-comment'>-- </span>
+<a name="line-20901"></a><span class='hs-comment'>-- Computes a tensor such that</span>
+<a name="line-20902"></a><span class='hs-comment'>-- \\(output_i = \min_j(data_j)\\) where `min` is over `j` such</span>
+<a name="line-20903"></a><span class='hs-comment'>-- that `segment_ids[j] == i`.</span>
+<a name="line-20904"></a><span class='hs-comment'>-- </span>
+<a name="line-20905"></a><span class='hs-comment'>-- &lt;div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;"&gt;</span>
+<a name="line-20906"></a><span class='hs-comment'>-- &lt;img style="width:100%" src="../../images/SegmentMin.png" alt&gt;</span>
+<a name="line-20907"></a><span class='hs-comment'>-- &lt;/div&gt;</span>
+<a name="line-20908"></a><span class='hs-definition'>segmentMin</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-20909"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-20910"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-20911"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-20912"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-20913"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-20914"></a>                                                               <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-20915"></a>                                         <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-20916"></a>                                         <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-20917"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20918"></a>              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__</span>
+<a name="line-20919"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __segment_ids__: A 1-D tensor whose rank is equal to the rank of `data`'s</span>
+<a name="line-20920"></a>                                    <span class='hs-comment'>-- first dimension.  Values should be sorted and can be repeated.</span>
+<a name="line-20921"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Has same shape as data, except for dimension 0 which</span>
+<a name="line-20922"></a>              <span class='hs-comment'>-- has size `k`, the number of segments.</span>
+<a name="line-20923"></a><span class='hs-definition'>segmentMin</span> <span class='hs-varid'>data'</span> <span class='hs-varid'>segment_ids</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-20924"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SegmentMin"</span>
+<a name="line-20925"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-20926"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-20927"></a>        <span class='hs-varid'>data'</span> <span class='hs-varid'>segment_ids</span>
+<a name="line-20928"></a><span class='hs-comment'>{-
+<a name="line-20929"></a>attr {
+<a name="line-20930"></a>  allowed_values {
+<a name="line-20931"></a>    list {
+<a name="line-20932"></a>      type: DT_FLOAT
+<a name="line-20933"></a>      type: DT_DOUBLE
+<a name="line-20934"></a>      type: DT_INT32
+<a name="line-20935"></a>      type: DT_INT64
+<a name="line-20936"></a>      type: DT_UINT8
+<a name="line-20937"></a>      type: DT_INT16
+<a name="line-20938"></a>      type: DT_INT8
+<a name="line-20939"></a>      type: DT_UINT16
+<a name="line-20940"></a>      type: DT_HALF
+<a name="line-20941"></a>    }
+<a name="line-20942"></a>  }
+<a name="line-20943"></a>  name: "T"
+<a name="line-20944"></a>  type: "type"
+<a name="line-20945"></a>}
+<a name="line-20946"></a>attr {
+<a name="line-20947"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-20948"></a>  name: "Tindices"
+<a name="line-20949"></a>  type: "type"
+<a name="line-20950"></a>}
+<a name="line-20951"></a>input_arg { name: "data" type_attr: "T" }
+<a name="line-20952"></a>input_arg {
+<a name="line-20953"></a>  description: "A 1-D tensor whose rank is equal to the rank of `data`\'s\nfirst dimension.  Values should be sorted and can be repeated."
+<a name="line-20954"></a>  name: "segment_ids"
+<a name="line-20955"></a>  type_attr: "Tindices"
+<a name="line-20956"></a>}
+<a name="line-20957"></a>output_arg {
+<a name="line-20958"></a>  description: "Has same shape as data, except for dimension 0 which\nhas size `k`, the number of segments."
+<a name="line-20959"></a>  name: "output"
+<a name="line-20960"></a>  type_attr: "T"
+<a name="line-20961"></a>}
+<a name="line-20962"></a>-}</span>
+<a name="line-20963"></a>
+<a name="line-20964"></a><a name="unsortedSegmentSum"></a><span class='hs-comment'>-- | Computes the sum along segments of a tensor.</span>
+<a name="line-20965"></a><span class='hs-comment'>--</span>
+<a name="line-20966"></a><span class='hs-comment'>-- Read [the section on</span>
+<a name="line-20967"></a><span class='hs-comment'>-- Segmentation](../../api_docs/python/math_ops.md#segmentation) for an explanation</span>
+<a name="line-20968"></a><span class='hs-comment'>-- of segments.</span>
+<a name="line-20969"></a><span class='hs-comment'>-- </span>
+<a name="line-20970"></a><span class='hs-comment'>-- Computes a tensor such that</span>
+<a name="line-20971"></a><span class='hs-comment'>-- `(output[i] = sum_{j...} data[j...]` where the sum is over tuples `j...` such</span>
+<a name="line-20972"></a><span class='hs-comment'>-- that `segment_ids[j...] == i`.  Unlike `SegmentSum`, `segment_ids`</span>
+<a name="line-20973"></a><span class='hs-comment'>-- need not be sorted and need not cover all values in the full</span>
+<a name="line-20974"></a><span class='hs-comment'>-- range of valid values.</span>
+<a name="line-20975"></a><span class='hs-comment'>-- </span>
+<a name="line-20976"></a><span class='hs-comment'>-- If the sum is empty for a given segment ID `i`, `output[i] = 0`.</span>
+<a name="line-20977"></a><span class='hs-comment'>-- </span>
+<a name="line-20978"></a><span class='hs-comment'>-- `num_segments` should equal the number of distinct segment IDs.</span>
+<a name="line-20979"></a><span class='hs-comment'>-- </span>
+<a name="line-20980"></a><span class='hs-comment'>-- &lt;div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;"&gt;</span>
+<a name="line-20981"></a><span class='hs-comment'>-- &lt;img style="width:100%" src="../../images/UnsortedSegmentSum.png" alt&gt;</span>
+<a name="line-20982"></a><span class='hs-comment'>-- &lt;/div&gt;</span>
+<a name="line-20983"></a><span class='hs-definition'>unsortedSegmentSum</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-20984"></a>                                                    <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-20985"></a>                                                            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-20986"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-20987"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-20988"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-20989"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-20990"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-20991"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-20992"></a>                                                            <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-20993"></a>                                                    <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-20994"></a>                                                    <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-20995"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-20996"></a>                      <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__</span>
+<a name="line-20997"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __segment_ids__: A tensor whose shape is a prefix of `data.shape`.</span>
+<a name="line-20998"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __num_segments__</span>
+<a name="line-20999"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Has same shape as data, except for the first `segment_ids.rank`</span>
+<a name="line-21000"></a>                      <span class='hs-comment'>-- dimensions, which are replaced with a single dimension which has size</span>
+<a name="line-21001"></a>                      <span class='hs-comment'>-- `num_segments`.</span>
+<a name="line-21002"></a><span class='hs-definition'>unsortedSegmentSum</span> <span class='hs-varid'>data'</span> <span class='hs-varid'>segment_ids</span> <span class='hs-varid'>num_segments</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21003"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"UnsortedSegmentSum"</span>
+<a name="line-21004"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-21005"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-21006"></a>        <span class='hs-varid'>data'</span> <span class='hs-varid'>segment_ids</span> <span class='hs-varid'>num_segments</span>
+<a name="line-21007"></a><span class='hs-comment'>{-
+<a name="line-21008"></a>attr {
+<a name="line-21009"></a>  allowed_values {
+<a name="line-21010"></a>    list {
+<a name="line-21011"></a>      type: DT_FLOAT
+<a name="line-21012"></a>      type: DT_DOUBLE
+<a name="line-21013"></a>      type: DT_INT64
+<a name="line-21014"></a>      type: DT_INT32
+<a name="line-21015"></a>      type: DT_UINT8
+<a name="line-21016"></a>      type: DT_UINT16
+<a name="line-21017"></a>      type: DT_INT16
+<a name="line-21018"></a>      type: DT_INT8
+<a name="line-21019"></a>      type: DT_COMPLEX64
+<a name="line-21020"></a>      type: DT_COMPLEX128
+<a name="line-21021"></a>      type: DT_QINT8
+<a name="line-21022"></a>      type: DT_QUINT8
+<a name="line-21023"></a>      type: DT_QINT32
+<a name="line-21024"></a>      type: DT_HALF
+<a name="line-21025"></a>    }
+<a name="line-21026"></a>  }
+<a name="line-21027"></a>  name: "T"
+<a name="line-21028"></a>  type: "type"
+<a name="line-21029"></a>}
+<a name="line-21030"></a>attr {
+<a name="line-21031"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-21032"></a>  name: "Tindices"
+<a name="line-21033"></a>  type: "type"
+<a name="line-21034"></a>}
+<a name="line-21035"></a>input_arg { name: "data" type_attr: "T" }
+<a name="line-21036"></a>input_arg {
+<a name="line-21037"></a>  description: "A tensor whose shape is a prefix of `data.shape`."
+<a name="line-21038"></a>  name: "segment_ids"
+<a name="line-21039"></a>  type_attr: "Tindices"
+<a name="line-21040"></a>}
+<a name="line-21041"></a>input_arg { name: "num_segments" type: DT_INT32 }
+<a name="line-21042"></a>output_arg {
+<a name="line-21043"></a>  description: "Has same shape as data, except for the first `segment_ids.rank`\ndimensions, which are replaced with a single dimension which has size\n`num_segments`."
+<a name="line-21044"></a>  name: "output"
+<a name="line-21045"></a>  type_attr: "T"
+<a name="line-21046"></a>}
+<a name="line-21047"></a>-}</span>
+<a name="line-21048"></a>
+<a name="line-21049"></a><span class='hs-comment'>-- | A Reader that outputs the records from a TensorFlow Records file.</span>
+<a name="line-21050"></a>
+<a name="line-21051"></a><a name="tFRecordReader"></a><span class='hs-definition'>tFRecordReader</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __reader_handle__: The handle to reference the Reader.</span>
+<a name="line-21052"></a><span class='hs-definition'>tFRecordReader</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21053"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"TFRecordReader"</span><span class='hs-layout'>)</span>
+<a name="line-21054"></a>        
+<a name="line-21055"></a><span class='hs-comment'>{-
+<a name="line-21056"></a>attr {
+<a name="line-21057"></a>  default_value { s: "" }
+<a name="line-21058"></a>  description: "If non-empty, this reader is placed in the given container.\nOtherwise, a default container is used."
+<a name="line-21059"></a>  name: "container"
+<a name="line-21060"></a>  type: "string"
+<a name="line-21061"></a>}
+<a name="line-21062"></a>attr {
+<a name="line-21063"></a>  default_value { s: "" }
+<a name="line-21064"></a>  description: "If non-empty, this reader is named in the given bucket\nwith this shared_name. Otherwise, the node name is used instead."
+<a name="line-21065"></a>  name: "shared_name"
+<a name="line-21066"></a>  type: "string"
+<a name="line-21067"></a>}
+<a name="line-21068"></a>attr {
+<a name="line-21069"></a>  default_value { s: "" } name: "compression_type" type: "string"
+<a name="line-21070"></a>}
+<a name="line-21071"></a>output_arg {
+<a name="line-21072"></a>  description: "The handle to reference the Reader."
+<a name="line-21073"></a>  is_ref: true
+<a name="line-21074"></a>  name: "reader_handle"
+<a name="line-21075"></a>  type: DT_STRING
+<a name="line-21076"></a>}
+<a name="line-21077"></a>-}</span>
+<a name="line-21078"></a>
+<a name="line-21079"></a><a name="sparseSegmentSum"></a><span class='hs-comment'>-- | Computes the sum along sparse segments of a tensor.</span>
+<a name="line-21080"></a><span class='hs-comment'>--</span>
+<a name="line-21081"></a><span class='hs-comment'>-- Read [the section on</span>
+<a name="line-21082"></a><span class='hs-comment'>-- Segmentation](../../api_docs/python/math_ops.md#segmentation) for an explanation</span>
+<a name="line-21083"></a><span class='hs-comment'>-- of segments.</span>
+<a name="line-21084"></a><span class='hs-comment'>-- </span>
+<a name="line-21085"></a><span class='hs-comment'>-- Like `SegmentSum`, but `segment_ids` can have rank less than `data`'s first</span>
+<a name="line-21086"></a><span class='hs-comment'>-- dimension, selecting a subset of dimension 0, specified by `indices`.</span>
+<a name="line-21087"></a><span class='hs-comment'>-- </span>
+<a name="line-21088"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-21089"></a><span class='hs-comment'>-- </span>
+<a name="line-21090"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-21091"></a><span class='hs-comment'>-- c = tf.constant([[1,2,3,4], [-1,-2,-3,-4], [5,6,7,8]])</span>
+<a name="line-21092"></a><span class='hs-comment'>-- </span>
+<a name="line-21093"></a><span class='hs-comment'>-- # Select two rows, one segment.</span>
+<a name="line-21094"></a><span class='hs-comment'>-- tf.sparse_segment_sum(c, tf.constant([0, 1]), tf.constant([0, 0]))</span>
+<a name="line-21095"></a><span class='hs-comment'>--   ==&gt; [[0 0 0 0]]</span>
+<a name="line-21096"></a><span class='hs-comment'>-- </span>
+<a name="line-21097"></a><span class='hs-comment'>-- # Select two rows, two segment.</span>
+<a name="line-21098"></a><span class='hs-comment'>-- tf.sparse_segment_sum(c, tf.constant([0, 1]), tf.constant([0, 1]))</span>
+<a name="line-21099"></a><span class='hs-comment'>--   ==&gt; [[ 1  2  3  4]</span>
+<a name="line-21100"></a><span class='hs-comment'>--        [-1 -2 -3 -4]]</span>
+<a name="line-21101"></a><span class='hs-comment'>-- </span>
+<a name="line-21102"></a><span class='hs-comment'>-- # Select all rows, two segments.</span>
+<a name="line-21103"></a><span class='hs-comment'>-- tf.sparse_segment_sum(c, tf.constant([0, 1, 2]), tf.constant([0, 0, 1]))</span>
+<a name="line-21104"></a><span class='hs-comment'>--   ==&gt; [[0 0 0 0]</span>
+<a name="line-21105"></a><span class='hs-comment'>--        [5 6 7 8]]</span>
+<a name="line-21106"></a><span class='hs-comment'>-- </span>
+<a name="line-21107"></a><span class='hs-comment'>-- # Which is equivalent to:</span>
+<a name="line-21108"></a><span class='hs-comment'>-- tf.segment_sum(c, tf.constant([0, 0, 1]))</span>
+<a name="line-21109"></a><span class='hs-comment'>-- ```</span>
+<a name="line-21110"></a><span class='hs-definition'>sparseSegmentSum</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-21111"></a>                                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-21112"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-21113"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-21114"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-21115"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-21116"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-21117"></a>                                                      <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span>
+<a name="line-21118"></a>                                              <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-21119"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-21120"></a>                    <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__</span>
+<a name="line-21121"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __indices__: A 1-D tensor. Has same rank as `segment_ids`.</span>
+<a name="line-21122"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __segment_ids__: A 1-D tensor. Values should be sorted and can be repeated.</span>
+<a name="line-21123"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Has same shape as data, except for dimension 0 which</span>
+<a name="line-21124"></a>                    <span class='hs-comment'>-- has size `k`, the number of segments.</span>
+<a name="line-21125"></a><span class='hs-definition'>sparseSegmentSum</span> <span class='hs-varid'>data'</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>segment_ids</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21126"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseSegmentSum"</span>
+<a name="line-21127"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-21128"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-21129"></a>        <span class='hs-varid'>data'</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>segment_ids</span>
+<a name="line-21130"></a><span class='hs-comment'>{-
+<a name="line-21131"></a>attr {
+<a name="line-21132"></a>  allowed_values {
+<a name="line-21133"></a>    list {
+<a name="line-21134"></a>      type: DT_FLOAT
+<a name="line-21135"></a>      type: DT_DOUBLE
+<a name="line-21136"></a>      type: DT_INT32
+<a name="line-21137"></a>      type: DT_INT64
+<a name="line-21138"></a>      type: DT_UINT8
+<a name="line-21139"></a>      type: DT_INT16
+<a name="line-21140"></a>      type: DT_INT8
+<a name="line-21141"></a>      type: DT_UINT16
+<a name="line-21142"></a>      type: DT_HALF
+<a name="line-21143"></a>    }
+<a name="line-21144"></a>  }
+<a name="line-21145"></a>  name: "T"
+<a name="line-21146"></a>  type: "type"
+<a name="line-21147"></a>}
+<a name="line-21148"></a>attr {
+<a name="line-21149"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-21150"></a>  default_value { type: DT_INT32 }
+<a name="line-21151"></a>  name: "Tidx"
+<a name="line-21152"></a>  type: "type"
+<a name="line-21153"></a>}
+<a name="line-21154"></a>input_arg { name: "data" type_attr: "T" }
+<a name="line-21155"></a>input_arg {
+<a name="line-21156"></a>  description: "A 1-D tensor. Has same rank as `segment_ids`."
+<a name="line-21157"></a>  name: "indices"
+<a name="line-21158"></a>  type_attr: "Tidx"
+<a name="line-21159"></a>}
+<a name="line-21160"></a>input_arg {
+<a name="line-21161"></a>  description: "A 1-D tensor. Values should be sorted and can be repeated."
+<a name="line-21162"></a>  name: "segment_ids"
+<a name="line-21163"></a>  type: DT_INT32
+<a name="line-21164"></a>}
+<a name="line-21165"></a>output_arg {
+<a name="line-21166"></a>  description: "Has same shape as data, except for dimension 0 which\nhas size `k`, the number of segments."
+<a name="line-21167"></a>  name: "output"
+<a name="line-21168"></a>  type_attr: "T"
+<a name="line-21169"></a>}
+<a name="line-21170"></a>-}</span>
+<a name="line-21171"></a>
+<a name="line-21172"></a><a name="sparseSegmentSqrtN"></a><span class='hs-comment'>-- | Computes the sum along sparse segments of a tensor divided by the sqrt of N.</span>
+<a name="line-21173"></a><span class='hs-comment'>--</span>
+<a name="line-21174"></a><span class='hs-comment'>-- N is the size of the segment being reduced.</span>
+<a name="line-21175"></a><span class='hs-comment'>-- </span>
+<a name="line-21176"></a><span class='hs-comment'>-- Read [the section on</span>
+<a name="line-21177"></a><span class='hs-comment'>-- Segmentation](../../api_docs/python/math_ops.md#segmentation) for an explanation</span>
+<a name="line-21178"></a><span class='hs-comment'>-- of segments.</span>
+<a name="line-21179"></a><span class='hs-definition'>sparseSegmentSqrtN</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-21180"></a>                                                                      <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-21181"></a>                                                <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span>
+<a name="line-21182"></a>                                                <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-21183"></a>                                                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-21184"></a>                      <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__</span>
+<a name="line-21185"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __indices__: A 1-D tensor. Has same rank as `segment_ids`.</span>
+<a name="line-21186"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __segment_ids__: A 1-D tensor. Values should be sorted and can be repeated.</span>
+<a name="line-21187"></a>                      <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Has same shape as data, except for dimension 0 which</span>
+<a name="line-21188"></a>                      <span class='hs-comment'>-- has size `k`, the number of segments.</span>
+<a name="line-21189"></a><span class='hs-definition'>sparseSegmentSqrtN</span> <span class='hs-varid'>data'</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>segment_ids</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21190"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseSegmentSqrtN"</span>
+<a name="line-21191"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-21192"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-21193"></a>        <span class='hs-varid'>data'</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>segment_ids</span>
+<a name="line-21194"></a><span class='hs-comment'>{-
+<a name="line-21195"></a>attr {
+<a name="line-21196"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-21197"></a>  name: "T"
+<a name="line-21198"></a>  type: "type"
+<a name="line-21199"></a>}
+<a name="line-21200"></a>attr {
+<a name="line-21201"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-21202"></a>  default_value { type: DT_INT32 }
+<a name="line-21203"></a>  name: "Tidx"
+<a name="line-21204"></a>  type: "type"
+<a name="line-21205"></a>}
+<a name="line-21206"></a>input_arg { name: "data" type_attr: "T" }
+<a name="line-21207"></a>input_arg {
+<a name="line-21208"></a>  description: "A 1-D tensor. Has same rank as `segment_ids`."
+<a name="line-21209"></a>  name: "indices"
+<a name="line-21210"></a>  type_attr: "Tidx"
+<a name="line-21211"></a>}
+<a name="line-21212"></a>input_arg {
+<a name="line-21213"></a>  description: "A 1-D tensor. Values should be sorted and can be repeated."
+<a name="line-21214"></a>  name: "segment_ids"
+<a name="line-21215"></a>  type: DT_INT32
+<a name="line-21216"></a>}
+<a name="line-21217"></a>output_arg {
+<a name="line-21218"></a>  description: "Has same shape as data, except for dimension 0 which\nhas size `k`, the number of segments."
+<a name="line-21219"></a>  name: "output"
+<a name="line-21220"></a>  type_attr: "T"
+<a name="line-21221"></a>}
+<a name="line-21222"></a>-}</span>
+<a name="line-21223"></a>
+<a name="line-21224"></a><a name="copyHost"></a><span class='hs-comment'>-- | Copy Host Op.</span>
+<a name="line-21225"></a><span class='hs-comment'>--</span>
+<a name="line-21226"></a><span class='hs-comment'>-- Performs CPU-to-CPU deep-copying of tensor.</span>
+<a name="line-21227"></a><span class='hs-comment'>-- </span>
+<a name="line-21228"></a><span class='hs-comment'>-- Unlike the Copy Op, this op has HostMemory constraint on its input or output.</span>
+<a name="line-21229"></a><span class='hs-definition'>copyHost</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-21230"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__: Input tensor.</span>
+<a name="line-21231"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Output tensor, deep-copied from input.</span>
+<a name="line-21232"></a><span class='hs-definition'>copyHost</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21233"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"CopyHost"</span>
+<a name="line-21234"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-21235"></a>        <span class='hs-varid'>input</span>
+<a name="line-21236"></a><span class='hs-comment'>{-
+<a name="line-21237"></a>attr { name: "T" type: "type" }
+<a name="line-21238"></a>attr {
+<a name="line-21239"></a>  default_value { s: "" }
+<a name="line-21240"></a>  description: "The name of the input tensor."
+<a name="line-21241"></a>  name: "tensor_name"
+<a name="line-21242"></a>  type: "string"
+<a name="line-21243"></a>}
+<a name="line-21244"></a>input_arg {
+<a name="line-21245"></a>  description: "Input tensor." name: "input" type_attr: "T"
+<a name="line-21246"></a>}
+<a name="line-21247"></a>output_arg {
+<a name="line-21248"></a>  description: "Output tensor, deep-copied from input."
+<a name="line-21249"></a>  name: "output"
+<a name="line-21250"></a>  type_attr: "T"
+<a name="line-21251"></a>}
+<a name="line-21252"></a>-}</span>
+<a name="line-21253"></a>
+<a name="line-21254"></a><a name="variable"></a><span class='hs-comment'>-- | Holds state in the form of a tensor that persists across steps.</span>
+<a name="line-21255"></a><span class='hs-comment'>--</span>
+<a name="line-21256"></a><span class='hs-comment'>-- Outputs a ref to the tensor state so it may be read or modified.</span>
+<a name="line-21257"></a><span class='hs-comment'>-- TODO(zhifengc/mrry): Adds a pointer to a more detail document</span>
+<a name="line-21258"></a><span class='hs-comment'>-- about sharing states in tensorflow.</span>
+<a name="line-21259"></a><span class='hs-definition'>variable</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-21260"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>dtype</span> <span class='hs-comment'>-- ^ __ref__: A reference to the variable tensor.</span>
+<a name="line-21261"></a><span class='hs-definition'>variable</span>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21262"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Variable"</span>
+<a name="line-21263"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-21264"></a>        
+<a name="line-21265"></a><span class='hs-comment'>{-
+<a name="line-21266"></a>attr {
+<a name="line-21267"></a>  description: "The shape of the variable tensor."
+<a name="line-21268"></a>  name: "shape"
+<a name="line-21269"></a>  type: "shape"
+<a name="line-21270"></a>}
+<a name="line-21271"></a>attr {
+<a name="line-21272"></a>  description: "The type of elements in the variable tensor."
+<a name="line-21273"></a>  name: "dtype"
+<a name="line-21274"></a>  type: "type"
+<a name="line-21275"></a>}
+<a name="line-21276"></a>attr {
+<a name="line-21277"></a>  default_value { s: "" }
+<a name="line-21278"></a>  description: "If non-empty, this variable is placed in the given container.\nOtherwise, a default container is used."
+<a name="line-21279"></a>  name: "container"
+<a name="line-21280"></a>  type: "string"
+<a name="line-21281"></a>}
+<a name="line-21282"></a>attr {
+<a name="line-21283"></a>  default_value { s: "" }
+<a name="line-21284"></a>  description: "If non-empty, this variable is named in the given bucket\nwith this shared_name. Otherwise, the node name is used instead."
+<a name="line-21285"></a>  name: "shared_name"
+<a name="line-21286"></a>  type: "string"
+<a name="line-21287"></a>}
+<a name="line-21288"></a>output_arg {
+<a name="line-21289"></a>  description: "A reference to the variable tensor."
+<a name="line-21290"></a>  is_ref: true
+<a name="line-21291"></a>  name: "ref"
+<a name="line-21292"></a>  type_attr: "dtype"
+<a name="line-21293"></a>}
+<a name="line-21294"></a>-}</span>
+<a name="line-21295"></a>
+<a name="line-21296"></a><a name="sparseSegmentSqrtNGrad"></a><span class='hs-comment'>-- | Computes gradients for SparseSegmentSqrtN.</span>
+<a name="line-21297"></a><span class='hs-comment'>--</span>
+<a name="line-21298"></a><span class='hs-comment'>-- Returns tensor "output" with same shape as grad, except for dimension 0 whose</span>
+<a name="line-21299"></a><span class='hs-comment'>-- value is output_dim0.</span>
+<a name="line-21300"></a><span class='hs-definition'>sparseSegmentSqrtNGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-21301"></a>                                                       <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-21302"></a>                                                       <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span>
+<a name="line-21303"></a>                                                       <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-21304"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-21305"></a>                          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __grad__: gradient propagated to the SparseSegmentSqrtN op.</span>
+<a name="line-21306"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __indices__: indices passed to the corresponding SparseSegmentSqrtN op.</span>
+<a name="line-21307"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __segment_ids__: segment_ids passed to the corresponding SparseSegmentSqrtN op.</span>
+<a name="line-21308"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __output_dim0__: dimension 0 of "data" passed to SparseSegmentSqrtN op.</span>
+<a name="line-21309"></a>                          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-21310"></a><span class='hs-definition'>sparseSegmentSqrtNGrad</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>segment_ids</span> <span class='hs-varid'>output_dim0</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21311"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseSegmentSqrtNGrad"</span>
+<a name="line-21312"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-21313"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-21314"></a>        <span class='hs-varid'>grad</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>segment_ids</span> <span class='hs-varid'>output_dim0</span>
+<a name="line-21315"></a><span class='hs-comment'>{-
+<a name="line-21316"></a>attr {
+<a name="line-21317"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-21318"></a>  name: "T"
+<a name="line-21319"></a>  type: "type"
+<a name="line-21320"></a>}
+<a name="line-21321"></a>attr {
+<a name="line-21322"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-21323"></a>  default_value { type: DT_INT32 }
+<a name="line-21324"></a>  name: "Tidx"
+<a name="line-21325"></a>  type: "type"
+<a name="line-21326"></a>}
+<a name="line-21327"></a>input_arg {
+<a name="line-21328"></a>  description: "gradient propagated to the SparseSegmentSqrtN op."
+<a name="line-21329"></a>  name: "grad"
+<a name="line-21330"></a>  type_attr: "T"
+<a name="line-21331"></a>}
+<a name="line-21332"></a>input_arg {
+<a name="line-21333"></a>  description: "indices passed to the corresponding SparseSegmentSqrtN op."
+<a name="line-21334"></a>  name: "indices"
+<a name="line-21335"></a>  type_attr: "Tidx"
+<a name="line-21336"></a>}
+<a name="line-21337"></a>input_arg {
+<a name="line-21338"></a>  description: "segment_ids passed to the corresponding SparseSegmentSqrtN op."
+<a name="line-21339"></a>  name: "segment_ids"
+<a name="line-21340"></a>  type: DT_INT32
+<a name="line-21341"></a>}
+<a name="line-21342"></a>input_arg {
+<a name="line-21343"></a>  description: "dimension 0 of \"data\" passed to SparseSegmentSqrtN op."
+<a name="line-21344"></a>  name: "output_dim0"
+<a name="line-21345"></a>  type: DT_INT32
+<a name="line-21346"></a>}
+<a name="line-21347"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-21348"></a>-}</span>
+<a name="line-21349"></a>
+<a name="line-21350"></a><a name="range"></a><span class='hs-comment'>-- | Creates a sequence of integers.</span>
+<a name="line-21351"></a><span class='hs-comment'>--</span>
+<a name="line-21352"></a><span class='hs-comment'>-- This operation creates a sequence of integers that begins at `start` and</span>
+<a name="line-21353"></a><span class='hs-comment'>-- extends by increments of `delta` up to but not including `limit`.</span>
+<a name="line-21354"></a><span class='hs-comment'>-- </span>
+<a name="line-21355"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-21356"></a><span class='hs-comment'>-- </span>
+<a name="line-21357"></a><span class='hs-comment'>-- ```</span>
+<a name="line-21358"></a><span class='hs-comment'>-- # 'start' is 3</span>
+<a name="line-21359"></a><span class='hs-comment'>-- # 'limit' is 18</span>
+<a name="line-21360"></a><span class='hs-comment'>-- # 'delta' is 3</span>
+<a name="line-21361"></a><span class='hs-comment'>-- tf.range(start, limit, delta) ==&gt; [3, 6, 9, 12, 15]</span>
+<a name="line-21362"></a><span class='hs-comment'>-- ```</span>
+<a name="line-21363"></a><span class='hs-definition'>range</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-21364"></a>                                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-21365"></a>         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __start__: 0-D (scalar). First entry in the sequence.</span>
+<a name="line-21366"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __limit__: 0-D (scalar). Upper limit of sequence, exclusive.</span>
+<a name="line-21367"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __delta__: 0-D (scalar). Optional. Default is 1. Number that increments `start`.</span>
+<a name="line-21368"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __output__: 1-D.</span>
+<a name="line-21369"></a><span class='hs-definition'>range</span> <span class='hs-varid'>start</span> <span class='hs-varid'>limit</span> <span class='hs-varid'>delta</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21370"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Range"</span>
+<a name="line-21371"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-21372"></a>        <span class='hs-varid'>start</span> <span class='hs-varid'>limit</span> <span class='hs-varid'>delta</span>
+<a name="line-21373"></a><span class='hs-comment'>{-
+<a name="line-21374"></a>attr {
+<a name="line-21375"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-21376"></a>  default_value { type: DT_INT32 }
+<a name="line-21377"></a>  name: "Tidx"
+<a name="line-21378"></a>  type: "type"
+<a name="line-21379"></a>}
+<a name="line-21380"></a>input_arg {
+<a name="line-21381"></a>  description: "0-D (scalar). First entry in the sequence."
+<a name="line-21382"></a>  name: "start"
+<a name="line-21383"></a>  type_attr: "Tidx"
+<a name="line-21384"></a>}
+<a name="line-21385"></a>input_arg {
+<a name="line-21386"></a>  description: "0-D (scalar). Upper limit of sequence, exclusive."
+<a name="line-21387"></a>  name: "limit"
+<a name="line-21388"></a>  type_attr: "Tidx"
+<a name="line-21389"></a>}
+<a name="line-21390"></a>input_arg {
+<a name="line-21391"></a>  description: "0-D (scalar). Optional. Default is 1. Number that increments `start`."
+<a name="line-21392"></a>  name: "delta"
+<a name="line-21393"></a>  type_attr: "Tidx"
+<a name="line-21394"></a>}
+<a name="line-21395"></a>output_arg { description: "1-D." name: "output" type_attr: "Tidx" }
+<a name="line-21396"></a>-}</span>
+<a name="line-21397"></a>
+<a name="line-21398"></a><a name="any"></a><span class='hs-comment'>-- | Computes the "logical or" of elements across dimensions of a tensor.</span>
+<a name="line-21399"></a><span class='hs-comment'>--</span>
+<a name="line-21400"></a><span class='hs-comment'>-- Reduces `input` along the dimensions given in `reduction_indices`. Unless</span>
+<a name="line-21401"></a><span class='hs-comment'>-- `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in</span>
+<a name="line-21402"></a><span class='hs-comment'>-- `reduction_indices`. If `keep_dims` is true, the reduced dimensions are</span>
+<a name="line-21403"></a><span class='hs-comment'>-- retained with length 1.</span>
+<a name="line-21404"></a><span class='hs-definition'>any</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-21405"></a>                                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-21406"></a>       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __input__: The tensor to reduce.</span>
+<a name="line-21407"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __reduction_indices__: The dimensions to reduce.</span>
+<a name="line-21408"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Bool</span> <span class='hs-comment'>-- ^ __output__: The reduced tensor.</span>
+<a name="line-21409"></a><span class='hs-definition'>any</span> <span class='hs-varid'>input</span> <span class='hs-varid'>reduction_indices</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21410"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Any"</span>
+<a name="line-21411"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-21412"></a>        <span class='hs-varid'>input</span> <span class='hs-varid'>reduction_indices</span>
+<a name="line-21413"></a><span class='hs-comment'>{-
+<a name="line-21414"></a>attr {
+<a name="line-21415"></a>  default_value { b: false }
+<a name="line-21416"></a>  description: "If true, retain reduced dimensions with length 1."
+<a name="line-21417"></a>  name: "keep_dims"
+<a name="line-21418"></a>  type: "bool"
+<a name="line-21419"></a>}
+<a name="line-21420"></a>attr {
+<a name="line-21421"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-21422"></a>  default_value { type: DT_INT32 }
+<a name="line-21423"></a>  name: "Tidx"
+<a name="line-21424"></a>  type: "type"
+<a name="line-21425"></a>}
+<a name="line-21426"></a>input_arg {
+<a name="line-21427"></a>  description: "The tensor to reduce." name: "input" type: DT_BOOL
+<a name="line-21428"></a>}
+<a name="line-21429"></a>input_arg {
+<a name="line-21430"></a>  description: "The dimensions to reduce."
+<a name="line-21431"></a>  name: "reduction_indices"
+<a name="line-21432"></a>  type_attr: "Tidx"
+<a name="line-21433"></a>}
+<a name="line-21434"></a>output_arg {
+<a name="line-21435"></a>  description: "The reduced tensor." name: "output" type: DT_BOOL
+<a name="line-21436"></a>}
+<a name="line-21437"></a>-}</span>
+<a name="line-21438"></a>
+<a name="line-21439"></a><a name="linSpace"></a><span class='hs-comment'>-- | Generates values in an interval.</span>
+<a name="line-21440"></a><span class='hs-comment'>--</span>
+<a name="line-21441"></a><span class='hs-comment'>-- A sequence of `num` evenly-spaced values are generated beginning at `start`.</span>
+<a name="line-21442"></a><span class='hs-comment'>-- If `num &gt; 1`, the values in the sequence increase by `stop - start / num - 1`,</span>
+<a name="line-21443"></a><span class='hs-comment'>-- so that the last one is exactly `stop`.</span>
+<a name="line-21444"></a><span class='hs-comment'>-- </span>
+<a name="line-21445"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-21446"></a><span class='hs-comment'>-- </span>
+<a name="line-21447"></a><span class='hs-comment'>-- ```</span>
+<a name="line-21448"></a><span class='hs-comment'>-- tf.linspace(10.0, 12.0, 3, name="linspace") =&gt; [ 10.0  11.0  12.0]</span>
+<a name="line-21449"></a><span class='hs-comment'>-- ```</span>
+<a name="line-21450"></a><span class='hs-definition'>linSpace</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-21451"></a>                                      <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-21452"></a>                                                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-21453"></a>            <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __start__: First entry in the range.</span>
+<a name="line-21454"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __stop__: Last entry in the range.</span>
+<a name="line-21455"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __num__: Number of values to generate.</span>
+<a name="line-21456"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 1-D. The generated values.</span>
+<a name="line-21457"></a><span class='hs-definition'>linSpace</span> <span class='hs-varid'>start</span> <span class='hs-varid'>stop</span> <span class='hs-varid'>num</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21458"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"LinSpace"</span>
+<a name="line-21459"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-21460"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-21461"></a>        <span class='hs-varid'>start</span> <span class='hs-varid'>stop</span> <span class='hs-varid'>num</span>
+<a name="line-21462"></a><span class='hs-comment'>{-
+<a name="line-21463"></a>attr {
+<a name="line-21464"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-21465"></a>  name: "T"
+<a name="line-21466"></a>  type: "type"
+<a name="line-21467"></a>}
+<a name="line-21468"></a>attr {
+<a name="line-21469"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-21470"></a>  default_value { type: DT_INT32 }
+<a name="line-21471"></a>  name: "Tidx"
+<a name="line-21472"></a>  type: "type"
+<a name="line-21473"></a>}
+<a name="line-21474"></a>input_arg {
+<a name="line-21475"></a>  description: "First entry in the range."
+<a name="line-21476"></a>  name: "start"
+<a name="line-21477"></a>  type_attr: "T"
+<a name="line-21478"></a>}
+<a name="line-21479"></a>input_arg {
+<a name="line-21480"></a>  description: "Last entry in the range." name: "stop" type_attr: "T"
+<a name="line-21481"></a>}
+<a name="line-21482"></a>input_arg {
+<a name="line-21483"></a>  description: "Number of values to generate."
+<a name="line-21484"></a>  name: "num"
+<a name="line-21485"></a>  type_attr: "Tidx"
+<a name="line-21486"></a>}
+<a name="line-21487"></a>output_arg {
+<a name="line-21488"></a>  description: "1-D. The generated values."
+<a name="line-21489"></a>  name: "output"
+<a name="line-21490"></a>  type_attr: "T"
+<a name="line-21491"></a>}
+<a name="line-21492"></a>-}</span>
+<a name="line-21493"></a>
+<a name="line-21494"></a><a name="resizeArea"></a><span class='hs-comment'>-- | Resize `images` to `size` using area interpolation.</span>
+<a name="line-21495"></a><span class='hs-comment'>--</span>
+<a name="line-21496"></a><span class='hs-comment'>-- Input images can be of different types but output images are always float.</span>
+<a name="line-21497"></a><span class='hs-definition'>resizeArea</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-21498"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-21499"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-21500"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-21501"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-21502"></a>                                                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-21503"></a>                                                      <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-21504"></a>              <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __images__: 4-D with shape `[batch, height, width, channels]`.</span>
+<a name="line-21505"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __size__: = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The</span>
+<a name="line-21506"></a>                                          <span class='hs-comment'>-- new size for the images.</span>
+<a name="line-21507"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Float</span> <span class='hs-comment'>-- ^ __resized_images__: 4-D with shape</span>
+<a name="line-21508"></a>              <span class='hs-comment'>-- `[batch, new_height, new_width, channels]`.</span>
+<a name="line-21509"></a><span class='hs-definition'>resizeArea</span> <span class='hs-varid'>images</span> <span class='hs-varid'>size</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21510"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"ResizeArea"</span>
+<a name="line-21511"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-21512"></a>        <span class='hs-varid'>images</span> <span class='hs-varid'>size</span>
+<a name="line-21513"></a><span class='hs-comment'>{-
+<a name="line-21514"></a>attr {
+<a name="line-21515"></a>  allowed_values {
+<a name="line-21516"></a>    list {
+<a name="line-21517"></a>      type: DT_UINT8
+<a name="line-21518"></a>      type: DT_INT8
+<a name="line-21519"></a>      type: DT_INT16
+<a name="line-21520"></a>      type: DT_INT32
+<a name="line-21521"></a>      type: DT_INT64
+<a name="line-21522"></a>      type: DT_HALF
+<a name="line-21523"></a>      type: DT_FLOAT
+<a name="line-21524"></a>      type: DT_DOUBLE
+<a name="line-21525"></a>    }
+<a name="line-21526"></a>  }
+<a name="line-21527"></a>  name: "T"
+<a name="line-21528"></a>  type: "type"
+<a name="line-21529"></a>}
+<a name="line-21530"></a>attr {
+<a name="line-21531"></a>  default_value { b: false }
+<a name="line-21532"></a>  description: "If true, rescale input by (new_height - 1) / (height - 1), which\nexactly aligns the 4 corners of images and resized images. If false, rescale\nby new_height / height. Treat similarly the width dimension."
+<a name="line-21533"></a>  name: "align_corners"
+<a name="line-21534"></a>  type: "bool"
+<a name="line-21535"></a>}
+<a name="line-21536"></a>input_arg {
+<a name="line-21537"></a>  description: "4-D with shape `[batch, height, width, channels]`."
+<a name="line-21538"></a>  name: "images"
+<a name="line-21539"></a>  type_attr: "T"
+<a name="line-21540"></a>}
+<a name="line-21541"></a>input_arg {
+<a name="line-21542"></a>  description: "= A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The\nnew size for the images."
+<a name="line-21543"></a>  name: "size"
+<a name="line-21544"></a>  type: DT_INT32
+<a name="line-21545"></a>}
+<a name="line-21546"></a>output_arg {
+<a name="line-21547"></a>  description: "4-D with shape\n`[batch, new_height, new_width, channels]`."
+<a name="line-21548"></a>  name: "resized_images"
+<a name="line-21549"></a>  type: DT_FLOAT
+<a name="line-21550"></a>}
+<a name="line-21551"></a>-}</span>
+<a name="line-21552"></a>
+<a name="line-21553"></a><a name="real"></a><span class='hs-comment'>-- | Returns the real part of a complex number.</span>
+<a name="line-21554"></a><span class='hs-comment'>--</span>
+<a name="line-21555"></a><span class='hs-comment'>-- Given a tensor `input` of complex numbers, this operation returns a tensor of</span>
+<a name="line-21556"></a><span class='hs-comment'>-- type `float` that is the real part of each element in `input`. All elements in</span>
+<a name="line-21557"></a><span class='hs-comment'>-- `input` must be complex numbers of the form \\(a + bj\\), where *a* is the real</span>
+<a name="line-21558"></a><span class='hs-comment'>--  part returned by this operation and *b* is the imaginary part.</span>
+<a name="line-21559"></a><span class='hs-comment'>-- </span>
+<a name="line-21560"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-21561"></a><span class='hs-comment'>-- </span>
+<a name="line-21562"></a><span class='hs-comment'>-- ```</span>
+<a name="line-21563"></a><span class='hs-comment'>-- # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]</span>
+<a name="line-21564"></a><span class='hs-comment'>-- tf.real(input) ==&gt; [-2.25, 3.25]</span>
+<a name="line-21565"></a><span class='hs-comment'>-- ```</span>
+<a name="line-21566"></a><span class='hs-definition'>real</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tout</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-21567"></a>                                                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-21568"></a>                            <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tout</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tout</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-21569"></a>        <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-21570"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>tout</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-21571"></a><span class='hs-definition'>real</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21572"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Real"</span>
+<a name="line-21573"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-21574"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tout"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tout</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-21575"></a>        <span class='hs-varid'>input</span>
+<a name="line-21576"></a><span class='hs-comment'>{-
+<a name="line-21577"></a>attr {
+<a name="line-21578"></a>  allowed_values { list { type: DT_COMPLEX64 type: DT_COMPLEX128 } }
+<a name="line-21579"></a>  default_value { type: DT_COMPLEX64 }
+<a name="line-21580"></a>  name: "T"
+<a name="line-21581"></a>  type: "type"
+<a name="line-21582"></a>}
+<a name="line-21583"></a>attr {
+<a name="line-21584"></a>  allowed_values { list { type: DT_FLOAT type: DT_DOUBLE } }
+<a name="line-21585"></a>  default_value { type: DT_FLOAT }
+<a name="line-21586"></a>  name: "Tout"
+<a name="line-21587"></a>  type: "type"
+<a name="line-21588"></a>}
+<a name="line-21589"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-21590"></a>output_arg { name: "output" type_attr: "Tout" }
+<a name="line-21591"></a>-}</span>
+<a name="line-21592"></a>
+<a name="line-21593"></a><a name="iFFT"></a><span class='hs-comment'>-- | Compute the inverse 1-dimensional discrete Fourier Transform over the inner-most</span>
+<a name="line-21594"></a><span class='hs-comment'>--</span>
+<a name="line-21595"></a><span class='hs-comment'>-- dimension of `input`.</span>
+<a name="line-21596"></a><span class='hs-definition'>iFFT</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __input__: A complex64 tensor.</span>
+<a name="line-21597"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __output__: A complex64 tensor of the same shape as `input`. The inner-most</span>
+<a name="line-21598"></a>        <span class='hs-comment'>-- dimension of `input` is replaced with its inverse 1D Fourier Transform.</span>
+<a name="line-21599"></a><span class='hs-definition'>iFFT</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21600"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"IFFT"</span><span class='hs-layout'>)</span>
+<a name="line-21601"></a>        <span class='hs-varid'>input</span>
+<a name="line-21602"></a><span class='hs-comment'>{-
+<a name="line-21603"></a>input_arg {
+<a name="line-21604"></a>  description: "A complex64 tensor." name: "input" type: DT_COMPLEX64
+<a name="line-21605"></a>}
+<a name="line-21606"></a>output_arg {
+<a name="line-21607"></a>  description: "A complex64 tensor of the same shape as `input`. The inner-most\ndimension of `input` is replaced with its inverse 1D Fourier Transform."
+<a name="line-21608"></a>  name: "output"
+<a name="line-21609"></a>  type: DT_COMPLEX64
+<a name="line-21610"></a>}
+<a name="line-21611"></a>-}</span>
+<a name="line-21612"></a>
+<a name="line-21613"></a><a name="iFFT3D"></a><span class='hs-comment'>-- | Compute the inverse 3-dimensional discrete Fourier Transform over the inner-most</span>
+<a name="line-21614"></a><span class='hs-comment'>--</span>
+<a name="line-21615"></a><span class='hs-comment'>-- 3 dimensions of `input`.</span>
+<a name="line-21616"></a><span class='hs-definition'>iFFT3D</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __input__: A complex64 tensor.</span>
+<a name="line-21617"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __output__: A complex64 tensor of the same shape as `input`. The inner-most 3</span>
+<a name="line-21618"></a>          <span class='hs-comment'>-- dimensions of `input` are replaced with their inverse 3D Fourier Transform.</span>
+<a name="line-21619"></a><span class='hs-definition'>iFFT3D</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21620"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"IFFT3D"</span><span class='hs-layout'>)</span>
+<a name="line-21621"></a>        <span class='hs-varid'>input</span>
+<a name="line-21622"></a><span class='hs-comment'>{-
+<a name="line-21623"></a>input_arg {
+<a name="line-21624"></a>  description: "A complex64 tensor." name: "input" type: DT_COMPLEX64
+<a name="line-21625"></a>}
+<a name="line-21626"></a>output_arg {
+<a name="line-21627"></a>  description: "A complex64 tensor of the same shape as `input`. The inner-most 3\ndimensions of `input` are replaced with their inverse 3D Fourier Transform."
+<a name="line-21628"></a>  name: "output"
+<a name="line-21629"></a>  type: DT_COMPLEX64
+<a name="line-21630"></a>}
+<a name="line-21631"></a>-}</span>
+<a name="line-21632"></a>
+<a name="line-21633"></a><a name="cross"></a><span class='hs-comment'>-- | Compute the pairwise cross product.</span>
+<a name="line-21634"></a><span class='hs-comment'>--</span>
+<a name="line-21635"></a><span class='hs-comment'>-- `a` and `b` must be the same shape; they can either be simple 3-element vectors,</span>
+<a name="line-21636"></a><span class='hs-comment'>-- or any shape where the innermost dimension is 3. In the latter case, each pair</span>
+<a name="line-21637"></a><span class='hs-comment'>-- of corresponding 3-element vectors is cross-multiplied independently.</span>
+<a name="line-21638"></a><span class='hs-definition'>cross</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-21639"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-21640"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-21641"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-21642"></a>                                                 <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-21643"></a>         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __a__: A tensor containing 3-element vectors.</span>
+<a name="line-21644"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __b__: Another tensor, of same type and shape as `a`.</span>
+<a name="line-21645"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __product__: Pairwise cross product of the vectors in `a` and `b`.</span>
+<a name="line-21646"></a><span class='hs-definition'>cross</span> <span class='hs-varid'>a</span> <span class='hs-varid'>b</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21647"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Cross"</span>
+<a name="line-21648"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-21649"></a>        <span class='hs-varid'>a</span> <span class='hs-varid'>b</span>
+<a name="line-21650"></a><span class='hs-comment'>{-
+<a name="line-21651"></a>attr {
+<a name="line-21652"></a>  allowed_values {
+<a name="line-21653"></a>    list {
+<a name="line-21654"></a>      type: DT_FLOAT
+<a name="line-21655"></a>      type: DT_DOUBLE
+<a name="line-21656"></a>      type: DT_INT32
+<a name="line-21657"></a>      type: DT_INT64
+<a name="line-21658"></a>      type: DT_UINT8
+<a name="line-21659"></a>      type: DT_INT16
+<a name="line-21660"></a>      type: DT_INT8
+<a name="line-21661"></a>      type: DT_UINT16
+<a name="line-21662"></a>      type: DT_HALF
+<a name="line-21663"></a>    }
+<a name="line-21664"></a>  }
+<a name="line-21665"></a>  name: "T"
+<a name="line-21666"></a>  type: "type"
+<a name="line-21667"></a>}
+<a name="line-21668"></a>input_arg {
+<a name="line-21669"></a>  description: "A tensor containing 3-element vectors."
+<a name="line-21670"></a>  name: "a"
+<a name="line-21671"></a>  type_attr: "T"
+<a name="line-21672"></a>}
+<a name="line-21673"></a>input_arg {
+<a name="line-21674"></a>  description: "Another tensor, of same type and shape as `a`."
+<a name="line-21675"></a>  name: "b"
+<a name="line-21676"></a>  type_attr: "T"
+<a name="line-21677"></a>}
+<a name="line-21678"></a>output_arg {
+<a name="line-21679"></a>  description: "Pairwise cross product of the vectors in `a` and `b`."
+<a name="line-21680"></a>  name: "product"
+<a name="line-21681"></a>  type_attr: "T"
+<a name="line-21682"></a>}
+<a name="line-21683"></a>-}</span>
+<a name="line-21684"></a>
+<a name="line-21685"></a><a name="cumsum"></a><span class='hs-comment'>-- | Compute the cumulative sum of the tensor `x` along `axis`.</span>
+<a name="line-21686"></a><span class='hs-comment'>--</span>
+<a name="line-21687"></a><span class='hs-comment'>-- By default, this op performs an inclusive cumsum, which means that the first</span>
+<a name="line-21688"></a><span class='hs-comment'>-- element of the input is identical to the first element of the output:</span>
+<a name="line-21689"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-21690"></a><span class='hs-comment'>-- tf.cumsum([a, b, c]) ==&gt; [a, a + b, a + b + c]</span>
+<a name="line-21691"></a><span class='hs-comment'>-- ```</span>
+<a name="line-21692"></a><span class='hs-comment'>-- </span>
+<a name="line-21693"></a><span class='hs-comment'>-- By setting the `exclusive` kwarg to `True`, an exclusive cumsum is</span>
+<a name="line-21694"></a><span class='hs-comment'>-- performed instead:</span>
+<a name="line-21695"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-21696"></a><span class='hs-comment'>-- tf.cumsum([a, b, c], exclusive=True) ==&gt; [0, a, a + b]</span>
+<a name="line-21697"></a><span class='hs-comment'>-- ```</span>
+<a name="line-21698"></a><span class='hs-comment'>-- </span>
+<a name="line-21699"></a><span class='hs-comment'>-- By setting the `reverse` kwarg to `True`, the cumsum is performed in the</span>
+<a name="line-21700"></a><span class='hs-comment'>-- opposite direction:</span>
+<a name="line-21701"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-21702"></a><span class='hs-comment'>-- tf.cumsum([a, b, c], reverse=True) ==&gt; [a + b + c, b + c, c]</span>
+<a name="line-21703"></a><span class='hs-comment'>-- ```</span>
+<a name="line-21704"></a><span class='hs-comment'>-- This is more efficient than using separate `tf.reverse` ops.</span>
+<a name="line-21705"></a><span class='hs-comment'>-- </span>
+<a name="line-21706"></a><span class='hs-comment'>-- The `reverse` and `exclusive` kwargs can also be combined:</span>
+<a name="line-21707"></a><span class='hs-comment'>-- ```prettyprint</span>
+<a name="line-21708"></a><span class='hs-comment'>-- tf.cumsum([a, b, c], exclusive=True, reverse=True) ==&gt; [b + c, c, 0]</span>
+<a name="line-21709"></a><span class='hs-comment'>-- ```</span>
+<a name="line-21710"></a><span class='hs-definition'>cumsum</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tidx</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-21711"></a>                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-21712"></a>                                         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-21713"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-21714"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-21715"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-21716"></a>                                         <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>,</span>
+<a name="line-21717"></a>                                 <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-21718"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-21719"></a>          <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-21720"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tidx</span> <span class='hs-comment'>-- ^ __axis__</span>
+<a name="line-21721"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __out__</span>
+<a name="line-21722"></a><span class='hs-definition'>cumsum</span> <span class='hs-varid'>x</span> <span class='hs-varid'>axis</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21723"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Cumsum"</span>
+<a name="line-21724"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-21725"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tidx"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tidx</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-21726"></a>        <span class='hs-varid'>x</span> <span class='hs-varid'>axis</span>
+<a name="line-21727"></a><span class='hs-comment'>{-
+<a name="line-21728"></a>attr { default_value { b: false } name: "exclusive" type: "bool" }
+<a name="line-21729"></a>attr { default_value { b: false } name: "reverse" type: "bool" }
+<a name="line-21730"></a>attr {
+<a name="line-21731"></a>  allowed_values {
+<a name="line-21732"></a>    list {
+<a name="line-21733"></a>      type: DT_FLOAT
+<a name="line-21734"></a>      type: DT_DOUBLE
+<a name="line-21735"></a>      type: DT_INT64
+<a name="line-21736"></a>      type: DT_INT32
+<a name="line-21737"></a>      type: DT_UINT8
+<a name="line-21738"></a>      type: DT_UINT16
+<a name="line-21739"></a>      type: DT_INT16
+<a name="line-21740"></a>      type: DT_INT8
+<a name="line-21741"></a>      type: DT_COMPLEX64
+<a name="line-21742"></a>      type: DT_COMPLEX128
+<a name="line-21743"></a>      type: DT_QINT8
+<a name="line-21744"></a>      type: DT_QUINT8
+<a name="line-21745"></a>      type: DT_QINT32
+<a name="line-21746"></a>      type: DT_HALF
+<a name="line-21747"></a>    }
+<a name="line-21748"></a>  }
+<a name="line-21749"></a>  name: "T"
+<a name="line-21750"></a>  type: "type"
+<a name="line-21751"></a>}
+<a name="line-21752"></a>attr {
+<a name="line-21753"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-21754"></a>  default_value { type: DT_INT32 }
+<a name="line-21755"></a>  name: "Tidx"
+<a name="line-21756"></a>  type: "type"
+<a name="line-21757"></a>}
+<a name="line-21758"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-21759"></a>input_arg { name: "axis" type_attr: "Tidx" }
+<a name="line-21760"></a>output_arg { name: "out" type_attr: "T" }
+<a name="line-21761"></a>-}</span>
+<a name="line-21762"></a>
+<a name="line-21763"></a><span class='hs-comment'>-- | </span>
+<a name="line-21764"></a>
+<a name="line-21765"></a><a name="batchIFFT"></a><span class='hs-definition'>batchIFFT</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-21766"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-21767"></a><span class='hs-definition'>batchIFFT</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21768"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchIFFT"</span><span class='hs-layout'>)</span>
+<a name="line-21769"></a>        <span class='hs-varid'>input</span>
+<a name="line-21770"></a><span class='hs-comment'>{-
+<a name="line-21771"></a>input_arg { name: "input" type: DT_COMPLEX64 }
+<a name="line-21772"></a>output_arg { name: "output" type: DT_COMPLEX64 }
+<a name="line-21773"></a>-}</span>
+<a name="line-21774"></a>
+<a name="line-21775"></a><span class='hs-comment'>-- | Computes the Gauss error function of `x` element-wise.</span>
+<a name="line-21776"></a>
+<a name="line-21777"></a><a name="erf"></a><span class='hs-definition'>erf</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-21778"></a>                                            <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-21779"></a>       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-21780"></a><span class='hs-definition'>erf</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21781"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Erf"</span>
+<a name="line-21782"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-21783"></a>        <span class='hs-varid'>x</span>
+<a name="line-21784"></a><span class='hs-comment'>{-
+<a name="line-21785"></a>attr {
+<a name="line-21786"></a>  allowed_values {
+<a name="line-21787"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-21788"></a>  }
+<a name="line-21789"></a>  name: "T"
+<a name="line-21790"></a>  type: "type"
+<a name="line-21791"></a>}
+<a name="line-21792"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-21793"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-21794"></a>-}</span>
+<a name="line-21795"></a>
+<a name="line-21796"></a><a name="barrierInsertMany"></a><span class='hs-comment'>-- | For each key, assigns the respective value to the specified component.</span>
+<a name="line-21797"></a><span class='hs-comment'>--</span>
+<a name="line-21798"></a><span class='hs-comment'>-- If a key is not found in the barrier, this operation will create a new</span>
+<a name="line-21799"></a><span class='hs-comment'>-- incomplete element. If a key is found in the barrier, and the element</span>
+<a name="line-21800"></a><span class='hs-comment'>-- already has a value at component_index, this operation will fail with</span>
+<a name="line-21801"></a><span class='hs-comment'>-- INVALID_ARGUMENT, and leave the barrier in an undefined state.</span>
+<a name="line-21802"></a><span class='hs-definition'>barrierInsertMany</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-21803"></a>                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __component_index__: The component of the barrier elements that is being assigned.</span>
+<a name="line-21804"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __handle__: The handle to a barrier.</span>
+<a name="line-21805"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __keys__: A one-dimensional tensor of keys, with length n.</span>
+<a name="line-21806"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __values__: An any-dimensional tensor of values, which are associated with the</span>
+<a name="line-21807"></a>                                    <span class='hs-comment'>-- respective keys. The 0th dimension must have length n.</span>
+<a name="line-21808"></a>                     <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-21809"></a><span class='hs-definition'>barrierInsertMany</span> <span class='hs-varid'>component_index</span> <span class='hs-varid'>handle</span> <span class='hs-varid'>keys</span> <span class='hs-varid'>values</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21810"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BarrierInsertMany"</span>
+<a name="line-21811"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-21812"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"component_index"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>component_index</span><span class='hs-layout'>)</span>
+<a name="line-21813"></a>        <span class='hs-varid'>handle</span> <span class='hs-varid'>keys</span> <span class='hs-varid'>values</span>
+<a name="line-21814"></a><span class='hs-comment'>{-
+<a name="line-21815"></a>attr { name: "T" type: "type" }
+<a name="line-21816"></a>attr {
+<a name="line-21817"></a>  description: "The component of the barrier elements that is being assigned."
+<a name="line-21818"></a>  name: "component_index"
+<a name="line-21819"></a>  type: "int"
+<a name="line-21820"></a>}
+<a name="line-21821"></a>input_arg {
+<a name="line-21822"></a>  description: "The handle to a barrier."
+<a name="line-21823"></a>  is_ref: true
+<a name="line-21824"></a>  name: "handle"
+<a name="line-21825"></a>  type: DT_STRING
+<a name="line-21826"></a>}
+<a name="line-21827"></a>input_arg {
+<a name="line-21828"></a>  description: "A one-dimensional tensor of keys, with length n."
+<a name="line-21829"></a>  name: "keys"
+<a name="line-21830"></a>  type: DT_STRING
+<a name="line-21831"></a>}
+<a name="line-21832"></a>input_arg {
+<a name="line-21833"></a>  description: "An any-dimensional tensor of values, which are associated with the\nrespective keys. The 0th dimension must have length n."
+<a name="line-21834"></a>  name: "values"
+<a name="line-21835"></a>  type_attr: "T"
+<a name="line-21836"></a>}
+<a name="line-21837"></a>-}</span>
+<a name="line-21838"></a>
+<a name="line-21839"></a><span class='hs-comment'>-- | Returns element-wise largest integer not greater than x.</span>
+<a name="line-21840"></a>
+<a name="line-21841"></a><a name="floor"></a><span class='hs-definition'>floor</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-21842"></a>                                              <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-21843"></a>         <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __x__</span>
+<a name="line-21844"></a>         <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __y__</span>
+<a name="line-21845"></a><span class='hs-definition'>floor</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21846"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Floor"</span>
+<a name="line-21847"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-21848"></a>        <span class='hs-varid'>x</span>
+<a name="line-21849"></a><span class='hs-comment'>{-
+<a name="line-21850"></a>attr {
+<a name="line-21851"></a>  allowed_values {
+<a name="line-21852"></a>    list { type: DT_HALF type: DT_FLOAT type: DT_DOUBLE }
+<a name="line-21853"></a>  }
+<a name="line-21854"></a>  name: "T"
+<a name="line-21855"></a>  type: "type"
+<a name="line-21856"></a>}
+<a name="line-21857"></a>input_arg { name: "x" type_attr: "T" }
+<a name="line-21858"></a>output_arg { name: "y" type_attr: "T" }
+<a name="line-21859"></a>-}</span>
+<a name="line-21860"></a>
+<a name="line-21861"></a><span class='hs-comment'>-- | </span>
+<a name="line-21862"></a>
+<a name="line-21863"></a><a name="batchFFT2D"></a><span class='hs-definition'>batchFFT2D</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-21864"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-21865"></a><span class='hs-definition'>batchFFT2D</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21866"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchFFT2D"</span><span class='hs-layout'>)</span>
+<a name="line-21867"></a>        <span class='hs-varid'>input</span>
+<a name="line-21868"></a><span class='hs-comment'>{-
+<a name="line-21869"></a>input_arg { name: "input" type: DT_COMPLEX64 }
+<a name="line-21870"></a>output_arg { name: "output" type: DT_COMPLEX64 }
+<a name="line-21871"></a>-}</span>
+<a name="line-21872"></a>
+<a name="line-21873"></a><a name="sparseAddGrad"></a><span class='hs-comment'>-- | The gradient operator for the SparseAdd op.</span>
+<a name="line-21874"></a><span class='hs-comment'>--</span>
+<a name="line-21875"></a><span class='hs-comment'>-- The SparseAdd op calculates A + B, where A, B, and the sum are all represented</span>
+<a name="line-21876"></a><span class='hs-comment'>-- as `SparseTensor` objects.  This op takes in the upstream gradient w.r.t.</span>
+<a name="line-21877"></a><span class='hs-comment'>-- non-empty values of the sum, and outputs the gradients w.r.t. the non-empty</span>
+<a name="line-21878"></a><span class='hs-comment'>-- values of A and B.</span>
+<a name="line-21879"></a><span class='hs-definition'>sparseAddGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-21880"></a>                                         <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-21881"></a>                                                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-21882"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-21883"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-21884"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-21885"></a>                                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-21886"></a>                                                 <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-21887"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __backprop_val_grad__: 1-D with shape `[nnz(sum)]`.  The gradient with respect to</span>
+<a name="line-21888"></a>                             <span class='hs-comment'>-- the non-empty values of the sum.</span>
+<a name="line-21889"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __a_indices__: 2-D.  The `indices` of the `SparseTensor` A, size `[nnz(A), ndims]`.</span>
+<a name="line-21890"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __b_indices__: 2-D.  The `indices` of the `SparseTensor` B, size `[nnz(B), ndims]`.</span>
+<a name="line-21891"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __sum_indices__: 2-D.  The `indices` of the sum `SparseTensor`, size</span>
+<a name="line-21892"></a>                                             <span class='hs-comment'>-- `[nnz(sum), ndims]`.</span>
+<a name="line-21893"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-21894"></a>                 <span class='hs-comment'>-- ^ (__a_val_grad__, __b_val_grad__)</span>
+<a name="line-21895"></a>                 <span class='hs-comment'>--</span>
+<a name="line-21896"></a>                 <span class='hs-comment'>-- * __a_val_grad__: 1-D with shape `[nnz(A)]`. The gradient with respect to the</span>
+<a name="line-21897"></a>                 <span class='hs-comment'>-- non-empty values of A.</span>
+<a name="line-21898"></a>                 <span class='hs-comment'>--</span>
+<a name="line-21899"></a>                 <span class='hs-comment'>-- * __b_val_grad__: 1-D with shape `[nnz(B)]`. The gradient with respect to the</span>
+<a name="line-21900"></a>                 <span class='hs-comment'>-- non-empty values of B.</span>
+<a name="line-21901"></a><span class='hs-definition'>sparseAddGrad</span> <span class='hs-varid'>backprop_val_grad</span> <span class='hs-varid'>a_indices</span> <span class='hs-varid'>b_indices</span>
+<a name="line-21902"></a>              <span class='hs-varid'>sum_indices</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-21903"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseAddGrad"</span>
+<a name="line-21904"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-21905"></a>        <span class='hs-varid'>backprop_val_grad</span> <span class='hs-varid'>a_indices</span> <span class='hs-varid'>b_indices</span> <span class='hs-varid'>sum_indices</span>
+<a name="line-21906"></a><span class='hs-comment'>{-
+<a name="line-21907"></a>attr {
+<a name="line-21908"></a>  allowed_values {
+<a name="line-21909"></a>    list {
+<a name="line-21910"></a>      type: DT_FLOAT
+<a name="line-21911"></a>      type: DT_DOUBLE
+<a name="line-21912"></a>      type: DT_INT64
+<a name="line-21913"></a>      type: DT_INT32
+<a name="line-21914"></a>      type: DT_UINT8
+<a name="line-21915"></a>      type: DT_UINT16
+<a name="line-21916"></a>      type: DT_INT16
+<a name="line-21917"></a>      type: DT_INT8
+<a name="line-21918"></a>      type: DT_COMPLEX64
+<a name="line-21919"></a>      type: DT_COMPLEX128
+<a name="line-21920"></a>      type: DT_QINT8
+<a name="line-21921"></a>      type: DT_QUINT8
+<a name="line-21922"></a>      type: DT_QINT32
+<a name="line-21923"></a>      type: DT_HALF
+<a name="line-21924"></a>    }
+<a name="line-21925"></a>  }
+<a name="line-21926"></a>  name: "T"
+<a name="line-21927"></a>  type: "type"
+<a name="line-21928"></a>}
+<a name="line-21929"></a>input_arg {
+<a name="line-21930"></a>  description: "1-D with shape `[nnz(sum)]`.  The gradient with respect to\nthe non-empty values of the sum."
+<a name="line-21931"></a>  name: "backprop_val_grad"
+<a name="line-21932"></a>  type_attr: "T"
+<a name="line-21933"></a>}
+<a name="line-21934"></a>input_arg {
+<a name="line-21935"></a>  description: "2-D.  The `indices` of the `SparseTensor` A, size `[nnz(A), ndims]`."
+<a name="line-21936"></a>  name: "a_indices"
+<a name="line-21937"></a>  type: DT_INT64
+<a name="line-21938"></a>}
+<a name="line-21939"></a>input_arg {
+<a name="line-21940"></a>  description: "2-D.  The `indices` of the `SparseTensor` B, size `[nnz(B), ndims]`."
+<a name="line-21941"></a>  name: "b_indices"
+<a name="line-21942"></a>  type: DT_INT64
+<a name="line-21943"></a>}
+<a name="line-21944"></a>input_arg {
+<a name="line-21945"></a>  description: "2-D.  The `indices` of the sum `SparseTensor`, size\n`[nnz(sum), ndims]`."
+<a name="line-21946"></a>  name: "sum_indices"
+<a name="line-21947"></a>  type: DT_INT64
+<a name="line-21948"></a>}
+<a name="line-21949"></a>output_arg {
+<a name="line-21950"></a>  description: "1-D with shape `[nnz(A)]`. The gradient with respect to the\nnon-empty values of A."
+<a name="line-21951"></a>  name: "a_val_grad"
+<a name="line-21952"></a>  type_attr: "T"
+<a name="line-21953"></a>}
+<a name="line-21954"></a>output_arg {
+<a name="line-21955"></a>  description: "1-D with shape `[nnz(B)]`. The gradient with respect to the\nnon-empty values of B."
+<a name="line-21956"></a>  name: "b_val_grad"
+<a name="line-21957"></a>  type_attr: "T"
+<a name="line-21958"></a>}
+<a name="line-21959"></a>-}</span>
+<a name="line-21960"></a>
+<a name="line-21961"></a><a name="sparseAdd"></a><span class='hs-comment'>-- | Adds two `SparseTensor` objects to produce another `SparseTensor`.</span>
+<a name="line-21962"></a><span class='hs-comment'>--</span>
+<a name="line-21963"></a><span class='hs-comment'>-- The input `SparseTensor` objects' indices are assumed ordered in standard</span>
+<a name="line-21964"></a><span class='hs-comment'>-- lexicographic order.  If this is not the case, before this step run</span>
+<a name="line-21965"></a><span class='hs-comment'>-- `SparseReorder` to restore index ordering.</span>
+<a name="line-21966"></a><span class='hs-comment'>-- </span>
+<a name="line-21967"></a><span class='hs-comment'>-- By default, if two values sum to zero at some index, the output `SparseTensor`</span>
+<a name="line-21968"></a><span class='hs-comment'>-- would still include that particular location in its index, storing a zero in the</span>
+<a name="line-21969"></a><span class='hs-comment'>-- corresponding value slot.  To override this, callers can specify `thresh`,</span>
+<a name="line-21970"></a><span class='hs-comment'>-- indicating that if the sum has a magnitude strictly smaller than `thresh`, its</span>
+<a name="line-21971"></a><span class='hs-comment'>-- corresponding value and index would then not be included.  In particular,</span>
+<a name="line-21972"></a><span class='hs-comment'>-- `thresh == 0` (default) means everything is kept and actual thresholding happens</span>
+<a name="line-21973"></a><span class='hs-comment'>-- only for a positive value.</span>
+<a name="line-21974"></a><span class='hs-comment'>-- </span>
+<a name="line-21975"></a><span class='hs-comment'>-- In the following shapes, `nnz` is the count after taking `thresh` into account.</span>
+<a name="line-21976"></a><span class='hs-definition'>sparseAdd</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>v6</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>t</span> <span class='hs-varid'>treal</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-21977"></a>                                                    <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-21978"></a>                                                            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-21979"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-21980"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-21981"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-21982"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-21983"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-21984"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-21985"></a>                                                            <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-21986"></a>                                                    <span class='hs-conid'>TensorType</span> <span class='hs-varid'>treal</span><span class='hs-layout'>,</span>
+<a name="line-21987"></a>                                                    <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-21988"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-21989"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-21990"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-21991"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-21992"></a>                                                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span>
+<a name="line-21993"></a>                                                            <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-21994"></a>                                                            <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>treal</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-21995"></a>             <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __a_indices__: 2-D.  The `indices` of the first `SparseTensor`, size `[nnz, ndims]` Matrix.</span>
+<a name="line-21996"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __a_values__: 1-D.  The `values` of the first `SparseTensor`, size `[nnz]` Vector.</span>
+<a name="line-21997"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __a_shape__: 1-D.  The `shape` of the first `SparseTensor`, size `[ndims]` Vector.</span>
+<a name="line-21998"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __b_indices__: 2-D.  The `indices` of the second `SparseTensor`, size `[nnz, ndims]` Matrix.</span>
+<a name="line-21999"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v5</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __b_values__: 1-D.  The `values` of the second `SparseTensor`, size `[nnz]` Vector.</span>
+<a name="line-22000"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v6</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __b_shape__: 1-D.  The `shape` of the second `SparseTensor`, size `[ndims]` Vector.</span>
+<a name="line-22001"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v7</span> <span class='hs-varid'>treal</span> <span class='hs-comment'>-- ^ __thresh__: 0-D.  The magnitude threshold that determines if an output value/index</span>
+<a name="line-22002"></a>                                <span class='hs-comment'>-- pair takes space.</span>
+<a name="line-22003"></a>             <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-22004"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-22005"></a>             <span class='hs-comment'>-- ^ (__sum_indices__, __sum_values__, __sum_shape__)</span>
+<a name="line-22006"></a>             <span class='hs-comment'>--</span>
+<a name="line-22007"></a>             <span class='hs-comment'>-- * __sum_indices__</span>
+<a name="line-22008"></a>             <span class='hs-comment'>--</span>
+<a name="line-22009"></a>             <span class='hs-comment'>-- * __sum_values__</span>
+<a name="line-22010"></a>             <span class='hs-comment'>--</span>
+<a name="line-22011"></a>             <span class='hs-comment'>-- * __sum_shape__</span>
+<a name="line-22012"></a><span class='hs-definition'>sparseAdd</span> <span class='hs-varid'>a_indices</span> <span class='hs-varid'>a_values</span> <span class='hs-varid'>a_shape</span> <span class='hs-varid'>b_indices</span> <span class='hs-varid'>b_values</span> <span class='hs-varid'>b_shape</span>
+<a name="line-22013"></a>          <span class='hs-varid'>thresh</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-22014"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseAdd"</span>
+<a name="line-22015"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-22016"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Treal"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>treal</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-22017"></a>        <span class='hs-varid'>a_indices</span> <span class='hs-varid'>a_values</span> <span class='hs-varid'>a_shape</span> <span class='hs-varid'>b_indices</span> <span class='hs-varid'>b_values</span> <span class='hs-varid'>b_shape</span> <span class='hs-varid'>thresh</span>
+<a name="line-22018"></a><span class='hs-comment'>{-
+<a name="line-22019"></a>attr {
+<a name="line-22020"></a>  allowed_values {
+<a name="line-22021"></a>    list {
+<a name="line-22022"></a>      type: DT_FLOAT
+<a name="line-22023"></a>      type: DT_DOUBLE
+<a name="line-22024"></a>      type: DT_INT64
+<a name="line-22025"></a>      type: DT_INT32
+<a name="line-22026"></a>      type: DT_UINT8
+<a name="line-22027"></a>      type: DT_UINT16
+<a name="line-22028"></a>      type: DT_INT16
+<a name="line-22029"></a>      type: DT_INT8
+<a name="line-22030"></a>      type: DT_COMPLEX64
+<a name="line-22031"></a>      type: DT_COMPLEX128
+<a name="line-22032"></a>      type: DT_QINT8
+<a name="line-22033"></a>      type: DT_QUINT8
+<a name="line-22034"></a>      type: DT_QINT32
+<a name="line-22035"></a>      type: DT_HALF
+<a name="line-22036"></a>    }
+<a name="line-22037"></a>  }
+<a name="line-22038"></a>  name: "T"
+<a name="line-22039"></a>  type: "type"
+<a name="line-22040"></a>}
+<a name="line-22041"></a>attr {
+<a name="line-22042"></a>  allowed_values {
+<a name="line-22043"></a>    list {
+<a name="line-22044"></a>      type: DT_FLOAT
+<a name="line-22045"></a>      type: DT_DOUBLE
+<a name="line-22046"></a>      type: DT_INT32
+<a name="line-22047"></a>      type: DT_INT64
+<a name="line-22048"></a>      type: DT_UINT8
+<a name="line-22049"></a>      type: DT_INT16
+<a name="line-22050"></a>      type: DT_INT8
+<a name="line-22051"></a>      type: DT_UINT16
+<a name="line-22052"></a>      type: DT_HALF
+<a name="line-22053"></a>    }
+<a name="line-22054"></a>  }
+<a name="line-22055"></a>  name: "Treal"
+<a name="line-22056"></a>  type: "type"
+<a name="line-22057"></a>}
+<a name="line-22058"></a>input_arg {
+<a name="line-22059"></a>  description: "2-D.  The `indices` of the first `SparseTensor`, size `[nnz, ndims]` Matrix."
+<a name="line-22060"></a>  name: "a_indices"
+<a name="line-22061"></a>  type: DT_INT64
+<a name="line-22062"></a>}
+<a name="line-22063"></a>input_arg {
+<a name="line-22064"></a>  description: "1-D.  The `values` of the first `SparseTensor`, size `[nnz]` Vector."
+<a name="line-22065"></a>  name: "a_values"
+<a name="line-22066"></a>  type_attr: "T"
+<a name="line-22067"></a>}
+<a name="line-22068"></a>input_arg {
+<a name="line-22069"></a>  description: "1-D.  The `shape` of the first `SparseTensor`, size `[ndims]` Vector."
+<a name="line-22070"></a>  name: "a_shape"
+<a name="line-22071"></a>  type: DT_INT64
+<a name="line-22072"></a>}
+<a name="line-22073"></a>input_arg {
+<a name="line-22074"></a>  description: "2-D.  The `indices` of the second `SparseTensor`, size `[nnz, ndims]` Matrix."
+<a name="line-22075"></a>  name: "b_indices"
+<a name="line-22076"></a>  type: DT_INT64
+<a name="line-22077"></a>}
+<a name="line-22078"></a>input_arg {
+<a name="line-22079"></a>  description: "1-D.  The `values` of the second `SparseTensor`, size `[nnz]` Vector."
+<a name="line-22080"></a>  name: "b_values"
+<a name="line-22081"></a>  type_attr: "T"
+<a name="line-22082"></a>}
+<a name="line-22083"></a>input_arg {
+<a name="line-22084"></a>  description: "1-D.  The `shape` of the second `SparseTensor`, size `[ndims]` Vector."
+<a name="line-22085"></a>  name: "b_shape"
+<a name="line-22086"></a>  type: DT_INT64
+<a name="line-22087"></a>}
+<a name="line-22088"></a>input_arg {
+<a name="line-22089"></a>  description: "0-D.  The magnitude threshold that determines if an output value/index\npair takes space."
+<a name="line-22090"></a>  name: "thresh"
+<a name="line-22091"></a>  type_attr: "Treal"
+<a name="line-22092"></a>}
+<a name="line-22093"></a>output_arg { name: "sum_indices" type: DT_INT64 }
+<a name="line-22094"></a>output_arg { name: "sum_values" type_attr: "T" }
+<a name="line-22095"></a>output_arg { name: "sum_shape" type: DT_INT64 }
+<a name="line-22096"></a>-}</span>
+<a name="line-22097"></a>
+<a name="line-22098"></a><span class='hs-comment'>-- | </span>
+<a name="line-22099"></a>
+<a name="line-22100"></a><a name="batchCholesky"></a><span class='hs-definition'>batchCholesky</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-22101"></a>                 <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __input__</span>
+<a name="line-22102"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__</span>
+<a name="line-22103"></a><span class='hs-definition'>batchCholesky</span> <span class='hs-varid'>input</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-22104"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"BatchCholesky"</span>
+<a name="line-22105"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-22106"></a>        <span class='hs-varid'>input</span>
+<a name="line-22107"></a><span class='hs-comment'>{-
+<a name="line-22108"></a>attr {
+<a name="line-22109"></a>  allowed_values { list { type: DT_DOUBLE type: DT_FLOAT } }
+<a name="line-22110"></a>  name: "T"
+<a name="line-22111"></a>  type: "type"
+<a name="line-22112"></a>}
+<a name="line-22113"></a>input_arg { name: "input" type_attr: "T" }
+<a name="line-22114"></a>output_arg { name: "output" type_attr: "T" }
+<a name="line-22115"></a>-}</span>
+<a name="line-22116"></a>
+<a name="line-22117"></a><a name="dynamicPartition"></a><span class='hs-comment'>-- | Partitions `data` into `num_partitions` tensors using indices from `partitions`.</span>
+<a name="line-22118"></a><span class='hs-comment'>--</span>
+<a name="line-22119"></a><span class='hs-comment'>-- For each index tuple `js` of size `partitions.ndim`, the slice `data[js, ...]`</span>
+<a name="line-22120"></a><span class='hs-comment'>-- becomes part of `outputs[partitions[js]]`.  The slices with `partitions[js] = i`</span>
+<a name="line-22121"></a><span class='hs-comment'>-- are placed in `outputs[i]` in lexicographic order of `js`, and the first</span>
+<a name="line-22122"></a><span class='hs-comment'>-- dimension of `outputs[i]` is the number of entries in `partitions` equal to `i`.</span>
+<a name="line-22123"></a><span class='hs-comment'>-- In detail,</span>
+<a name="line-22124"></a><span class='hs-comment'>-- </span>
+<a name="line-22125"></a><span class='hs-comment'>--     outputs[i].shape = [sum(partitions == i)] + data.shape[partitions.ndim:]</span>
+<a name="line-22126"></a><span class='hs-comment'>-- </span>
+<a name="line-22127"></a><span class='hs-comment'>--     outputs[i] = pack([data[js, ...] for js if partitions[js] == i])</span>
+<a name="line-22128"></a><span class='hs-comment'>-- </span>
+<a name="line-22129"></a><span class='hs-comment'>-- `data.shape` must start with `partitions.shape`.</span>
+<a name="line-22130"></a><span class='hs-comment'>-- </span>
+<a name="line-22131"></a><span class='hs-comment'>-- For example:</span>
+<a name="line-22132"></a><span class='hs-comment'>-- </span>
+<a name="line-22133"></a><span class='hs-comment'>--     # Scalar partitions</span>
+<a name="line-22134"></a><span class='hs-comment'>--     partitions = 1</span>
+<a name="line-22135"></a><span class='hs-comment'>--     num_partitions = 2</span>
+<a name="line-22136"></a><span class='hs-comment'>--     data = [10, 20]</span>
+<a name="line-22137"></a><span class='hs-comment'>--     outputs[0] = []  # Empty with shape [0, 2]</span>
+<a name="line-22138"></a><span class='hs-comment'>--     outputs[1] = [[10, 20]]</span>
+<a name="line-22139"></a><span class='hs-comment'>-- </span>
+<a name="line-22140"></a><span class='hs-comment'>--     # Vector partitions</span>
+<a name="line-22141"></a><span class='hs-comment'>--     partitions = [0, 0, 1, 1, 0]</span>
+<a name="line-22142"></a><span class='hs-comment'>--     num_partitions = 2</span>
+<a name="line-22143"></a><span class='hs-comment'>--     data = [10, 20, 30, 40, 50]</span>
+<a name="line-22144"></a><span class='hs-comment'>--     outputs[0] = [10, 20, 50]</span>
+<a name="line-22145"></a><span class='hs-comment'>--     outputs[1] = [30, 40]</span>
+<a name="line-22146"></a><span class='hs-comment'>-- </span>
+<a name="line-22147"></a><span class='hs-comment'>-- &lt;div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;"&gt;</span>
+<a name="line-22148"></a><span class='hs-comment'>-- &lt;img style="width:100%" src="../../images/DynamicPartition.png" alt&gt;</span>
+<a name="line-22149"></a><span class='hs-comment'>-- &lt;/div&gt;</span>
+<a name="line-22150"></a><span class='hs-definition'>dynamicPartition</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-22151"></a>                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __num_partitions__: The number of partitions to output.</span>
+<a name="line-22152"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__</span>
+<a name="line-22153"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span> <span class='hs-comment'>-- ^ __partitions__: Any shape.  Indices in the range `[0, num_partitions)`.</span>
+<a name="line-22154"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __outputs__</span>
+<a name="line-22155"></a><span class='hs-definition'>dynamicPartition</span> <span class='hs-varid'>num_partitions</span> <span class='hs-varid'>data'</span> <span class='hs-varid'>partitions</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-22156"></a>    <span class='hs-varid'>buildListOp</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>num_partitions</span><span class='hs-keyglyph'>]</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"DynamicPartition"</span>
+<a name="line-22157"></a>                                  <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-22158"></a>                                  <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"num_partitions"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>num_partitions</span><span class='hs-layout'>)</span>
+<a name="line-22159"></a>        <span class='hs-varid'>data'</span> <span class='hs-varid'>partitions</span>
+<a name="line-22160"></a><span class='hs-comment'>{-
+<a name="line-22161"></a>attr {
+<a name="line-22162"></a>  description: "The number of partitions to output."
+<a name="line-22163"></a>  has_minimum: true
+<a name="line-22164"></a>  minimum: 1
+<a name="line-22165"></a>  name: "num_partitions"
+<a name="line-22166"></a>  type: "int"
+<a name="line-22167"></a>}
+<a name="line-22168"></a>attr { name: "T" type: "type" }
+<a name="line-22169"></a>input_arg { name: "data" type_attr: "T" }
+<a name="line-22170"></a>input_arg {
+<a name="line-22171"></a>  description: "Any shape.  Indices in the range `[0, num_partitions)`."
+<a name="line-22172"></a>  name: "partitions"
+<a name="line-22173"></a>  type: DT_INT32
+<a name="line-22174"></a>}
+<a name="line-22175"></a>output_arg {
+<a name="line-22176"></a>  name: "outputs" number_attr: "num_partitions" type_attr: "T"
+<a name="line-22177"></a>}
+<a name="line-22178"></a>-}</span>
+<a name="line-22179"></a>
+<a name="line-22180"></a><span class='hs-comment'>-- | Serialize a `SparseTensor` into a string 3-vector (1-D `Tensor`) object.</span>
+<a name="line-22181"></a>
+<a name="line-22182"></a><a name="serializeSparse"></a><span class='hs-definition'>serializeSparse</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-22183"></a>                   <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __sparse_indices__: 2-D.  The `indices` of the `SparseTensor`.</span>
+<a name="line-22184"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __sparse_values__: 1-D.  The `values` of the `SparseTensor`.</span>
+<a name="line-22185"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __sparse_shape__: 1-D.  The `shape` of the `SparseTensor`.</span>
+<a name="line-22186"></a>                   <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ __serialized_sparse__</span>
+<a name="line-22187"></a><span class='hs-definition'>serializeSparse</span> <span class='hs-varid'>sparse_indices</span> <span class='hs-varid'>sparse_values</span> <span class='hs-varid'>sparse_shape</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-22188"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SerializeSparse"</span>
+<a name="line-22189"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-22190"></a>        <span class='hs-varid'>sparse_indices</span> <span class='hs-varid'>sparse_values</span> <span class='hs-varid'>sparse_shape</span>
+<a name="line-22191"></a><span class='hs-comment'>{-
+<a name="line-22192"></a>attr { name: "T" type: "type" }
+<a name="line-22193"></a>input_arg {
+<a name="line-22194"></a>  description: "2-D.  The `indices` of the `SparseTensor`."
+<a name="line-22195"></a>  name: "sparse_indices"
+<a name="line-22196"></a>  type: DT_INT64
+<a name="line-22197"></a>}
+<a name="line-22198"></a>input_arg {
+<a name="line-22199"></a>  description: "1-D.  The `values` of the `SparseTensor`."
+<a name="line-22200"></a>  name: "sparse_values"
+<a name="line-22201"></a>  type_attr: "T"
+<a name="line-22202"></a>}
+<a name="line-22203"></a>input_arg {
+<a name="line-22204"></a>  description: "1-D.  The `shape` of the `SparseTensor`."
+<a name="line-22205"></a>  name: "sparse_shape"
+<a name="line-22206"></a>  type: DT_INT64
+<a name="line-22207"></a>}
+<a name="line-22208"></a>output_arg { name: "serialized_sparse" type: DT_STRING }
+<a name="line-22209"></a>-}</span>
+<a name="line-22210"></a>
+<a name="line-22211"></a><a name="sparseConcat"></a><span class='hs-comment'>-- | Concatenates a list of `SparseTensor` along the specified dimension.</span>
+<a name="line-22212"></a><span class='hs-comment'>--</span>
+<a name="line-22213"></a><span class='hs-comment'>-- Concatenation is with respect to the dense versions of these sparse tensors.</span>
+<a name="line-22214"></a><span class='hs-comment'>-- It is assumed that each input is a `SparseTensor` whose elements are ordered</span>
+<a name="line-22215"></a><span class='hs-comment'>-- along increasing dimension number.</span>
+<a name="line-22216"></a><span class='hs-comment'>-- </span>
+<a name="line-22217"></a><span class='hs-comment'>-- All inputs' shapes must match, except for the concat dimension.  The</span>
+<a name="line-22218"></a><span class='hs-comment'>-- `indices`, `values`, and `shapes` lists must have the same length.</span>
+<a name="line-22219"></a><span class='hs-comment'>-- </span>
+<a name="line-22220"></a><span class='hs-comment'>-- The output shape is identical to the inputs', except along the concat</span>
+<a name="line-22221"></a><span class='hs-comment'>-- dimension, where it is the sum of the inputs' sizes along that dimension.</span>
+<a name="line-22222"></a><span class='hs-comment'>-- </span>
+<a name="line-22223"></a><span class='hs-comment'>-- The output elements will be resorted to preserve the sort order along</span>
+<a name="line-22224"></a><span class='hs-comment'>-- increasing dimension number.</span>
+<a name="line-22225"></a><span class='hs-comment'>-- </span>
+<a name="line-22226"></a><span class='hs-comment'>-- This op runs in `O(M log M)` time, where `M` is the total number of non-empty</span>
+<a name="line-22227"></a><span class='hs-comment'>-- values across all inputs. This is due to the need for an internal sort in</span>
+<a name="line-22228"></a><span class='hs-comment'>-- order to concatenate efficiently across an arbitrary dimension.</span>
+<a name="line-22229"></a><span class='hs-comment'>-- </span>
+<a name="line-22230"></a><span class='hs-comment'>-- For example, if `concat_dim = 1` and the inputs are</span>
+<a name="line-22231"></a><span class='hs-comment'>-- </span>
+<a name="line-22232"></a><span class='hs-comment'>--     sp_inputs[0]: shape = [2, 3]</span>
+<a name="line-22233"></a><span class='hs-comment'>--     [0, 2]: "a"</span>
+<a name="line-22234"></a><span class='hs-comment'>--     [1, 0]: "b"</span>
+<a name="line-22235"></a><span class='hs-comment'>--     [1, 1]: "c"</span>
+<a name="line-22236"></a><span class='hs-comment'>-- </span>
+<a name="line-22237"></a><span class='hs-comment'>--     sp_inputs[1]: shape = [2, 4]</span>
+<a name="line-22238"></a><span class='hs-comment'>--     [0, 1]: "d"</span>
+<a name="line-22239"></a><span class='hs-comment'>--     [0, 2]: "e"</span>
+<a name="line-22240"></a><span class='hs-comment'>-- </span>
+<a name="line-22241"></a><span class='hs-comment'>-- then the output will be</span>
+<a name="line-22242"></a><span class='hs-comment'>-- </span>
+<a name="line-22243"></a><span class='hs-comment'>--     shape = [2, 7]</span>
+<a name="line-22244"></a><span class='hs-comment'>--     [0, 2]: "a"</span>
+<a name="line-22245"></a><span class='hs-comment'>--     [0, 4]: "d"</span>
+<a name="line-22246"></a><span class='hs-comment'>--     [0, 5]: "e"</span>
+<a name="line-22247"></a><span class='hs-comment'>--     [1, 0]: "b"</span>
+<a name="line-22248"></a><span class='hs-comment'>--     [1, 1]: "c"</span>
+<a name="line-22249"></a><span class='hs-comment'>-- </span>
+<a name="line-22250"></a><span class='hs-comment'>-- Graphically this is equivalent to doing</span>
+<a name="line-22251"></a><span class='hs-comment'>-- </span>
+<a name="line-22252"></a><span class='hs-comment'>--     [    a] concat [  d e  ] = [    a   d e  ]</span>
+<a name="line-22253"></a><span class='hs-comment'>--     [b c  ]        [       ]   [b c          ]</span>
+<a name="line-22254"></a><span class='hs-definition'>sparseConcat</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-22255"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __concat_dim__: Dimension to concatenate along. Must be in range [-rank, rank),</span>
+<a name="line-22256"></a>                               <span class='hs-comment'>-- where rank is the number of dimensions in each input `SparseTensor`.</span>
+<a name="line-22257"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __indices__: 2-D.  Indices of each input `SparseTensor`.</span>
+<a name="line-22258"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __values__: 1-D.  Non-empty values of each `SparseTensor`.</span>
+<a name="line-22259"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-comment'>-- ^ __shapes__: 1-D.  Shapes of each `SparseTensor`.</span>
+<a name="line-22260"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-22261"></a>                    <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-22262"></a>                <span class='hs-comment'>-- ^ (__output_indices__, __output_values__, __output_shape__)</span>
+<a name="line-22263"></a>                <span class='hs-comment'>--</span>
+<a name="line-22264"></a>                <span class='hs-comment'>-- * __output_indices__: 2-D.  Indices of the concatenated `SparseTensor`.</span>
+<a name="line-22265"></a>                <span class='hs-comment'>--</span>
+<a name="line-22266"></a>                <span class='hs-comment'>-- * __output_values__: 1-D.  Non-empty values of the concatenated `SparseTensor`.</span>
+<a name="line-22267"></a>                <span class='hs-comment'>--</span>
+<a name="line-22268"></a>                <span class='hs-comment'>-- * __output_shape__: 1-D.  Shape of the concatenated `SparseTensor`.</span>
+<a name="line-22269"></a><span class='hs-definition'>sparseConcat</span> <span class='hs-varid'>concat_dim</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>values</span>
+<a name="line-22270"></a>             <span class='hs-varid'>shapes</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"N"</span><span class='hs-layout'>,</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"shapes"</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>shapes</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-22271"></a>                                            <span class='hs-layout'>(</span><span class='hs-str'>"values"</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>values</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-22272"></a>                                            <span class='hs-layout'>(</span><span class='hs-str'>"indices"</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>indices</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-22273"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseConcat"</span>
+<a name="line-22274"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-22275"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"concat_dim"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>concat_dim</span>
+<a name="line-22276"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"N"</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromIntegral</span> <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>shapes</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-22277"></a>        <span class='hs-varid'>indices</span> <span class='hs-varid'>values</span> <span class='hs-varid'>shapes</span>
+<a name="line-22278"></a><span class='hs-comment'>{-
+<a name="line-22279"></a>attr {
+<a name="line-22280"></a>  description: "Dimension to concatenate along. Must be in range [-rank, rank),\nwhere rank is the number of dimensions in each input `SparseTensor`."
+<a name="line-22281"></a>  name: "concat_dim"
+<a name="line-22282"></a>  type: "int"
+<a name="line-22283"></a>}
+<a name="line-22284"></a>attr { has_minimum: true minimum: 2 name: "N" type: "int" }
+<a name="line-22285"></a>attr { name: "T" type: "type" }
+<a name="line-22286"></a>input_arg {
+<a name="line-22287"></a>  description: "2-D.  Indices of each input `SparseTensor`."
+<a name="line-22288"></a>  name: "indices"
+<a name="line-22289"></a>  number_attr: "N"
+<a name="line-22290"></a>  type: DT_INT64
+<a name="line-22291"></a>}
+<a name="line-22292"></a>input_arg {
+<a name="line-22293"></a>  description: "1-D.  Non-empty values of each `SparseTensor`."
+<a name="line-22294"></a>  name: "values"
+<a name="line-22295"></a>  number_attr: "N"
+<a name="line-22296"></a>  type_attr: "T"
+<a name="line-22297"></a>}
+<a name="line-22298"></a>input_arg {
+<a name="line-22299"></a>  description: "1-D.  Shapes of each `SparseTensor`."
+<a name="line-22300"></a>  name: "shapes"
+<a name="line-22301"></a>  number_attr: "N"
+<a name="line-22302"></a>  type: DT_INT64
+<a name="line-22303"></a>}
+<a name="line-22304"></a>output_arg {
+<a name="line-22305"></a>  description: "2-D.  Indices of the concatenated `SparseTensor`."
+<a name="line-22306"></a>  name: "output_indices"
+<a name="line-22307"></a>  type: DT_INT64
+<a name="line-22308"></a>}
+<a name="line-22309"></a>output_arg {
+<a name="line-22310"></a>  description: "1-D.  Non-empty values of the concatenated `SparseTensor`."
+<a name="line-22311"></a>  name: "output_values"
+<a name="line-22312"></a>  type_attr: "T"
+<a name="line-22313"></a>}
+<a name="line-22314"></a>output_arg {
+<a name="line-22315"></a>  description: "1-D.  Shape of the concatenated `SparseTensor`."
+<a name="line-22316"></a>  name: "output_shape"
+<a name="line-22317"></a>  type: DT_INT64
+<a name="line-22318"></a>}
+<a name="line-22319"></a>-}</span>
+<a name="line-22320"></a>
+<a name="line-22321"></a><a name="segmentProd"></a><span class='hs-comment'>-- | Computes the product along segments of a tensor.</span>
+<a name="line-22322"></a><span class='hs-comment'>--</span>
+<a name="line-22323"></a><span class='hs-comment'>-- Read [the section on</span>
+<a name="line-22324"></a><span class='hs-comment'>-- Segmentation](../../api_docs/python/math_ops.md#segmentation) for an explanation</span>
+<a name="line-22325"></a><span class='hs-comment'>-- of segments.</span>
+<a name="line-22326"></a><span class='hs-comment'>-- </span>
+<a name="line-22327"></a><span class='hs-comment'>-- Computes a tensor such that</span>
+<a name="line-22328"></a><span class='hs-comment'>-- \\(output_i = \prod_j data_j\\) where the product is over `j` such</span>
+<a name="line-22329"></a><span class='hs-comment'>-- that `segment_ids[j] == i`.</span>
+<a name="line-22330"></a><span class='hs-comment'>-- </span>
+<a name="line-22331"></a><span class='hs-comment'>-- &lt;div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;"&gt;</span>
+<a name="line-22332"></a><span class='hs-comment'>-- &lt;img style="width:100%" src="../../images/SegmentProd.png" alt&gt;</span>
+<a name="line-22333"></a><span class='hs-comment'>-- &lt;/div&gt;</span>
+<a name="line-22334"></a><span class='hs-definition'>segmentProd</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-varid'>tindices</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-22335"></a>                                          <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-22336"></a>                                                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-22337"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-22338"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-22339"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-22340"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-22341"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-22342"></a>                                                  <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>,</span>
+<a name="line-22343"></a>                                          <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-22344"></a>                                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-22345"></a>               <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __data__</span>
+<a name="line-22346"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>tindices</span> <span class='hs-comment'>-- ^ __segment_ids__: A 1-D tensor whose rank is equal to the rank of `data`'s</span>
+<a name="line-22347"></a>                                     <span class='hs-comment'>-- first dimension.  Values should be sorted and can be repeated.</span>
+<a name="line-22348"></a>               <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: Has same shape as data, except for dimension 0 which</span>
+<a name="line-22349"></a>               <span class='hs-comment'>-- has size `k`, the number of segments.</span>
+<a name="line-22350"></a><span class='hs-definition'>segmentProd</span> <span class='hs-varid'>data'</span> <span class='hs-varid'>segment_ids</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-22351"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SegmentProd"</span>
+<a name="line-22352"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span>
+<a name="line-22353"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tindices"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>tindices</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-22354"></a>        <span class='hs-varid'>data'</span> <span class='hs-varid'>segment_ids</span>
+<a name="line-22355"></a><span class='hs-comment'>{-
+<a name="line-22356"></a>attr {
+<a name="line-22357"></a>  allowed_values {
+<a name="line-22358"></a>    list {
+<a name="line-22359"></a>      type: DT_FLOAT
+<a name="line-22360"></a>      type: DT_DOUBLE
+<a name="line-22361"></a>      type: DT_INT64
+<a name="line-22362"></a>      type: DT_INT32
+<a name="line-22363"></a>      type: DT_UINT8
+<a name="line-22364"></a>      type: DT_UINT16
+<a name="line-22365"></a>      type: DT_INT16
+<a name="line-22366"></a>      type: DT_INT8
+<a name="line-22367"></a>      type: DT_COMPLEX64
+<a name="line-22368"></a>      type: DT_COMPLEX128
+<a name="line-22369"></a>      type: DT_QINT8
+<a name="line-22370"></a>      type: DT_QUINT8
+<a name="line-22371"></a>      type: DT_QINT32
+<a name="line-22372"></a>      type: DT_HALF
+<a name="line-22373"></a>    }
+<a name="line-22374"></a>  }
+<a name="line-22375"></a>  name: "T"
+<a name="line-22376"></a>  type: "type"
+<a name="line-22377"></a>}
+<a name="line-22378"></a>attr {
+<a name="line-22379"></a>  allowed_values { list { type: DT_INT32 type: DT_INT64 } }
+<a name="line-22380"></a>  name: "Tindices"
+<a name="line-22381"></a>  type: "type"
+<a name="line-22382"></a>}
+<a name="line-22383"></a>input_arg { name: "data" type_attr: "T" }
+<a name="line-22384"></a>input_arg {
+<a name="line-22385"></a>  description: "A 1-D tensor whose rank is equal to the rank of `data`\'s\nfirst dimension.  Values should be sorted and can be repeated."
+<a name="line-22386"></a>  name: "segment_ids"
+<a name="line-22387"></a>  type_attr: "Tindices"
+<a name="line-22388"></a>}
+<a name="line-22389"></a>output_arg {
+<a name="line-22390"></a>  description: "Has same shape as data, except for dimension 0 which\nhas size `k`, the number of segments."
+<a name="line-22391"></a>  name: "output"
+<a name="line-22392"></a>  type_attr: "T"
+<a name="line-22393"></a>}
+<a name="line-22394"></a>-}</span>
+<a name="line-22395"></a>
+<a name="line-22396"></a><a name="sparseReshape"></a><span class='hs-comment'>-- | Reshapes a SparseTensor to represent values in a new dense shape.</span>
+<a name="line-22397"></a><span class='hs-comment'>--</span>
+<a name="line-22398"></a><span class='hs-comment'>-- This operation has the same semantics as reshape on the represented dense</span>
+<a name="line-22399"></a><span class='hs-comment'>-- tensor.  The `input_indices` are recomputed based on the requested `new_shape`.</span>
+<a name="line-22400"></a><span class='hs-comment'>-- </span>
+<a name="line-22401"></a><span class='hs-comment'>-- If one component of `new_shape` is the special value -1, the size of that</span>
+<a name="line-22402"></a><span class='hs-comment'>-- dimension is computed so that the total dense size remains constant.  At</span>
+<a name="line-22403"></a><span class='hs-comment'>-- most one component of `new_shape` can be -1.  The number of dense elements</span>
+<a name="line-22404"></a><span class='hs-comment'>-- implied by `new_shape` must be the same as the number of dense elements</span>
+<a name="line-22405"></a><span class='hs-comment'>-- originally implied by `input_shape`.</span>
+<a name="line-22406"></a><span class='hs-comment'>-- </span>
+<a name="line-22407"></a><span class='hs-comment'>-- Reshaping does not affect the order of values in the SparseTensor.</span>
+<a name="line-22408"></a><span class='hs-comment'>-- </span>
+<a name="line-22409"></a><span class='hs-comment'>-- If the input tensor has rank `R_in` and `N` non-empty values, and `new_shape`</span>
+<a name="line-22410"></a><span class='hs-comment'>-- has length `R_out`, then `input_indices` has shape `[N, R_in]`,</span>
+<a name="line-22411"></a><span class='hs-comment'>-- `input_shape` has length `R_in`, `output_indices` has shape `[N, R_out]`, and</span>
+<a name="line-22412"></a><span class='hs-comment'>-- `output_shape` has length `R_out`.</span>
+<a name="line-22413"></a><span class='hs-definition'>sparseReshape</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __input_indices__: 2-D.  `N x R_in` matrix with the indices of non-empty values in a</span>
+<a name="line-22414"></a>                                          <span class='hs-comment'>-- SparseTensor.</span>
+<a name="line-22415"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __input_shape__: 1-D.  `R_in` vector with the input SparseTensor's dense shape.</span>
+<a name="line-22416"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __new_shape__: 1-D.  `R_out` vector with the requested new dense shape.</span>
+<a name="line-22417"></a>                 <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-22418"></a>                 <span class='hs-comment'>-- ^ (__output_indices__, __output_shape__)</span>
+<a name="line-22419"></a>                 <span class='hs-comment'>--</span>
+<a name="line-22420"></a>                 <span class='hs-comment'>-- * __output_indices__: 2-D.  `N x R_out` matrix with the updated indices of non-empty</span>
+<a name="line-22421"></a>                 <span class='hs-comment'>-- values in the output SparseTensor.</span>
+<a name="line-22422"></a>                 <span class='hs-comment'>--</span>
+<a name="line-22423"></a>                 <span class='hs-comment'>-- * __output_shape__: 1-D.  `R_out` vector with the full dense shape of the output</span>
+<a name="line-22424"></a>                 <span class='hs-comment'>-- SparseTensor.  This is the same as `new_shape` but with any -1 dimensions</span>
+<a name="line-22425"></a>                 <span class='hs-comment'>-- filled in.</span>
+<a name="line-22426"></a><span class='hs-definition'>sparseReshape</span> <span class='hs-varid'>input_indices</span> <span class='hs-varid'>input_shape</span> <span class='hs-varid'>new_shape</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-22427"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseReshape"</span><span class='hs-layout'>)</span>
+<a name="line-22428"></a>        <span class='hs-varid'>input_indices</span> <span class='hs-varid'>input_shape</span> <span class='hs-varid'>new_shape</span>
+<a name="line-22429"></a><span class='hs-comment'>{-
+<a name="line-22430"></a>input_arg {
+<a name="line-22431"></a>  description: "2-D.  `N x R_in` matrix with the indices of non-empty values in a\nSparseTensor."
+<a name="line-22432"></a>  name: "input_indices"
+<a name="line-22433"></a>  type: DT_INT64
+<a name="line-22434"></a>}
+<a name="line-22435"></a>input_arg {
+<a name="line-22436"></a>  description: "1-D.  `R_in` vector with the input SparseTensor\'s dense shape."
+<a name="line-22437"></a>  name: "input_shape"
+<a name="line-22438"></a>  type: DT_INT64
+<a name="line-22439"></a>}
+<a name="line-22440"></a>input_arg {
+<a name="line-22441"></a>  description: "1-D.  `R_out` vector with the requested new dense shape."
+<a name="line-22442"></a>  name: "new_shape"
+<a name="line-22443"></a>  type: DT_INT64
+<a name="line-22444"></a>}
+<a name="line-22445"></a>output_arg {
+<a name="line-22446"></a>  description: "2-D.  `N x R_out` matrix with the updated indices of non-empty\nvalues in the output SparseTensor."
+<a name="line-22447"></a>  name: "output_indices"
+<a name="line-22448"></a>  type: DT_INT64
+<a name="line-22449"></a>}
+<a name="line-22450"></a>output_arg {
+<a name="line-22451"></a>  description: "1-D.  `R_out` vector with the full dense shape of the output\nSparseTensor.  This is the same as `new_shape` but with any -1 dimensions\nfilled in."
+<a name="line-22452"></a>  name: "output_shape"
+<a name="line-22453"></a>  type: DT_INT64
+<a name="line-22454"></a>}
+<a name="line-22455"></a>-}</span>
+<a name="line-22456"></a>
+<a name="line-22457"></a><a name="sparseDenseCwiseMul"></a><span class='hs-comment'>-- | Component-wise multiplies a SparseTensor by a dense Tensor.</span>
+<a name="line-22458"></a><span class='hs-comment'>--</span>
+<a name="line-22459"></a><span class='hs-comment'>-- The output locations corresponding to the implicitly zero elements in the sparse</span>
+<a name="line-22460"></a><span class='hs-comment'>-- tensor will be zero (i.e., will not take up storage space), regardless of the</span>
+<a name="line-22461"></a><span class='hs-comment'>-- contents of the dense tensor (even if it's +/-INF and that INF*0 == NaN).</span>
+<a name="line-22462"></a><span class='hs-comment'>-- </span>
+<a name="line-22463"></a><span class='hs-comment'>-- *Limitation*: this Op only broadcasts the dense side to the sparse side, but not</span>
+<a name="line-22464"></a><span class='hs-comment'>-- the other direction.</span>
+<a name="line-22465"></a><span class='hs-definition'>sparseDenseCwiseMul</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-22466"></a>                                               <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-22467"></a>                                                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-22468"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-22469"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-22470"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-22471"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-22472"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-22473"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-22474"></a>                                                       <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-22475"></a>                       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __sp_indices__: 2-D.  `N x R` matrix with the indices of non-empty values in a</span>
+<a name="line-22476"></a>                                                <span class='hs-comment'>-- SparseTensor, possibly not in canonical ordering.</span>
+<a name="line-22477"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __sp_values__: 1-D.  `N` non-empty values corresponding to `sp_indices`.</span>
+<a name="line-22478"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __sp_shape__: 1-D.  Shape of the input SparseTensor.</span>
+<a name="line-22479"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __dense__: `R`-D.  The dense Tensor operand.</span>
+<a name="line-22480"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 1-D.  The `N` values that are operated on.</span>
+<a name="line-22481"></a><span class='hs-definition'>sparseDenseCwiseMul</span> <span class='hs-varid'>sp_indices</span> <span class='hs-varid'>sp_values</span> <span class='hs-varid'>sp_shape</span> <span class='hs-varid'>dense</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-22482"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseDenseCwiseMul"</span>
+<a name="line-22483"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-22484"></a>        <span class='hs-varid'>sp_indices</span> <span class='hs-varid'>sp_values</span> <span class='hs-varid'>sp_shape</span> <span class='hs-varid'>dense</span>
+<a name="line-22485"></a><span class='hs-comment'>{-
+<a name="line-22486"></a>attr {
+<a name="line-22487"></a>  allowed_values {
+<a name="line-22488"></a>    list {
+<a name="line-22489"></a>      type: DT_FLOAT
+<a name="line-22490"></a>      type: DT_DOUBLE
+<a name="line-22491"></a>      type: DT_INT64
+<a name="line-22492"></a>      type: DT_INT32
+<a name="line-22493"></a>      type: DT_UINT8
+<a name="line-22494"></a>      type: DT_UINT16
+<a name="line-22495"></a>      type: DT_INT16
+<a name="line-22496"></a>      type: DT_INT8
+<a name="line-22497"></a>      type: DT_COMPLEX64
+<a name="line-22498"></a>      type: DT_COMPLEX128
+<a name="line-22499"></a>      type: DT_QINT8
+<a name="line-22500"></a>      type: DT_QUINT8
+<a name="line-22501"></a>      type: DT_QINT32
+<a name="line-22502"></a>      type: DT_HALF
+<a name="line-22503"></a>    }
+<a name="line-22504"></a>  }
+<a name="line-22505"></a>  name: "T"
+<a name="line-22506"></a>  type: "type"
+<a name="line-22507"></a>}
+<a name="line-22508"></a>input_arg {
+<a name="line-22509"></a>  description: "2-D.  `N x R` matrix with the indices of non-empty values in a\nSparseTensor, possibly not in canonical ordering."
+<a name="line-22510"></a>  name: "sp_indices"
+<a name="line-22511"></a>  type: DT_INT64
+<a name="line-22512"></a>}
+<a name="line-22513"></a>input_arg {
+<a name="line-22514"></a>  description: "1-D.  `N` non-empty values corresponding to `sp_indices`."
+<a name="line-22515"></a>  name: "sp_values"
+<a name="line-22516"></a>  type_attr: "T"
+<a name="line-22517"></a>}
+<a name="line-22518"></a>input_arg {
+<a name="line-22519"></a>  description: "1-D.  Shape of the input SparseTensor."
+<a name="line-22520"></a>  name: "sp_shape"
+<a name="line-22521"></a>  type: DT_INT64
+<a name="line-22522"></a>}
+<a name="line-22523"></a>input_arg {
+<a name="line-22524"></a>  description: "`R`-D.  The dense Tensor operand."
+<a name="line-22525"></a>  name: "dense"
+<a name="line-22526"></a>  type_attr: "T"
+<a name="line-22527"></a>}
+<a name="line-22528"></a>output_arg {
+<a name="line-22529"></a>  description: "1-D.  The `N` values that are operated on."
+<a name="line-22530"></a>  name: "output"
+<a name="line-22531"></a>  type_attr: "T"
+<a name="line-22532"></a>}
+<a name="line-22533"></a>-}</span>
+<a name="line-22534"></a>
+<a name="line-22535"></a><a name="sparseDenseCwiseDiv"></a><span class='hs-comment'>-- | Component-wise divides a SparseTensor by a dense Tensor.</span>
+<a name="line-22536"></a><span class='hs-comment'>--</span>
+<a name="line-22537"></a><span class='hs-comment'>-- *Limitation*: this Op only broadcasts the dense side to the sparse side, but not</span>
+<a name="line-22538"></a><span class='hs-comment'>-- the other direction.</span>
+<a name="line-22539"></a><span class='hs-definition'>sparseDenseCwiseDiv</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>v3</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>,</span>
+<a name="line-22540"></a>                                               <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-22541"></a>                                                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-22542"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int16</span><span class='hs-layout'>,</span>
+<a name="line-22543"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-22544"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-22545"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int8</span><span class='hs-layout'>,</span>
+<a name="line-22546"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word16</span><span class='hs-layout'>,</span>
+<a name="line-22547"></a>                                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-22548"></a>                                                       <span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-22549"></a>                       <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __sp_indices__: 2-D.  `N x R` matrix with the indices of non-empty values in a</span>
+<a name="line-22550"></a>                                                <span class='hs-comment'>-- SparseTensor, possibly not in canonical ordering.</span>
+<a name="line-22551"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __sp_values__: 1-D.  `N` non-empty values corresponding to `sp_indices`.</span>
+<a name="line-22552"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v3</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span> <span class='hs-comment'>-- ^ __sp_shape__: 1-D.  Shape of the input SparseTensor.</span>
+<a name="line-22553"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v4</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __dense__: `R`-D.  The dense Tensor operand.</span>
+<a name="line-22554"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span> <span class='hs-comment'>-- ^ __output__: 1-D.  The `N` values that are operated on.</span>
+<a name="line-22555"></a><span class='hs-definition'>sparseDenseCwiseDiv</span> <span class='hs-varid'>sp_indices</span> <span class='hs-varid'>sp_values</span> <span class='hs-varid'>sp_shape</span> <span class='hs-varid'>dense</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>eqLengthGuard</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-22556"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"SparseDenseCwiseDiv"</span>
+<a name="line-22557"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-22558"></a>        <span class='hs-varid'>sp_indices</span> <span class='hs-varid'>sp_values</span> <span class='hs-varid'>sp_shape</span> <span class='hs-varid'>dense</span>
+<a name="line-22559"></a><span class='hs-comment'>{-
+<a name="line-22560"></a>attr {
+<a name="line-22561"></a>  allowed_values {
+<a name="line-22562"></a>    list {
+<a name="line-22563"></a>      type: DT_FLOAT
+<a name="line-22564"></a>      type: DT_DOUBLE
+<a name="line-22565"></a>      type: DT_INT64
+<a name="line-22566"></a>      type: DT_INT32
+<a name="line-22567"></a>      type: DT_UINT8
+<a name="line-22568"></a>      type: DT_UINT16
+<a name="line-22569"></a>      type: DT_INT16
+<a name="line-22570"></a>      type: DT_INT8
+<a name="line-22571"></a>      type: DT_COMPLEX64
+<a name="line-22572"></a>      type: DT_COMPLEX128
+<a name="line-22573"></a>      type: DT_QINT8
+<a name="line-22574"></a>      type: DT_QUINT8
+<a name="line-22575"></a>      type: DT_QINT32
+<a name="line-22576"></a>      type: DT_HALF
+<a name="line-22577"></a>    }
+<a name="line-22578"></a>  }
+<a name="line-22579"></a>  name: "T"
+<a name="line-22580"></a>  type: "type"
+<a name="line-22581"></a>}
+<a name="line-22582"></a>input_arg {
+<a name="line-22583"></a>  description: "2-D.  `N x R` matrix with the indices of non-empty values in a\nSparseTensor, possibly not in canonical ordering."
+<a name="line-22584"></a>  name: "sp_indices"
+<a name="line-22585"></a>  type: DT_INT64
+<a name="line-22586"></a>}
+<a name="line-22587"></a>input_arg {
+<a name="line-22588"></a>  description: "1-D.  `N` non-empty values corresponding to `sp_indices`."
+<a name="line-22589"></a>  name: "sp_values"
+<a name="line-22590"></a>  type_attr: "T"
+<a name="line-22591"></a>}
+<a name="line-22592"></a>input_arg {
+<a name="line-22593"></a>  description: "1-D.  Shape of the input SparseTensor."
+<a name="line-22594"></a>  name: "sp_shape"
+<a name="line-22595"></a>  type: DT_INT64
+<a name="line-22596"></a>}
+<a name="line-22597"></a>input_arg {
+<a name="line-22598"></a>  description: "`R`-D.  The dense Tensor operand."
+<a name="line-22599"></a>  name: "dense"
+<a name="line-22600"></a>  type_attr: "T"
+<a name="line-22601"></a>}
+<a name="line-22602"></a>output_arg {
+<a name="line-22603"></a>  description: "1-D.  The `N` values that are operated on."
+<a name="line-22604"></a>  name: "output"
+<a name="line-22605"></a>  type_attr: "T"
+<a name="line-22606"></a>}
+<a name="line-22607"></a>-}</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/src/hscolour.css b/docs/haddock/tensorflow-core-ops-0.1.0.0/src/hscolour.css
new file mode 100644
index 0000000..c15919e
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/src/hscolour.css
@@ -0,0 +1,5 @@
+.hs-keyglyph, .hs-layout {color: red;}
+.hs-keyword {color: blue;}
+.hs-comment, .hs-comment a {color: green;}
+.hs-str, .hs-chr {color: teal;}
+.hs-keyword, .hs-conid, .hs-varid, .hs-conop, .hs-varop, .hs-num, .hs-cpp, .hs-sel, .hs-definition {}
diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/synopsis.png b/docs/haddock/tensorflow-core-ops-0.1.0.0/synopsis.png
new file mode 100644
index 0000000000000000000000000000000000000000..85fb86ec84907bcc86531dc82871948ff4d471fa
GIT binary patch
literal 11327
zcmV-FEWp!=P)<h;3K|Lk000e1NJLTq004jh003wR1^@s6IVt!g000+IX+uL$Nkc;*
zP;zf(X>4Tx0C)k_S!GyNTeqHT_l8Y(cXyX`gGi?cY`Qxn1VID|MJXwjPC)?)F$h6K
zMMOd+6hs7sqbPzXbr*U(-*=zy-hcPcUC*=TdiNM(jyd-lv&OpsU|J&v2m2!^0SE{T
z54F(O;<U4O;KWV=44?vZKnREv-NJ%3&CHI1|Ml9M1kj;!OzhCt|7!L>E2!K(!rTCW
z%wV;vdzf1QjBf#e&~gh74F>?Z4a=WLg$KhJ^$5nap>PLbJadS>e&h8+?D`9%QNL`g
zEVKbYGXj7k5Q(8)0Fd#*a?VIMFW3*64geVHKzE-&0BG!BtmfuTbO(T`0Jaeg2nagF
z{V*1E{Wm{e|AvV~*MEExiC+KU-~R=!2{)|c6Bg`GjQ;iG|FQ`1kAUCTuZtQk34#8{
z4r4<P-^=bq9b@>(3g7#|{=Z@d+d#}7f!3C=>=26vx*jwA8>@MS>RG@Tt_zt3hie^T
z_?<zX!Qp0Bf6L=zytR+Pm;lRjJ;HSV_$l=z8Xd-r9G16)Mp#(Gm>0%9VUd=)Fos7I
z^ghPh%Jy%YZ|)vCf6EaFPai$Q-!=$ppK!y&wrJs)bNdAuANB!m3n34Tfj{s75g-&U
z1A!Pg3bcXF-=!Gv1VmU93G2duANT;{0JugFTqg*|oPXPC|A$2HS3NJd-hcPV3EW`Y
zh=1Dr-5Mv{<{zIvz#Yba<V3iQL)~HT!km4Q{?+Cmi$j0EdL!8&xgxpyr%xz;lr%~X
zrG-*OsiBkrCyE}$48%~fC`FVeN*$#F>y&^Vcn^E_`qRfl{{bzYkp)4~$~NAx_VB;E
z{?P)PU)DbV{Qi#~0H0@T9czDj06@6MNq8OrpdAz(9qQxd9nPr<&s+~tPQySqaZyfb
zNh!%g_5YjeaLxMN*$s<dYY{|0@-La=c>v_p;d%b#U$Wpz0Geb0U>E+EOsEQ;I<Re%
zqFY30<Zpkd!v`9W019}7G68nL0|epsEDhv=3eW($zzCQED_{?t;PK-PkHJt71>!&=
zNC6q(BFF<p;2Nj^wV)BSfDUjM^n!=r33v`Bz%-Zxi(nP3gDtQFK@b)qgJ>Wohy&t-
zL?CHM5mJM6p`(xmWDmJOUQi$u0mVUQpbRJ*DuT+OI;a`C4fR4p&?xj8nuk`Puh35f
z5<x)FBG?f82nmEDLK9((utJ<bcp*X%F^FWu1w;X&9C01dj_5@^K};a#5UYqyBtViP
z>5*JWF{C0=8)=GkKzbrWk@3iMWInPS*@Wyu4kE{pbI3L14-^JPgW^Pq!Q<2bWsPz}
zg`nb5nW!REEvg;Wj~YYGqt;RTXfiY_S_G|(HbmQ@z0gtU6m&ki8r_B-Ku@3-(OVb{
zh8`n;QNS2r>@mKWSWG773g!l;2Q!LUz-(f%SSG9pRuyZCC1S&|DcC~nb!<2G1$Gg;
zjU&Zz;G}VSI0sxHE(w>9tH<5Py}&KucJP#VKD;vC6z`6Y#%JLx@m=^4{33pbgo;Ff
zM3uyf#Fr$Iq=2M}WPoIbWP_BHl$%tE)ST3Z^fYM!=}po{r1PXd2-E~&f;PdC5J9*=
zs3G(aUK2LR$jJD~G{_vt!pSa>)sa0QdqcKOPD3tEZb<G<o<Lqq-cJ64{1XM9f{#Lz
z!ige=BA=p_VuWIi5>LrbsZB|wjHfK7yiNI%a+8XNN{Y&qDu61Js-9|yYMB~K%}=dM
z?M|IcT|xbTdVvN>!$YG@<3@9arjllWW|0;{D?n>V>r0zK+erJ2cAbuzPL|Gw?j&6?
z-95TFdL%tRy&=6neHMKS{UrTQ1~vvw1`mcbh9-s=4Br`97&RC@7}FVVFitT3Wa4Df
zW%6UX#MHqw%Zy?cW;SPzV!p~ez`Vvn%c8>K#*)s`!ZO8*U=?PyV2x$1V13HE$;Qs6
z&lb#9$o7D3jh&udgWZ=sm;FBb3I`2`8ix-@E=M=VM@~9UO-_H#0?vNUbuLye1Fi_J
zGOlM_JKO@?*4#<lH@IhcNO%-^Jb7|?`gqoPIe1NY<9Hi*r}*%EihSODm-z<yzVi$4
z+w!OKcknL@FbWt7#0WG9%m|VRstbk)mJ5yxVT2Tg{DiIuJr_m@%L#i67YmPwAVlOv
zd_=B@jEbT~l|%zY%SB&`5yUjbBE=fS=Edp7kBTRWcZh$I;FWNYxFj(su`ekn=`UF+
zIW0vkWh9j()g|>+T3Fgmx>$N#hD=6JCPAiC=8LR|tcUDX*;jHjawc-Aa(!}p@(S{y
z@=fw93cLy~3MC3J6=@aC6f+ecDWR3LloFKgD*aHFR}NQhQU0tVrsAhkud;kZ;E2bO
z$|DP^+^R&?GSxXXPBj;`QnfjCE_I@Mx%xW|9u0SmYK<jLAx&S+>zbdmB(*}d+O)oF
zD{G(9?$JT&=<B5GJkzDtwbm`teWS;%=cU)6x2`XzAFJPEfHp8TxMc9skj>D|u+DJZ
zNWtioQNJ<4*wVPj_}x+AqoGH;Ob{kUCOIZE$M}u~9_ug#riP|Drn6=OW+7&G%rWL>
z=Ede8ETk;rECwxUES)XuEw`++tg@`8tp%+ktov*zY#eRsY`)v-*k;?#*-6-)vU_6B
zZ0}>=>40^xaj16KJg$2@@A#sloMVdP<OzWjCr=DHF+2G?-E*dJCOWsepj~WT>Ron;
zro?jMrmLZAiR-$Xw%cX5Rd)^dT=x|ZRgY|sB~Mk)Y|mvcRj(Yc6>oL#eD5_MZJ#2a
zFTMu8*L=VGnflfE9r)Y&-w413xCGn|qz?28><?lOIvF$?EE1d&JQt!Ik{7ZbY8+Y}
z28Epn>kOxb4~I`91S8Hy%txw47DsMJ*+jLTq&gXR@@ceibXxRMj9yGtEGpJ5wl9t=
zE-`NYl;)|jcqraAzAu3%Avt03wEpSZM3O|m#Ni~#r0k?`XKc@OC9@@;PF^^xf3_io
zJS8;cWvWW*wR5O*KIfjL$)pvg?Wen^KhBWM$j{i#bjy5vUg~_o`GX6d7oKIwXI;IB
zxfpnH@{;<c+H9)qnCyidlbrTk&fN6eZ+XsnL;14#*DjM>j<`HmaI~Pakhkz+;ck(4
z(L}LU@r@GJlC+ZVSKP0>xT<lr=^ER$jB7tjeM(=I>6f*a^OxsWU@9UjK2+LN4pu2v
z)m1ZBXH@Ui1lG*eTGaN}Db&@~v({%dAQ~bXR<1i<f8MCm*x4l3RCa^@M#fEWGvemQ
zW|!v47UPz_R>jt)TYR@l+GyI++oAU8_Vo_$j=4_z&e7XOxBI$Oy4voD->JFFb+`B)
z-My^)B=?i=A9TlbZ}tTDto3^JF7!F~O+T=EFy3$8|7^f`;L$_9hYtod2fH7sKDs-k
zJaqf9;^U4d@=w~I$~|oxmK$z+CjYE`L}8@!xzh8l(IcbxU#P$69n%?mIBq!pWa8Mw
z=%n@JtCx;1=U%zLT7K>S`pZ=0)Xwzj8T3s0Eahze8`d}FZ-w68n3JEoH?K4Q^qu9q
z=>@li)%RiVcNddCkbTHs;#jI%mR`QQqPOz=<MEG+tHG-~Ybk$G{Z;Zw@YAi&8lRu8
z+pRBt3H@^LHFJY$qvo6Jx4uoY%{N;CTRY#=x0$wUe<=PK{Av4hX(wtIyIZg)wAZz7
zygzdgaBvVz3?=?L3n1YV;pqv$*BUs7u!nPp=KzrO{LU$US-|NE1lz#a2I7HkB7Bib
zC_HKj-GeE>CgGy+9whdp4g`BLCvp!8U&;uov(!a2t+bEnRv6HXyi9t`-YglcEo`$K
zI8GTZXYLH1F5YE+b^&9-c%dfYc~N>X1MygiCdpZ8N*OKLV7W5+5rusvVP$KTgd_E;
zV`@J%*flk^Jhjj1)aX9cTQC5ItVZ(2W=FkE;*aH-)|+*kk6SET?pjmWaNEk+>D${o
z_#cmV%sNr-bj$gX%QW$m8{|&wA?SI;%go!uC))SCU%7vKz~jI-L0?1Ap^RZ7;i?hG
zB3+__P9{WW#uUa@#oavB8Q+`m==5;nXwvwZiR6j1<0+%5!{;8Q^`_s>XwIxTUvlAM
z)|rdpmprp=bM$iM@_6#8@((Vr7Q8HcP;{fXs3iGH;8nY8TBRaov}JqcixtC_ZB<vQ
z&(`?YTGlDob2eZb{<=Qi*ws{eBju)Fvqg(SE88vPt>w07?YBCLI#1vB=rX<|d6)j~
z?!9;SA9XkN<n~7QIX}?p7apJ-*nBuW*z>4rDD83J6N{$`!z{xG&lW}=KCd6md=WHe
zF)la3F!5t@`sLkMS6?Sg5vR3gcxTbGOK%>(y*_twKH{Cjg64anMViI^4{J-a%g<LH
ze(YYoyLRud-cLiHC*aZaV}tse#HQ(1*!QCCfgc+?yt~$WY5Tng2XOpFKtGg+ut(A%
z-=j*=VHi8C5{?;<#;=l0k&Y4u$@<72Q1np_Pz_U$(Jatz(CstOF!D30Fk7<(v!=0C
zv-fel=R$LH^XTz<^JVZi3XBSF3Nwow5p@<z60edNko+Kxkr9;Dk#m)gS143$Q5sTy
zqw-A^tHz`*q@kp#r)8miT*qBEKrc)`${^M-!6?x<@o1vS>0=3|@n*5+(H4=G;Z`Bm
z0XDw2UUnY#t`5ZG&W<NeoN#h<c62#GbaHieBf7hKczXJJ`FjWZMEFMgo$^l#ND0gg
zx)_`nQW$zAtSr1TqB^oJs{Uj{bVE#iY<*nqsjB$0ge#{D6LXTzpE;Wxe>ObDFO_)C
zCe0{aEki1k_dNXt+=U-mA1_W_8p^(%Q<IyQmz*DR*{Q&|P@#yYn4);6Wa-M-)$VKc
zrTJw^<pCA;mAX|@)oe8+HQ#IB*NxV9H#A-^X-sd5zTtDzzS*!vsa50_Ya2Pdf^Bpx
zbxz-Y-ZgON&fVsFb@$7<i+XZ;v-&a~r1qx_oPC%wnEEJVDC=?lldDf_hFhL>j|@Mb
z9sM+h7-yIepVWIvd=>Y)XzKR#)XeT1j<?!#-t!mU-CCGlggyu?nJkB_<bCW|eYv*z
ziR!cPy2cmFuWlPb-=a6;w@!ag*pB@X{?m8o#IEU{#=iIg)35lmfin;%)QHeRj3K>H
zI8-@&65hs?W6g0$Tn9b?K9MevmJ{6JljSOT6GbGYHWfM5G<6M41g#z&E8Qx6H$yI?
z50eHn6Z1ODBi1suSavH8F-{EUJXaTYHjh8AJ|73)7XPq7g<!OhsW6Z5mdKE3k(jTz
zvII_ILb6OMKw41-$PCI}kh79!lAl#5RdiS4Q~ID>t>OirQ5IDz)!g7S$y<#pnvPn`
zTCcP(>sag3>W=B<=vx}l7>pa{8`&AN7|$LpGx0noeC)GnyV)so9SefRgyl6WA8Q%w
zeVfO&`F8I1(hk7k+3~B6fhW|RD4pIpx4EPekGo2^q1>k2n?25Xx_BviQ+coYJoGK~
zi}SY&kPV~?{2VkK+z^r;>Jw%VE)ao<Sd8qCsyuln+BHTamLqmAZW>-y@)AN%A4?QY
z!X(X~xtpASHaNvFl_z!g+(cSqdP;^<ru=#G^Gg@HvkESTTr$e$&EC&>mD`$^mG5`i
zpn$&+Rk%>pUtCp^dd2Um*){o6wlZ|t=klqF!OHfk>gs};%-W>7nEHr@(CeX%5lwM7
zQg7xp*S7SwzHLLbOLn+*Uc0?`NAB*$d)wWCJsW)~{h|X4gV%@BpPU*_8L1qd8t0!(
zdySmVd!st{bK%K{=9Rj&=Ffv)KX1|hFxkC)82{hg(&3(fkq6-NB>?O?0kGBtAd?QJ
zm0$~|LIBLj0I*U5i1iA9XzK$|?dCuG2lOlFq=GX}9v}f{nuc(O=>uZH1yBw;!3bD_
zU{(i`gLA_m=mOLPjX+-zbO<GcGvX|w5itp8bfQQ*WD2qkUVrG}Of3%84DUniXj}9p
z^ivET&bv}EkFcb0hLww*!SUjPaCh){ydAy{ze{3HQcDVv9w+S}&=VpFugO%&D#;1t
zQRE9078DOC<tZDexT%V%>8W#QsA+O&>1m7Uxak_`<>>nu%o*kx!<ev4CCswS!z|vc
zXx2(L9kvhb=^T<AuQ(IA#JHxo(|J^QKJ!-bo#1ES9~a0Gv=U+xdMjKf5+SN1MlZG|
z-Y;=YGG59>T2DqomQ{`*59GHMHWa@qZ7S~^!Kl)z@vEz7SZjuAWovinywxMoS2FN7
zEH|1t%4A}H?2754xrD_j%Moi{n>gE7_6iP##}7_;J59Lg5Ifz(-D^B~y{dc!eQ)?H
z1`GsQ2d{)Cgfm98MOmHv9&;s5@6?xs(nO0hxa6LcxN|CLdl`M_GqP+i31t7w9nHU9
zkY40hVt!S*RG^%pl2DDR1@+)Ms)_U_Lks^c#r9*J-d)LeEAIFAEIl9{kQ}rbihXiz
zxOZfJbZ?wtQtXx5l+ld&8>=~scSi5kK8P(dtn9DO{nh=s_)Emb(M`^+uiKA)7VrA)
zEB#tO5ODlSVZM$P@WWh#2Fx+Iz|6u~m`%6|24UXdCqxG`1g0=2kOkd@#-Q&AR(P%P
zMdTpvAy(jBM;jT2tUyk{D~~EF3{{U>K(nFk;T(JdLx-`&6l3PF0@xsI7Y>87!d2q7
z@J9GD{0|aKlAELyq`{<P1SLWl8G$UCY?<7Fe1t-eqMK5MvWrTEsvFK%hiUBK+%$p?
zLsvvEML)#g&IlN<F{v}XV@_oeW0_<<%_hn=1;_jd$2w;fmm@bl_bAT=UK2hFz8Cxj
z0?vX$f}2A9!g(UzqH1CcVqe9dOEgR7NySNf$XLi~$;ru!DDWzBDsd|Fs0baAQPog0
zQ+L;h)y&gs)gIT`*5lMSHi$92ZZv<C*~A7u*Ss=kxA3yOWsS12wQaP=IJh3~J;CFY
z>in5#@A}YP&ZEYQ#XH-V)Gsvv6_^~14ao?j4lj=6k7|w9iW!UZJhhvUlPHq(FxfQ)
zq?V>>q`%8dxgeZ1aw#H*HTOZjUjc35y<*QR6jwV-iRB~}tyPXS=-S45n}+?ysv9OZ
zzqJ(K(rR1j$hs}xHG4PtzG(M&@2Lj@{VyISJQ5#z^W@U7{<DMUcV5Jft50HIPQ0$2
zPMURnYc#L4Ah{^9q_ASN>hV|l=i6Vte3RLV-yYuK+dKCw{z!laG%#N$3ABJM%p<0O
zYA^skKqQbP%m$r-WBwLFh0ujLomRwONMWQ8vL5*f<`CmhgJ?Rm2f718hVj63W7)9r
z*mpQXTq~XnpG|@xNg&xFjU_!Gq>|CVvs#J#1w}9=HDxE2J2egUAWZ`85!yYvKKcv>
zJ4PYKJ*G+KW|m8=VQlv7TJY|}%0<qN;Qq<;jdz`Io&T%Amf)T+R)kiRM@&}SK*Cw_
zq*ShSi_DnpraX&+wqk%%q4J>0wyKDli~41a=UN19Bb{{JVSQ=?d&3H&&qviwE*<+|
zre<Dfv2EpG-EAvqm*;>!9^?4cDF}{Txa*#Kx+jZQvyZXwvVVG@WYFu7)G)>HwaCho
zPBE;pGpDX4cqED@Z6)`nTsY^LE}F4-ek7|Lj+#LpTmF}Vfuf?4z^j_2v}GSEI;v7@
ztn0YySFg7=Mcq_r{?^*qM(m*I?Cd&z=li|$-7G!jeOwO;25=992S<m5AICiD8%90T
z7zuw~Iy(4bWgIcVFe&g-@|E0cxhctM!5QXRlG$%>X5MzsmCeV$vtN*Wk9q%cvGzm6
zlGZYQ`Nc~9M~79`)tR-DzwAEIeH!_EZe4SI`^$~5?i-97Prt=)N^Q<<UHq>3ePg@o
zht*Hi&(<B6UGm+oJ@<X`{k8*#!}y1J%19sX9snX}!Sv&sgM$rz0I=Br>|HuI*eO3a
z*sFk(4fq>KkN@xQ6^F(cm~$_2K14li9;XkV|9<@!M&f%8Nam8p00009a7bBm000XU
z000XU0RWnu7ytkil}SWFRCodHT?u#;Rkr@KbUNvfeG_5`YY-wNfPp{+o{ADgGcxep
z5O;8ydCWk3pWowCbe1RjK4lzy;4&jKqk}U-a1=+ud7z@;LLwlFC>S)v1jwFrI_XY2
zop;WyuIf%_F<t2n-Ic!IS6#QNZr{59Ip>~x?x|CCgE~7q5lBOq0>MKUdH^|7ARquk
zTn+*P5D<VNPJtl~KcE2u0x-lWFvQ`9_7{IpP!Ma2Ag8UrfAvN=8&%31Xc{{x^H{Co
z5Bf3s5%3VAbLh&IH$@c|uDl{TIw}*k&TyNplZ9bWQ*bafH>lHMG@8ELxbaVWHf?&T
znHpfF&E_pZ&^rD;1;7qozi0Q$(`V)7{8<+kI>wdbHk%E>!9AN2eO+^{$KB)hHtVU6
z4;0@%KYw`%{kM%aj|)L>`1``u*EM%B_Ep|f_<QCY-I>7iHT~t6&rZsneaT;XVt##n
z3*O&%0=#!k4Gq$@x_XoAC663)d$?Wm=U<vc73yj8s7C4ye?8%O?|sJH9hZdp8`Rda
zV)AbC6#gCsfWR{`**=IUEL^cLOs^a53WvcKmwX4axn1m?g`lRU8J`}Mu9cL0kEW(p
z*S~w@nA^2_^<7tUJ89F|+dbgAd&6H>XTrha?_sgD)BZa!4dhf)W5g$)o+5f!@!6p=
z7>#E6lGpa0z~7?)*juclePn!mT$U>W2F?VqT7?}(LqHHhL#3+DoNXk5_#Pb{(lwSP
zZ<=X|iSbjYeFoatR`H}3=!RdX3qeSTbc>FTPC<i4yKox}dW8l|7rAq|($3Sz1I6E)
z0PuMxfT6vuJ<7S!uFMY$3&G48V-OV;j??9JIDGgtTjt^~;o&;W${mB481A6nNfvWk
zG}X)k!0qh~kGnVT5kF&`L0DN3I5DFSlBQz~XE21}l1oOBfJjYXO~BgEYVC0M-fop2
zDE{6E#$wJW!`duuVRDQ5Vt#39HNO0^LN1EPNWEbGuwmiKId5c0K$-Du1N3`J+mCki
zh@bx#O+t{9<6v}qdxy{j&CM3X#uzYdYMN4g6Ac50s6((}=&dwR{Cx?4=mpk}4k7Lh
zja}8gbI(8i^%;C~;;e!(COSeeKO#br9p##*Gxqi|^SQmA8UnbUF8H0>&5WKoW3<xS
z+uAzF<~aqE$%3S$Xk2`8s#2WoA?J>vT<}n4p!j<e0B)WjD5-NnRG#WdwxQH=@&k}$
zOR8R0dom`Lc=?)q&RcuB7q?Ti)x-Dbg5R}n#wZVUe2wiO-Z*u-^!Q-=uplma_-CN_
z8Dw8c=~1luSn8d}KeqH68(WbvJ_VB|4Oi+hk~;kO6RAfiC;c;{;CF65PXUWd$`m2j
z>ve)Qtntp05&Y$`N~L&mauhNrjZlt#E%Rdnz*4RdA(~WsS0P~4Cker*^h9K3rID79
zAhx!)2_f*-6tD+E@|~5o_HbR*DQEm#fix64W;xPOIEsuwz3>ej`Mg}wlx+M?%^s;7
zt7<_1|D+24j|zb6{d*Duo)R*nQ%A&N`m}UK6}Gim#oV|jr-^I5{&3u6Y!z0&JjK=N
zf~iA{0UNr_&1RH*=FkdaRxmwXu@ih1pW6b!KwO1@&&hNB<m70YDveg=#B4Xn!v>f0
z=VYU~zns|bF>|Ig{pE8Oi&e4q8Sf>;d>$HnJ*g4^2E{@!BWJXj|MK2>t{)#4iCiKM
z_X3_Wd3!22SVWGECF_5t9Wx1ebd<xtk&rM9vu2GY%sRn*f5AMUxid3;-rS5H$2b;{
z-mVQ6!@O~ym}o<<xL6?O&7IIAJ-pEtPt>Ve1IRabo*K&M<b7Y)<TC!a?+aI7*L)G{
zfCa!8K?T~>e+mp(08G`jsI~A7O*rz=A?*I(Ym_y4*Z<vrc-}ABZsXaPWE_55X~M~}
zTIJaLcOUp224n-2zVikie7yn%YnEfm!5#SRe*5FmzkQB}p11+o{Fp|=>BHj<`2EIL
z@XCfeu<hfsczydWJ6bb}o?S)x0g-o5D5qWnbezFoiTRwunk%qdX1+L6i)|mA5jidY
z{0}jolhGWEzbnk=GvPS&`Ekzpd{Qp}jg2@(@Fp)^ip7aWa9kF~U5c$wF9JUgt8d{b
zE$t=?X?=dwwZ3ox^kn_>GtW8G6RGFlFM<@CjE-OtU#5a;0kB%yXw(N%<3n(~<Qsu8
zZ|+#=eP8Q&gdZK`A?MLQF(CmT9+sYjF8fKIN9rETN>sBeG(H{~)Y9EAyo%kT#Rg2j
zpdOnacnjrpoDswQL%S&=xD)LJZ^c?^7~tUKxVSW2U-+UJ`I8c2{Q|sd4FLUcTr-0M
zaqMa26wFKpz7U~s3AlNV<LRJWL`!?_{RP)vp+K58DGni_Hng{J{G%C}6UQspd^w|9
z18~65>^qhrHMbm9<`9gTLcVV_VCkY<jGCK;u}-q#%LEum+wQu00v693hqg8|E+F?e
zr#1I3!kny;lw%R`{p7X)SelGzum&R`J{*=hGlr!kVHo3VG@_P-B6~JTzO@CWjwnQm
zSR&sif;%k~Gk*Hv$KL#GDAqtA4UzOxf>cW$bp+1aV?*4j`n;5NQvl5P$NHC1)DVqF
ze?14Uta}S5dTDmrRR#Fn;tPAZ>c6M&cw`%zt17X5(`x+mXPZPMYENh$xHA{IIn#Q&
z<WuL|wgH?ccYQ_s*J*Sb=&abC^R8TNFm``&#I=ks9J{Fj!m;GPmSYjgeN40tkN)>^
zG}YF_5*3HIuofIEDMeLB1jc8M#;C+D(d52>)gx`#@~i9ZqkAV_+e~x*&R~QFvHtHw
zX=O8P?QIyJ9<S9QfZh~|rm@{l%9`SLom|1Yuu=8~?mQv6f;~cV->Ss9*B|&g;0hMp
z3Alm-uHb+xn7Ts16&!E{`__2XkJh+p1UhOAxPk+&;D9SQ;0g}7f`^<d*f9ujPKWd{
zNwA(ei(>~4p*<oBsnf?}Y_tY_z+HhWM{(ruZlLc|d{fqlC~^zWp3Kn@&XYx@s8428
z+W?HF@nkVpZ73FT0nX?mGjp-|=hGB_f%7#j<?C8Quwu<K-PpY#ZzNXSa*ba=b=wuJ
zprVklr=6G|NiR*v4kFXBPw?d4Dy67aGq*W0e|U1NJv0<}=fG*4{I=9DV5`;u*19?o
z5{B?pJoNC@$d2dNpxSZdpKs&0Z&je7$@!^)U(mv>Mp`Hum_uHM8Ep9TllPO>m-^Cs
zpVwg1bK6i`-w1z*2vDs7WXVaJJHyU=rk@Vk3#W^iKzdl}7D4^3u#E2B8*>%rGlt8u
z5=Bg)^vMF>N2OW-kTeo=C=#;#Uwg6hiz=At%UPznGuZL$9uX3jIcgXzEoL+}ne7De
zePX!NLIZ__1sfvp<d*$}QF`*n@c|qD`l$#Bn=o#aJJp@HvT{duzX6s;924PRO?YW@
z5nghs!8um&p{H&~w#Fia`1v1)e}&ROaMy!>aY5fTR<P*VpVM{qeXU#N9z3DGRUK^<
z`8ISD3xNMj_Il@&dJSMm!=vkNfHgFv_xE4u6KU4aaHMFg_cJ-U=WN?_Kka)L1Nw>(
zUH5HKQ7-^w@TCk-ATqS$+;^2Y-9Yg{p~En8>~LcE&~OCN2SO-y!qgT7qsff0kWR!$
z^D81!lBm$TfXL;}=Y9YJK+SF{!{d*<g%X#sAaChntiD3qVd2a){!IDu!avzV!TF4o
z9xGgkjB%soL!aKOw*koiw_`(h<}=y@MgPFg@A$O|ZKq#6{$@AkGvc*R?Si#UlQDr#
z3~i~w{&Resf;DUNF*}X?fEFV@D0-Vh!)meXqjD6C9^>=}ZDsk}pA}{0WdF3_)n|T5
zFNK7P(SF;zrP#jx9qieE2>F-K@p;gyHGt(@rI_!hEt)McpP}lbFn3v=a0JCAI=-Ld
z^HfmLKw}#PgVO)j-n&3BpR3@}{)WrPilHHGIK3w22T8R6=u<`rMwjnBh~jFy5zt}A
zN81hv!KkMXNNPDnh1mq7H@>uwma1@k3;2!wtQCOj+9tn%uigkWBw{AL|5)BofhX2&
zA+XZ302%fCsUzg9CimQPVv`f;C6O8|{n>ML#6sZcPqU_9<QKRdEq1%Pk+|_c^D$ZZ
zpTp?8<hlYZ8Sj`5DMd7vdH&JsA-;|ya5+f%=&CEBr`QO8Na{5Jql+JdN1k1QvXkYg
zs3w&k5rOovX-K7)MutMt)1dz*eGsClwG%b<Mnp!CZzRsYY@Y!MK;y}yczWw0oN9FR
ze!+NJMke*DMR;s+x~Aozt~D6zfBzb)s_VMoY^H}`58OE$vvRV#ocF~8^#ULQLiXc}
zUQkuE$PY|=DWHMVx2^eKu~LOA6rK$<8jY@}eezhf09Y%}*qLuY{&mxkmF%SY=ox@6
zF~uhVd#X`JdPSa>DPe!$!>g7c<pOq#ONp0~Ns93gcerW+2#u6Nz)abBxO;&=#J<Sv
z@ytFfgdg<)OXiP7*2GlLRC*q2ahR^|-aI}H3;8oRM~>oyleK6R!5=0O9Kit+4(r(6
ziv6QJ8-P(X4Sa3SakRGjFIv?a0G4_jZD3}d!^RD-cH>&cq5?d2jrKkeAp_;!Ur#;&
z9W7Y4e9epUX=T6m-g%gom8l&2YDT>Vpn#D2K2TLOYC9;D1)wkDRn>N#8T3J_^Lk0W
z2GEDo5^3Wxdgdfd9w7&WOIUcVywJ$#^9sz{H)rNATQUdN%*}+3f?}K#TL)6Cfb&`3
z%&Qjw3IaWJ_$1z;4dDsM&%YQ~=42pUgopbkSWmW!9lu+5e2Bl(Hp~!=)psw#l#5d7
z<59t4!9`Er%bRtn7l4p3WRMY9&31sf7Q0{HC$^-K>G(;07G_Pk5PmWfQbk{$>nD;C
z$aX+;iw(co_@<~Qn^p+B=a%_MiWA>XQ&sn1{z<(6(1#*dufHEF>#Fe8m!&8!F2%dw
zHlg}-8UFYJZG<8tdn)d^eHPNC3G-m$^7_440RBMV3*u1l6Q_-M<NrRRMWepLk*_Lo
z-`bbN{R`%2;?^Z|DbJ;SGxcT1^sB{@@yW=TI0_Rc#C0ow;>ckXuK!rmQ$k)#dR$sG
z@^U71!@qOSF|2)@pOpG;Qm+AE#NKTmpy<6aRJ-8I$ex7UR10>zRSMI&Dx4*+aC%oe
z$>ksZdHCl3@33X-u5M#~!F>8s>bP;(@Z1iZ5DQ57E<W`F&{Exn`H?9pD7b1W{*9JP
zOVBg7&1l$Lik;?q)RUhiU1Oo)Yy>(pe>^RmdH=2Rkv1Y;;r0f4a|kUQI?AO7tZbEf
zJ(*E203jiWBR5FKRnt*$=_L9l06hS)bRb+XpPQ(|6)W>G1u?i-W6WoCJgU<Q$tb+@
zD$Jffj@}LJ{)WH{zrK~CgV-P9o6Kh%ssy0+_)+}f)#J#erD6&w@S}9^J6Kv&%Joq)
zVCl;1aO2ciq)r?I|MikfXnOjsHU;7FXSU<j4Ywhj0zZnsHR6e<UbYvt;`zsJK|)+S
z62!-nogR|)RM~kx=eB4z9^a39Htt8sxmx;YG^zT!qj*qZKBH+t(F@!0k$*VC+$MjO
zDHQiTunA>lRkTWYJ9y;~2lKhQP~5|72z2_#^8q&npdI^OKWZnM4)jd~l<t#{4Y+Cf
zJmk?iep_h!z8~!6w6%M3%WgX$8wEf`;a??H_G2r4_wX%<Bi{h4{7*`b`2}z{1;B(K
z1YRBPTX!e&hj*a(pcMU@bm?3ynVpH1H&2p)w`(&X<ooZ-37-u`n^q8?b=dLG_XXa&
z|9vVMfYDf2c#}A8B~Ph8B=-cMr4A=W1b8$?C6oGZsl=!338XY{hCK?{ZdF8R@clrx
z<=e!HAY|AD>xFIKK%PKOm(9u+`!IG4P>PAtq9@Rh0<!HW6ucJiefk6H>JE!{0DuH!
zkK`y|6ZXDM&ju*fYcM2?dkd?0BQd?AvKl9=rI$l^%Bzo%82<d<-V^wnjLl;2>pwp_
z3!t@d`N^j}MPee&>2}gr!FRvB)4o^~UCPYDMfxiI>b@c+MsVI_ZG?n%#SdILF9)yD
z8iBv~&32h6$j=)^`5;_--)1F7aK==Pycf`JwRRcIa&EjD`NGhX@h9M+TM4YCmA;oJ
zrO3=nv3MeD1n(z%`&dZj&7(JU#eehVv~0XE^yJ%^arZ3+;^s6cinJi_LRv*8MlRsh
z{Xp^er2%-zvwii|iPQND<~cxwB;)S&_u$&{D%8_7aQMh%>8YP30yAe!z=De>;j*0J
zN>6b7(K|VAAJyy)=J$-BZpMp7n5{I{+sN@1<}jm{UYm<6az<l#QupT~-a!ux6n{`>
zC)2KLBDKeY!To$ha&qG2BZqfAotPNM^BbQ^H8u4$*;5z(vZ|_v=c1LgH4&aJ8cR)s
zhZ25=_;#ffO9d0sLd30K^&jiDoI6+3R|Htse-FYDw`bL=buUu;*yY6jR@v$9iMtOO
z{Jm)a77X@ba<Dr`c8-aSjUKyf*>%$<P8>f%7edh>l!!{woQDqvAyLn?wOiY*$B%zo
zv32X~pEWczvH$rLZ56cfy6vr`0a$epDA9d}4E`PkfT>4BU?%e$j!CrfB%e1P1~}M{
zuQ8DZRRHLI>|J6XE5CNbPoY`u^Tv~L_DESt0J@K9biv&;RPgs@1TwMtC4bqg&n_U&
z^RqpU@fmCZV8(Krcxd8Db|Y=v9v+%_sqO*ye5%7a4GH|cY5=AL^#T?U?(IAraOf}Z
znfd(s?_l?Sx}{(;kM%5!ES&ry9?r8?uz9NYQ(Ynr1^j&q08@d8z|&jaWMSaE-1`Sx
z2*lKk?$1KN8*2mJGw(g3`l+riN$dE3Q~;P7L<D9yC)d4ncJO{>Cd=wx?7hW&8J3pu
z_e%g|LIn2Oqk!C_wTCQ#s9zKa2tdEcq}@UR0njdQ`-LnZ0R1A9b_)drK)bx{7qWl=
z^ovZ|Eff#{?eex?$N~b;FEVMjP(T2*%iDe-`+v|7m{y$1dn*6{002ovPDHLkV1lnB
B5rhB$

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-core-ops-0.1.0.0/tensorflow-core-ops.txt b/docs/haddock/tensorflow-core-ops-0.1.0.0/tensorflow-core-ops.txt
new file mode 100644
index 0000000..a68f2c8
--- /dev/null
+++ b/docs/haddock/tensorflow-core-ops-0.1.0.0/tensorflow-core-ops.txt
@@ -0,0 +1,4576 @@
+-- Hoogle documentation, generated by Haddock
+-- See Hoogle, http://www.haskell.org/hoogle/
+
+
+-- | Haskell wrappers for Core Tensorflow Ops.
+--   
+--   Code generated signatures for the Ops in libtensorflow_c.
+@package tensorflow-core-ops
+@version 0.1.0.0
+
+module TensorFlow.GenOps.Core
+
+-- | Receives the named tensor from send_device on recv_device.
+--   
+--   _HostRecv requires its input on host memory whereas _Recv requires its
+--   input on device memory.
+_HostRecv :: (TensorType tensor_type) => Int64 -> Tensor Value tensor_type
+
+-- | Receives the named tensor from send_device on recv_device.
+_Recv :: (TensorType tensor_type) => Int64 -> Tensor Value tensor_type
+
+-- | Sends the named tensor from send_device to recv_device.
+_Send :: (TensorType t) => Int64 -> Tensor v1 t -> ControlNode
+
+-- | A graph node which represents an argument to a function.
+_Arg :: (TensorType t) => Int64 -> Tensor Value t
+
+-- | Update '*var' according to the RMSProp algorithm.
+--   
+--   Note that in dense implement of this algorithm, ms and mom will update
+--   even if the grad is zero, but in this sparse implement, ms and mom
+--   will not update in iterations the grad is zero.
+--   
+--   mean_square = decay * mean_square + (1-decay) * gradient ** 2 Delta =
+--   learning_rate * gradient / sqrt(mean_square + epsilon)
+--   
+--   ms &lt;- rho * ms_{t-1} + (1-rho) * grad * grad mom &lt;- momentum *
+--   mom_{t-1} + lr * grad / sqrt(ms + epsilon) var &lt;- var - mom
+sparseApplyRMSProp :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 t -> Tensor v6 t -> Tensor v7 t -> Tensor v8 t -> Tensor v9 tindices -> Tensor Value t
+
+-- | Update '*var' according to the Adam algorithm.
+--   
+--   lr_t &lt;- learning_rate * sqrt(1 - beta2^t) / (1 - beta1^t) m_t &lt;-
+--   beta1 * m_{t-1} + (1 - beta1) * g_t v_t &lt;- beta2 * v_{t-1} + (1 -
+--   beta2) * g_t * g_t variable &lt;- variable - lr_t * m_t / (sqrt(v_t) +
+--   epsilon)
+applyAdam :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 t -> Tensor v6 t -> Tensor v7 t -> Tensor v8 t -> Tensor v9 t -> Tensor v10 t -> Tensor Value t
+
+-- | Update relevant entries in '*var' and '*accum' according to the
+--   momentum scheme.
+--   
+--   Set use_nesterov = True if you want to use Nesterov momentum.
+--   
+--   That is for rows we have grad for, we update var and accum as follows:
+--   
+--   accum = accum * momentum + grad var -= lr * accum
+sparseApplyMomentum :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 tindices -> Tensor v6 t -> Tensor Value t
+
+-- | Update '*var' according to the momentum scheme. Set use_nesterov =
+--   True if you
+--   
+--   want to use Nesterov momentum.
+--   
+--   accum = accum * momentum + grad var -= lr * accum
+applyMomentum :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 t -> Tensor Value t
+
+-- | Update '*var' according to the Ftrl-proximal scheme.
+--   
+--   accum_new = accum + grad * grad linear += grad +
+--   (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var quadratic = 1.0
+--   / (accum_new^(lr_power) * lr) + 2 * l2 var = (sign(linear) * l1 -
+--   linear) / quadratic if |linear| &gt; l1 else 0.0 accum = accum_new
+applyFtrl :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 t -> Tensor v6 t -> Tensor v7 t -> Tensor v8 t -> Tensor Value t
+
+-- | Update entries in '*var' and '*accum' according to the proximal
+--   adagrad scheme.
+sparseApplyAdagradDA :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 tindices -> Tensor v6 t -> Tensor v7 t -> Tensor v8 t -> Tensor v9 Int64 -> Tensor Value t
+
+-- | Update relevant entries in '*var' and '*accum' according to the
+--   adagrad scheme.
+--   
+--   That is for rows we have grad for, we update var and accum as follows:
+--   accum += grad * grad var -= lr * grad * (1 / sqrt(accum))
+sparseApplyAdagrad :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 tindices -> Tensor Value t
+
+-- | Update '*var' and '*accum' according to FOBOS with Adagrad learning
+--   rate.
+--   
+--   accum += grad * grad prox_v = var - lr * grad * (1 / sqrt(accum)) var
+--   = sign(prox_v)/(1+lr*l2) * max{|prox_v|-lr*l1,0}
+applyProximalAdagrad :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 t -> Tensor v6 t -> Tensor Value t
+
+-- | Update '*var' according to the adagrad scheme.
+--   
+--   accum += grad * grad var -= lr * grad * (1 / sqrt(accum))
+applyAdagrad :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor Value t
+
+-- | Update '*var' according to the adadelta scheme.
+--   
+--   accum = rho() * accum + (1 - rho()) * grad.square(); update =
+--   (update_accum + epsilon).sqrt() * (accum + epsilon()).rsqrt() * grad;
+--   update_accum = rho() * update_accum + (1 - rho()) * update.square();
+--   var -= update;
+applyAdadelta :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 t -> Tensor v6 t -> Tensor v7 t -> Tensor Value t
+
+-- | Sparse update '*var' as FOBOS algorithm with fixed learning rate.
+--   
+--   That is for rows we have grad for, we update var as follows: prox_v =
+--   var - alpha * grad var = sign(prox_v)/(1+alpha*l2) *
+--   max{|prox_v|-alpha*l1,0}
+sparseApplyProximalGradientDescent :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 t -> Tensor v6 tindices -> Tensor Value t
+
+-- | Update '*var' as FOBOS algorithm with fixed learning rate.
+--   
+--   prox_v = var - alpha * delta var = sign(prox_v)/(1+alpha*l2) *
+--   max{|prox_v|-alpha*l1,0}
+applyProximalGradientDescent :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 t -> Tensor Value t
+
+-- | Encode strings into web-safe base64 format.
+--   
+--   Refer to the following article for more information on base64 format:
+--   en.wikipedia.org<i>wiki</i>Base64. Base64 strings may have padding
+--   with '=' at the end so that the encoded has length multiple of 4. See
+--   Padding section of the link above.
+--   
+--   Web-safe means that the encoder uses - and _ instead of + and /.
+encodeBase64 :: Tensor v1 ByteString -> Tensor Value ByteString
+
+-- | Split elements of <tt>input</tt> based on <tt>delimiter</tt> into a
+--   <tt>SparseTensor</tt>.
+--   
+--   Let N be the size of source (typically N will be the batch size).
+--   Split each element of <tt>input</tt> based on <tt>delimiter</tt> and
+--   return a <tt>SparseTensor</tt> containing the splitted tokens. Empty
+--   tokens are ignored.
+--   
+--   <tt>delimiter</tt> can be empty or a single character. If
+--   <tt>delimiter</tt> is an empty string, each element of <tt>input</tt>
+--   is split into individual 1 character strings.
+--   
+--   For example: N = 2, input[0] is 'hello world' and input[1] is 'a b c',
+--   then the output will be
+--   
+--   indices = [0, 0; 0, 1; 1, 0; 1, 1; 1, 2] shape = [2, 3] values =
+--   [<tt>hello</tt>, <tt>world</tt>, <tt>a</tt>, <tt>b</tt>, <tt>c</tt>]
+stringSplit :: Tensor v1 ByteString -> Tensor v2 ByteString -> (Tensor Value Int64, Tensor Value ByteString, Tensor Value Int64)
+
+-- | Joins the strings in the given list of string tensors into one tensor;
+--   
+--   with the given separator (default is an empty separator).
+stringJoin :: [Tensor v1 ByteString] -> Tensor Value ByteString
+
+-- | Converts each entry in the given tensor to strings. Supports many
+--   numeric
+--   
+--   types and boolean.
+asString :: (TensorType t, OneOf '[Complex Float, Bool, Int32, Int64, Int8, Double, Float] t) => Tensor v1 t -> Tensor Value ByteString
+
+-- | Converts each string in the input Tensor to its hash mod by a number
+--   of buckets.
+--   
+--   The hash function is deterministic on the content of the string within
+--   the process. The hash function is a keyed hash function, where
+--   attribute <tt>key</tt> defines the key of the hash function.
+--   <tt>key</tt> is an array of 2 elements.
+--   
+--   A strong hash is important when inputs may be malicious, e.g. URLs
+--   with additional components. Adversaries could try to make their inputs
+--   hash to the same bucket for a denial-of-service attack or to skew the
+--   results. A strong hash prevents this by making it dificult, if not
+--   infeasible, to compute inputs that hash to the same bucket. This comes
+--   at a cost of roughly 4x higher compute time than
+--   tf.string_to_hash_bucket_fast.
+stringToHashBucketStrong :: Int64 -> Tensor v1 ByteString -> Tensor Value Int64
+
+-- | Multiplies sparse updates into a variable reference.
+--   
+--   This operation computes
+--   
+--   # Scalar indices ref[indices, ...] *= updates[...]
+--   
+--   # Vector indices (for each i) ref[indices[i], ...] *= updates[i, ...]
+--   
+--   # High rank indices (for each i, ..., j) ref[indices[i, ..., j], ...]
+--   *= updates[i, ..., j, ...]
+--   
+--   This operation outputs <tt>ref</tt> after the update is done. This
+--   makes it easier to chain operations that need to use the reset value.
+--   
+--   Duplicate entries are handled correctly: if multiple <tt>indices</tt>
+--   reference the same location, their contributions multiply.
+--   
+--   Requires `updates.shape = indices.shape + ref.shape[1:]`.
+scatterMul :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 tindices -> Tensor v3 t -> Tensor Value t
+
+-- | Joins a string Tensor across the given dimensions.
+--   
+--   Computes the string join across dimensions in the given string Tensor
+--   of shape `[d_0, d_1, ..., d_n-1]`. Returns a new Tensor created by
+--   joining the input strings with the given separator (default: empty
+--   string). Negative indices are counted backwards from the end, with
+--   `-1` being equivalent to `n - 1`. Passing an empty
+--   <tt>reduction_indices</tt> joins all strings in linear index order and
+--   outputs a scalar string.
+--   
+--   For example:
+--   
+--   ``` # tensor <tt>a</tt> is [["a", "b"], ["c", "d"]] tf.reduce_join(a,
+--   0) ==&gt; ["ac", "bd"] tf.reduce_join(a, 1) ==&gt; ["ab", "cd"]
+--   tf.reduce_join(a, -2) = tf.reduce_join(a, 0) ==&gt; ["ac", "bd"]
+--   tf.reduce_join(a, -1) = tf.reduce_join(a, 1) ==&gt; ["ab", "cd"]
+--   tf.reduce_join(a, 0, keep_dims=True) ==&gt; [["ac", "bd"]]
+--   tf.reduce_join(a, 1, keep_dims=True) ==&gt; [["ab"], ["cd"]]
+--   tf.reduce_join(a, 0, separator=".") ==&gt; ["a.c", "b.d"]
+--   tf.reduce_join(a, [0, 1]) ==&gt; ["acbd"] tf.reduce_join(a, [1, 0])
+--   ==&gt; ["abcd"] tf.reduce_join(a, []) ==&gt; ["abcd"] ```
+reduceJoin :: Tensor v1 ByteString -> Tensor v2 Int32 -> Tensor Value ByteString
+
+-- | Subtracts sparse updates to a variable reference.
+--   
+--   # Scalar indices ref[indices, ...] -= updates[...]
+--   
+--   # Vector indices (for each i) ref[indices[i], ...] -= updates[i, ...]
+--   
+--   # High rank indices (for each i, ..., j) ref[indices[i, ..., j], ...]
+--   -= updates[i, ..., j, ...]
+--   
+--   This operation outputs <tt>ref</tt> after the update is done. This
+--   makes it easier to chain operations that need to use the reset value.
+--   
+--   Duplicate entries are handled correctly: if multiple <tt>indices</tt>
+--   reference the same location, their (negated) contributions add.
+--   
+--   Requires `updates.shape = indices.shape + ref.shape[1:]`.
+--   
+--   <a>style="width:70%; margin:auto; margin-bottom:10px;
+--   margin-top:20px;"</a> <a>style="width:100%"
+--   src="../../images/ScatterSub.png" alt</a> <a>/div</a>
+scatterSub :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 tindices -> Tensor v3 t -> Tensor Value t
+
+-- | Adds sparse updates to a variable reference.
+--   
+--   This operation computes
+--   
+--   # Scalar indices ref[indices, ...] += updates[...]
+--   
+--   # Vector indices (for each i) ref[indices[i], ...] += updates[i, ...]
+--   
+--   # High rank indices (for each i, ..., j) ref[indices[i, ..., j], ...]
+--   += updates[i, ..., j, ...]
+--   
+--   This operation outputs <tt>ref</tt> after the update is done. This
+--   makes it easier to chain operations that need to use the reset value.
+--   
+--   Duplicate entries are handled correctly: if multiple <tt>indices</tt>
+--   reference the same location, their contributions add.
+--   
+--   Requires `updates.shape = indices.shape + ref.shape[1:]`.
+--   
+--   <a>style="width:70%; margin:auto; margin-bottom:10px;
+--   margin-top:20px;"</a> <a>style="width:100%"
+--   src="../../images/ScatterAdd.png" alt</a> <a>/div</a>
+scatterAdd :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 tindices -> Tensor v3 t -> Tensor Value t
+
+-- | Applies sparse updates to a variable reference.
+--   
+--   This operation computes
+--   
+--   # Scalar indices ref[indices, ...] = updates[...]
+--   
+--   # Vector indices (for each i) ref[indices[i], ...] = updates[i, ...]
+--   
+--   # High rank indices (for each i, ..., j) ref[indices[i, ..., j], ...]
+--   = updates[i, ..., j, ...]
+--   
+--   This operation outputs <tt>ref</tt> after the update is done. This
+--   makes it easier to chain operations that need to use the reset value.
+--   
+--   If values in <tt>ref</tt> is to be updated more than once, because
+--   there are duplicate entires in <tt>indices</tt>, the order at which
+--   the updates happen for each value is undefined.
+--   
+--   Requires `updates.shape = indices.shape + ref.shape[1:]`.
+--   
+--   <a>style="width:70%; margin:auto; margin-bottom:10px;
+--   margin-top:20px;"</a> <a>style="width:100%"
+--   src="../../images/ScatterUpdate.png" alt</a> <a>/div</a>
+scatterUpdate :: (TensorType t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 tindices -> Tensor v3 t -> Tensor Value t
+
+-- | Update <tt>ref</tt> by subtracting <a>value</a> from it.
+--   
+--   This operation outputs "ref" after the update is done. This makes it
+--   easier to chain operations that need to use the reset value.
+assignSub :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Update <tt>ref</tt> by adding <a>value</a> to it.
+--   
+--   This operation outputs "ref" after the update is done. This makes it
+--   easier to chain operations that need to use the reset value.
+assignAdd :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Computes gradients for SparseSegmentMean.
+--   
+--   Returns tensor "output" with same shape as grad, except for dimension
+--   0 whose value is output_dim0.
+sparseSegmentMeanGrad :: (TensorType t, OneOf '[Double, Float] t, TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 t -> Tensor v2 tidx -> Tensor v3 Int32 -> Tensor v4 Int32 -> Tensor Value t
+
+-- | Applies softmax to a batched N-D <tt>SparseTensor</tt>.
+--   
+--   The inputs represent an N-D SparseTensor with logical shape `[..., B,
+--   C]` (where `N &gt;= 2`), and with indices sorted in the canonical
+--   lexicographic order.
+--   
+--   This op is equivalent to applying the normal `tf.nn.softmax()` to each
+--   innermost logical submatrix with shape `[B, C]`, but with the catch
+--   that *the implicitly zero elements do not participate*. Specifically,
+--   the algorithm is equivalent to the following:
+--   
+--   <ol>
+--   <li>Applies `tf.nn.softmax()` to a densified view of each innermost
+--   submatrix with shape `[B, C]`, along the size-C dimension;</li>
+--   <li>Masks out the original implicitly-zero locations;</li>
+--   <li>Renormalizes the remaining elements.</li>
+--   </ol>
+--   
+--   Hence, the <tt>SparseTensor</tt> result has exactly the same non-zero
+--   indices and shape.
+sparseSoftmax :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 Int64 -> Tensor v2 t -> Tensor v3 Int64 -> Tensor Value t
+
+-- | Solves systems of linear equations.
+--   
+--   <tt>Matrix</tt> is a tensor of shape `[..., M, M]` whose inner-most 2
+--   dimensions form square matrices. <tt>Rhs</tt> is a tensor of shape
+--   `[..., M, K]`. The <tt>output</tt> is a tensor shape `[..., M, K]`. If
+--   <tt>adjoint</tt> is <a>False</a> then each output matrix satisfies
+--   `matrix[..., :, :] * output[..., :, :] = rhs[..., :, :]`. If
+--   <tt>adjoint</tt> is <a>True</a> then each output matrix satisfies
+--   `adjoint(matrix[..., :, :]) * output[..., :, :] = rhs[..., :, :]`.
+matrixSolve :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Computes the eigen decomposition of one or more square self-adjoint
+--   matrices.
+--   
+--   Computes the eigenvalues and (optionally) eigenvectors of each inner
+--   matrix in <tt>input</tt> such that `input[..., :, :] = v[..., :, :] *
+--   diag(e[..., :])`.
+--   
+--   ```prettyprint # a is a tensor. # e is a tensor of eigenvalues. # v is
+--   a tensor of eigenvectors. e, v = self_adjoint_eig(a) e =
+--   self_adjoint_eig(a, compute_v=False) ```
+selfAdjointEigV2 :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> (Tensor Value t, Tensor Value t)
+
+-- | Computes the Eigen Decomposition of a batch of square self-adjoint
+--   matrices.
+--   
+--   The input is a tensor of shape `[..., M, M]` whose inner-most 2
+--   dimensions form square matrices, with the same constraints as the
+--   single matrix SelfAdjointEig.
+--   
+--   The result is a [..., M+1, M] matrix with [..., 0,:] containing the
+--   eigenvalues, and subsequent [...,1:, :] containing the eigenvectors.
+selfAdjointEig :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Update '*var' by subtracting <tt>alpha</tt> * <tt>delta</tt> from it.
+applyGradientDescent :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor Value t
+
+-- | Push an element onto the stack.
+stackPush :: (TensorType t) => Tensor v1 ByteString -> Tensor v2 t -> Tensor Value t
+
+-- | Computes the Cholesky decomposition of one or more square matrices.
+--   
+--   The input is a tensor of shape `[..., M, M]` whose inner-most 2
+--   dimensions form square matrices, with the same constraints as the
+--   single matrix Cholesky decomposition above. The output is a tensor of
+--   the same shape as the input containing the Cholesky decompositions for
+--   all input submatrices `[..., :, :]`.
+cholesky :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Interleave the values from the `data` tensors into a single tensor.
+--   
+--   Builds a merged tensor such that
+--   
+--   merged[indices[m][i, ..., j], ...] = data[m][i, ..., j, ...]
+--   
+--   For example, if each `indices[m]` is scalar or vector, we have
+--   
+--   # Scalar indices merged[indices[m], ...] = data[m][...]
+--   
+--   # Vector indices merged[indices[m][i], ...] = data[m][i, ...]
+--   
+--   Each `data[i].shape` must start with the corresponding
+--   `indices[i].shape`, and the rest of `data[i].shape` must be constant
+--   w.r.t. <tt>i</tt>. That is, we must have `data[i].shape =
+--   indices[i].shape + constant`. In terms of this <tt>constant</tt>, the
+--   output shape is
+--   
+--   merged.shape = [max(indices)] + constant
+--   
+--   Values are merged in order, so if an index appears in both
+--   `indices[m][i]` and `indices[n][j]` for `(m,i) &lt; (n,j)` the slice
+--   `data[n][j]` will appear in the merged result.
+--   
+--   For example:
+--   
+--   indices[0] = 6 indices[1] = [4, 1] indices[2] = [[5, 2], [0, 3]]
+--   data[0] = [61, 62] data[1] = [[41, 42], [11, 12]] data[2] = [[[51,
+--   52], [21, 22]], [[1, 2], [31, 32]]] merged = [[1, 2], [11, 12], [21,
+--   22], [31, 32], [41, 42], [51, 52], [61, 62]]
+--   
+--   <a>style="width:70%; margin:auto; margin-bottom:10px;
+--   margin-top:20px;"</a> <a>style="width:100%"
+--   src="../../images/DynamicStitch.png" alt</a> <a>/div</a>
+dynamicStitch :: (TensorType t) => [Tensor v1 Int32] -> [Tensor v2 t] -> Tensor Value t
+
+-- | Returns the number of work units this Reader has finished processing.
+readerNumWorkUnitsCompleted :: Tensor v1 ByteString -> Tensor Value Int64
+
+-- | Returns the next record (key, value pair) produced by a Reader.
+--   
+--   Will dequeue from the input queue if necessary (e.g. when the Reader
+--   needs to start reading from a new file since it has finished with the
+--   previous file).
+readerRead :: Tensor v1 ByteString -> Tensor v2 ByteString -> (Tensor Value ByteString, Tensor Value ByteString)
+
+-- | Compute the 2-dimensional discrete Fourier Transform over the
+--   inner-most
+--   
+--   2 dimensions of <tt>input</tt>.
+fFT2D :: Tensor v1 (Complex Float) -> Tensor Value (Complex Float)
+
+-- | A Reader that outputs fixed-length records from a file.
+fixedLengthRecordReader :: Int64 -> Tensor Value ByteString
+
+-- | A placeholder op for a value that will be fed into the computation.
+--   
+--   N.B. This operation will fail with an error if it is executed. It is
+--   intended as a way to represent a value that will always be fed, and to
+--   provide attrs that enable the fed value to be checked at runtime.
+placeholder :: (TensorType dtype) => Tensor Value dtype
+
+-- | Outputs a <tt>Summary</tt> protocol buffer with scalar values.
+--   
+--   The input <tt>tags</tt> and <tt>values</tt> must have the same shape.
+--   The generated summary has a summary value for each tag-value pair in
+--   <tt>tags</tt> and <tt>values</tt>.
+scalarSummary :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 ByteString -> Tensor v2 t -> Tensor Value ByteString
+
+-- | Computes softmax activations.
+--   
+--   For each batch <tt>i</tt> and class <tt>j</tt> we have
+--   
+--   softmax[i, j] = exp(logits[i, j]) / sum_j(exp(logits[i, j]))
+softmax :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Generate a sharded filename. The filename is printf formatted as
+--   
+--   %s-%05d-of-%05d, basename, shard, num_shards.
+shardedFilename :: Tensor v1 ByteString -> Tensor v2 Int32 -> Tensor v3 Int32 -> Tensor Value ByteString
+
+-- | Sends the named tensor from send_device to recv_device.
+--   
+--   _HostSend requires its input on host memory whereas _Send requires its
+--   input on device memory.
+_HostSend :: (TensorType t) => Int64 -> Tensor v1 t -> ControlNode
+
+-- | Computes the gradient of the sigmoid of <tt>x</tt> wrt its input.
+--   
+--   Specifically, `grad = dy * y * (1 - y)`, where `y = sigmoid(x)`, and
+--   <tt>dy</tt> is the corresponding input gradient.
+sigmoidGrad :: (TensorType t, OneOf '[Complex Double, Complex Float, Word16, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Greedily selects a subset of bounding boxes in descending order of
+--   score,
+--   
+--   pruning away boxes that have high intersection-over-union (IOU)
+--   overlap with previously selected boxes. Bounding boxes are supplied as
+--   [y1, x1, y2, x2], where (y1, x1) and (y2, x2) are the coordinates of
+--   any diagonal pair of box corners and the coordinates can be provided
+--   as normalized (i.e., lying in the interval [0, 1]) or absolute. Note
+--   that this algorithm is agnostic to where the origin is in the
+--   coordinate system. Note that this algorithm is invariant to orthogonal
+--   transformations and translations of the coordinate system; thus
+--   translating or reflections of the coordinate system result in the same
+--   boxes being selected by the algorithm.
+--   
+--   The output of this operation is a set of integers indexing into the
+--   input collection of bounding boxes representing the selected boxes.
+--   The bounding box coordinates corresponding to the selected indices can
+--   then be obtained using the tf.gather operation. For example:
+--   
+--   selected_indices = tf.image.non_max_suppression( boxes, scores,
+--   max_output_size, iou_threshold) selected_boxes = tf.gather(boxes,
+--   selected_indices)
+nonMaxSuppression :: Tensor v1 Float -> Tensor v2 Float -> Tensor v3 Int32 -> Tensor Value Int32
+
+-- | A Reader that outputs the queued work as both the key and value.
+--   
+--   To use, enqueue strings in a Queue. ReaderRead will take the front
+--   work string and output (work, work).
+identityReader :: Tensor Value ByteString
+
+-- | Extracts a glimpse from the input tensor.
+--   
+--   Returns a set of windows called glimpses extracted at location
+--   <tt>offsets</tt> from the input tensor. If the windows only partially
+--   overlaps the inputs, the non overlapping areas will be filled with
+--   random noise.
+--   
+--   The result is a 4-D tensor of shape `[batch_size, glimpse_height,
+--   glimpse_width, channels]`. The channels and batch dimensions are the
+--   same as that of the input tensor. The height and width of the output
+--   windows are specified in the <a>size</a> parameter.
+--   
+--   The argument <tt>normalized</tt> and <tt>centered</tt> controls how
+--   the windows are built:
+--   
+--   <ul>
+--   <li>If the coordinates are normalized but not centered, 0.0 and 1.0
+--   correspond to the minimum and maximum of each height and width
+--   dimension.</li>
+--   <li>If the coordinates are both normalized and centered, they range
+--   from</li>
+--   <li>1.0 to 1.0. The coordinates (-1.0, -1.0) correspond to the upper
+--   left corner, the lower right corner is located at (1.0, 1.0) and the
+--   center is at (0, 0).</li>
+--   <li>If the coordinates are not normalized they are interpreted as
+--   numbers of pixels.</li>
+--   </ul>
+extractGlimpse :: Tensor v1 Float -> Tensor v2 Int32 -> Tensor v3 Float -> Tensor Value Float
+
+-- | Computes the gradients of 3-D convolution with respect to the input.
+conv3DBackpropInput :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor Value t
+
+-- | Solves one or more linear least-squares problems.
+--   
+--   <tt>matrix</tt> is a tensor of shape `[..., M, N]` whose inner-most 2
+--   dimensions form matrices of size `[M, N]`. Rhs is a tensor of shape
+--   `[..., M, K]`. The output is a tensor shape `[..., N, K]` where each
+--   output matrix solves each of the equations matrix[..., :, :] *
+--   output[..., :, :] = rhs[..., :, :] in the least squares sense.
+--   
+--   matrix and right-hand sides in the batch:
+--   
+--   <tt>matrix</tt>=\(A in Re^{m times n}\), <tt>rhs</tt>=\(B in Re^{m
+--   times k}\), <tt>output</tt>=\(X in Re^{n times k}\),
+--   <tt>l2_regularizer</tt>=\(lambda\).
+--   
+--   If <tt>fast</tt> is <a>True</a>, then the solution is computed by
+--   solving the normal equations using Cholesky decomposition.
+--   Specifically, if \(m ge n\) then \(X = (A^T A + lambda I)^{-1} A^T
+--   B\), which solves the least-squares problem \(X = mathrm{argmin}_{Z in
+--   Re^{n times k}} ||A Z - B||_F^2 + lambda ||Z||_F^2\). If \(m lt n\)
+--   then <tt>output</tt> is computed as \(X = A^T (A A^T + lambda I)^{-1}
+--   B\), which (for \(lambda = 0\)) is the minimum-norm solution to the
+--   under-determined linear system, i.e. \(X = mathrm{argmin}_{Z in Re^{n
+--   times k}} ||Z||_F^2 \), subject to \(A Z = B\). Notice that the fast
+--   path is only numerically stable when \(A\) is numerically full rank
+--   and has a condition number \(mathrm{cond}(A) lt
+--   frac{1}{sqrt{epsilon_{mach}}}\) or\(lambda\) is sufficiently large.
+--   
+--   If <tt>fast</tt> is <a>False</a> an algorithm based on the numerically
+--   robust complete orthogonal decomposition is used. This computes the
+--   minimum-norm least-squares solution, even when \(A\) is rank
+--   deficient. This path is typically 6-7 times slower than the fast path.
+--   If <tt>fast</tt> is <a>False</a> then <tt>l2_regularizer</tt> is
+--   ignored.
+matrixSolveLs :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 Double -> Tensor Value t
+
+-- | Converts one or more images from RGB to HSV.
+--   
+--   Outputs a tensor of the same shape as the <tt>images</tt> tensor,
+--   containing the HSV value of the pixels. The output is only well
+--   defined if the value in <tt>images</tt> are in `[0,1]`.
+--   
+--   `output[..., 0]` contains hue, `output[..., 1]` contains saturation,
+--   and `output[..., 2]` contains value. All HSV values are in `[0,1]`. A
+--   hue of 0 corresponds to pure red, hue 1<i>3 is pure green, and 2</i>3
+--   is pure blue.
+rGBToHSV :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Decode the first frame of a GIF-encoded image to a uint8 tensor.
+--   
+--   GIF with frame or transparency compression are not supported convert
+--   animated GIF from compressed to uncompressed by:
+--   
+--   convert $src.gif -coalesce $dst.gif
+decodeGif :: Tensor v1 ByteString -> Tensor Value Word8
+
+-- | Deprecated. Disallowed in GraphDef version &gt;= 2.
+adjustContrast :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 Float -> Tensor v3 Float -> Tensor v4 Float -> Tensor Value Float
+
+-- | DepthToSpace for tensors of type T.
+--   
+--   Rearranges data from depth into blocks of spatial data. This is the
+--   reverse transformation of SpaceToDepth. More specifically, this op
+--   outputs a copy of the input tensor where values from the
+--   <tt>depth</tt> dimension are moved in spatial blocks to the
+--   <tt>height</tt> and <tt>width</tt> dimensions. The attr
+--   <tt>block_size</tt> indicates the input block size and how the data is
+--   moved.
+--   
+--   <ul>
+--   <li>Chunks of data of size `block_size * block_size` from depth are
+--   rearranged into non-overlapping blocks of size `block_size x
+--   block_size`</li>
+--   <li>The width the output tensor is `input_depth * block_size`, whereas
+--   the height is `input_height * block_size`.</li>
+--   <li>The depth of the input tensor must be divisible by `block_size *
+--   block_size`.</li>
+--   </ul>
+--   
+--   That is, assuming the input is in the shape: `[batch, height, width,
+--   depth]`, the shape of the output will be: `[batch, height*block_size,
+--   width*block_size, depth/(block_size*block_size)]`
+--   
+--   This operation requires that the input tensor be of rank 4, and that
+--   <tt>block_size</tt> be &gt;=1 and that `block_size * block_size` be a
+--   divisor of the input depth.
+--   
+--   This operation is useful for resizing the activations between
+--   convolutions (but keeping all data), e.g. instead of pooling. It is
+--   also useful for training purely convolutional models.
+--   
+--   For example, given this input of shape `[1, 1, 1, 4]`, and a block
+--   size of 2:
+--   
+--   ```prettyprint x = [[[[1, 2, 3, 4]]]]
+--   
+--   ```
+--   
+--   This operation will output a tensor of shape `[1, 2, 2, 1]`:
+--   
+--   ```prettyprint [[[[1], [2]], [[3], [4]]]] ```
+--   
+--   Here, the input has a batch of 1 and each batch element has shape `[1,
+--   1, 4]`, the corresponding output will have 2x2 elements and will have
+--   a depth of 1 channel (1 = `4 / (block_size * block_size)`). The output
+--   element shape is `[2, 2, 1]`.
+--   
+--   For an input tensor with larger depth, here of shape `[1, 1, 1, 12]`,
+--   e.g.
+--   
+--   ```prettyprint x = [[[[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]]]] ```
+--   
+--   This operation, for block size of 2, will return the following tensor
+--   of shape `[1, 2, 2, 3]`
+--   
+--   ```prettyprint [[[[1, 2, 3], [4, 5, 6]], [[7, 8, 9], [10, 11, 12]]]]
+--   
+--   ```
+--   
+--   Similarly, for the following input of shape `[1 2 2 4]`, and a block
+--   size of 2:
+--   
+--   ```prettyprint x = [[[[1, 2, 3, 4], [5, 6, 7, 8]], [[9, 10, 11, 12],
+--   [13, 14, 15, 16]]]] ```
+--   
+--   the operator will return the following tensor of shape `[1 4 4 1]`:
+--   
+--   ```prettyprint x = [[ [1], [2], [5], [6]], [ [3], [4], [7], [8]], [
+--   [9], [10], [13], [14]], [ [11], [12], [15], [16]]]
+--   
+--   ```
+depthToSpace :: (TensorType t) => Int64 -> Tensor v1 t -> Tensor Value t
+batchMatrixSolve :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Computes the complementary error function of <tt>x</tt> element-wise.
+erfc :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes the gradient of bilinear interpolation.
+resizeBilinearGrad :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 Float -> Tensor v2 t -> Tensor Value t
+
+-- | Output a fact about factorials.
+fact :: Tensor Value ByteString
+
+-- | Delete the tensor specified by its handle in the session.
+deleteSessionTensor :: Tensor v1 ByteString -> ControlNode
+
+-- | Returns the truth value of x OR y element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>LogicalOr</tt> supports broadcasting. More about
+--   broadcasting <a>here</a></li>
+--   </ul>
+logicalOr :: Tensor v1 Bool -> Tensor v2 Bool -> Tensor Value Bool
+
+-- | Get the value of the tensor specified by its handle.
+getSessionTensor :: (TensorType dtype) => Tensor v1 ByteString -> Tensor Value dtype
+batchMatrixInverse :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Generate a glob pattern matching all sharded file names.
+shardedFilespec :: Tensor v1 ByteString -> Tensor v2 Int32 -> Tensor Value ByteString
+
+-- | Decode web-safe base64-encoded strings.
+--   
+--   Input may or may not have padding at the end. See EncodeBase64 for
+--   padding. Web-safe means that input must use - and _ instead of + and
+--   /.
+decodeBase64 :: Tensor v1 ByteString -> Tensor Value ByteString
+
+-- | Store the input tensor in the state of the current session.
+getSessionHandle :: (TensorType t) => Tensor v1 t -> Tensor Value ByteString
+
+-- | Table initializer that takes two tensors for keys and values
+--   respectively.
+initializeTable :: (TensorType tkey, TensorType tval) => Tensor v1 ByteString -> Tensor v2 tkey -> Tensor v3 tval -> ControlNode
+
+-- | Computes tan of x element-wise.
+tan :: (TensorType t, OneOf '[Complex Double, Complex Float, Int32, Int64, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes hyperbolic tangent of <tt>x</tt> element-wise.
+tanh :: (TensorType t, OneOf '[Complex Double, Complex Float, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Update '*var' according to the proximal adagrad scheme.
+applyAdagradDA :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 t -> Tensor v6 t -> Tensor v7 t -> Tensor v8 Int64 -> Tensor Value t
+
+-- | Converts each string in the input Tensor to its hash mod by a number
+--   of buckets.
+--   
+--   The hash function is deterministic on the content of the string within
+--   the process.
+--   
+--   Note that the hash function may change from time to time. This
+--   functionality will be deprecated and it's recommended to use
+--   `tf.string_to_hash_bucket_fast()` or
+--   `tf.string_to_hash_bucket_strong()`.
+stringToHashBucket :: Int64 -> Tensor v1 ByteString -> Tensor Value Int64
+
+-- | Computes gradients for the exponential linear (Elu) operation.
+eluGrad :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Computes gradient of the FractionalAvgPool function.
+--   
+--   Unlike FractionalMaxPoolGrad, we don't need to find arg_max for
+--   FractionalAvgPoolGrad, we just need to evenly back-propagate each
+--   element of out_backprop to those indices that form the same pooling
+--   cell. Therefore, we just need to know the shape of original input
+--   tensor, instead of the whole tensor.
+fractionalAvgPoolGrad :: (TensorType t, OneOf '[Int32, Int64, Double, Float] t) => Tensor v1 Int64 -> Tensor v2 t -> Tensor v3 Int64 -> Tensor v4 Int64 -> Tensor Value t
+
+-- | Solves systems of linear equations with upper or lower triangular
+--   matrices by
+--   
+--   backsubstitution.
+--   
+--   <tt>matrix</tt> is a tensor of shape `[..., M, M]` whose inner-most 2
+--   dimensions form square matrices. If <tt>lower</tt> is <a>True</a> then
+--   the strictly upper triangular part of each inner-most matrix is
+--   assumed to be zero and not accessed. If <tt>lower</tt> is False then
+--   the strictly lower triangular part of each inner-most matrix is
+--   assumed to be zero and not accessed. <tt>rhs</tt> is a tensor of shape
+--   `[..., M, K]`.
+--   
+--   The output is a tensor of shape `[..., M, K]`. If <tt>adjoint</tt> is
+--   <a>True</a> then the innermost matrices in output` satisfy matrix
+--   equations `matrix[..., :, :] * output[..., :, :] = rhs[..., :, :]`. If
+--   <tt>adjoint</tt> is <a>False</a> then the strictly then the innermost
+--   matrices in <tt>output</tt> satisfy matrix equations
+--   `adjoint(matrix[..., i, k]) * output[..., k, j] = rhs[..., i, j]`.
+matrixTriangularSolve :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Computes the (possibly normalized) Levenshtein Edit Distance.
+--   
+--   The inputs are variable-length sequences provided by SparseTensors
+--   (hypothesis_indices, hypothesis_values, hypothesis_shape) and
+--   (truth_indices, truth_values, truth_shape).
+--   
+--   The inputs are:
+editDistance :: (TensorType t) => Tensor v1 Int64 -> Tensor v2 t -> Tensor v3 Int64 -> Tensor v4 Int64 -> Tensor v5 t -> Tensor v6 Int64 -> Tensor Value Float
+
+-- | Computes the number of incomplete elements in the given barrier.
+barrierIncompleteSize :: Tensor v1 ByteString -> Tensor Value Int32
+
+-- | Generates labels for candidate sampling with a learned unigram
+--   distribution.
+--   
+--   See explanations of candidate sampling and the data formats at
+--   go/candidate-sampling.
+--   
+--   For each batch, this op picks a single set of sampled candidate
+--   labels.
+--   
+--   The advantages of sampling candidates per-batch are simplicity and the
+--   possibility of efficient dense matrix multiplication. The disadvantage
+--   is that the sampled candidates must be chosen independently of the
+--   context and of the true labels.
+threadUnsafeUnigramCandidateSampler :: Int64 -> Int64 -> Int64 -> Bool -> Tensor v1 Int64 -> (Tensor Value Int64, Tensor Value Float, Tensor Value Float)
+
+-- | Computes the number of complete elements in the given barrier.
+barrierReadySize :: Tensor v1 ByteString -> Tensor Value Int32
+
+-- | Closes the given barrier.
+--   
+--   This operation signals that no more new elements will be inserted in
+--   the given barrier. Subsequent InsertMany that try to introduce a new
+--   key will fail. Subsequent InsertMany operations that just add missing
+--   components to already existing elements will continue to succeed.
+--   Subsequent TakeMany operations will continue to succeed if sufficient
+--   completed elements remain in the barrier. Subsequent TakeMany
+--   operations that would block will fail immediately.
+barrierClose :: Tensor v1 ByteString -> ControlNode
+
+-- | A Reader that outputs the lines of a file delimited by '\n'.
+textLineReader :: Tensor Value ByteString
+
+-- | Compute the 3-dimensional discrete Fourier Transform over the
+--   inner-most 3
+--   
+--   dimensions of <tt>input</tt>.
+fFT3D :: Tensor v1 (Complex Float) -> Tensor Value (Complex Float)
+
+-- | Exits the current frame to its parent frame.
+--   
+--   Exit makes its input `data` available to the parent frame.
+refExit :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes exponential of x element-wise. \(y = e^x\).
+exp :: (TensorType t, OneOf '[Complex Double, Complex Float, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Restores a tensor from checkpoint files.
+--   
+--   This is like <tt>Restore</tt> except that restored tensor can be
+--   listed as filling only a slice of a larger tensor.
+--   <tt>shape_and_slice</tt> specifies the shape of the larger tensor and
+--   the slice that the restored tensor covers.
+--   
+--   The <tt>shape_and_slice</tt> input has the same format as the elements
+--   of the <tt>shapes_and_slices</tt> input of the <tt>SaveSlices</tt> op.
+restoreSlice :: (TensorType dt) => Tensor v1 ByteString -> Tensor v2 ByteString -> Tensor v3 ByteString -> Tensor Value dt
+
+-- | Returns the complex conjugate of a complex number.
+--   
+--   Given a tensor <tt>input</tt> of complex numbers, this operation
+--   returns a tensor of complex numbers that are the complex conjugate of
+--   each element in <tt>input</tt>. The complex numbers in <tt>input</tt>
+--   must be of the form \(a + bj\), where *a* is the real part and *b* is
+--   the imaginary part.
+--   
+--   The complex conjugate returned by this operation is of the form \(a -
+--   bj\).
+--   
+--   For example:
+--   
+--   ``` # tensor <tt>input</tt> is [-2.25 + 4.75j, 3.25 + 5.75j]
+--   tf.conj(input) ==&gt; [-2.25 - 4.75j, 3.25 - 5.75j] ```
+conj :: (TensorType t, OneOf '[Complex Double, Complex Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes the gradient of nearest neighbor interpolation.
+resizeNearestNeighborGrad :: (TensorType t, OneOf '[Int32, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 Int32 -> Tensor Value t
+
+-- | Delete the TensorArray from its resource container. This enables
+--   
+--   the user to close and release the resource in the middle of a
+--   step/run.
+tensorArrayClose :: Tensor v1 ByteString -> ControlNode
+
+-- | Computes atan of x element-wise.
+atan :: (TensorType t, OneOf '[Complex Double, Complex Float, Int32, Int64, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Get the current size of the TensorArray.
+tensorArraySize :: Tensor v1 ByteString -> Tensor v2 Float -> Tensor Value Int32
+
+-- | Concat the elements from the TensorArray into value <a>value</a>.
+--   
+--   Takes <tt>T</tt> elements of shapes
+--   
+--   ``` (n0 x d0 x d1 x ...), (n1 x d0 x d1 x ...), ..., (n(T-1) x d0 x d1
+--   x ...) ```
+--   
+--   and concatenates them into a Tensor of shape:
+--   
+--   ```(n0 + n1 + ... + n(T-1) x d0 x d1 x ...)```
+--   
+--   All elements must have the same shape (excepting the first dimension).
+tensorArrayConcat :: (TensorType dtype) => Tensor v1 ByteString -> Tensor v2 Float -> (Tensor Value dtype, Tensor Value Int64)
+
+-- | Local Response Normalization.
+--   
+--   The 4-D <tt>input</tt> tensor is treated as a 3-D array of 1-D vectors
+--   (along the last dimension), and each vector is normalized
+--   independently. Within a given vector, each component is divided by the
+--   weighted, squared sum of inputs within <tt>depth_radius</tt>. In
+--   detail,
+--   
+--   sqr_sum[a, b, c, d] = sum(input[a, b, c, d - depth_radius : d +
+--   depth_radius + 1] ** 2) output = input / (bias + alpha * sqr_sum) **
+--   beta
+--   
+--   For details, see <a>Krizhevsky et al., ImageNet classification with
+--   deep convolutional neural networks (NIPS 2012)</a>.
+lRN :: (TensorType t, OneOf '[Word16, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Converts each string in the input Tensor to its hash mod by a number
+--   of buckets.
+--   
+--   The hash function is deterministic on the content of the string within
+--   the process and will never change. However, it is not suitable for
+--   cryptography. This function may be used when CPU time is scarce and
+--   inputs are trusted or unimportant. There is a risk of adversaries
+--   constructing inputs that all hash to the same bucket. To prevent this
+--   problem, use a strong hash function with
+--   `tf.string_to_hash_bucket_strong`.
+stringToHashBucketFast :: Int64 -> Tensor v1 ByteString -> Tensor Value Int64
+
+-- | Pack the elements from the TensorArray into output <a>value</a>.
+--   
+--   <ul>
+--   <li>*WARNING: This op is deprecated.**</li>
+--   </ul>
+--   
+--   Instead of this op, use <tt>TensorArrayGather</tt> with `indices =
+--   RangeOp(0, TensorArraySizeOp)`.
+--   
+--   All elements must have the same shape.
+tensorArrayPack :: (TensorType dtype) => Tensor v1 ByteString -> Tensor v2 Float -> Tensor Value dtype
+
+-- | Computes offsets of concat inputs within its output.
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>x</tt> is [2, 2, 7] # <tt>y</tt> is [2, 3, 7] #
+--   <tt>z</tt> is [2, 5, 7] concat_offset(2, [x, y, z]) =&gt; [0, 0, 0],
+--   [0, 2, 0], [0, 5, 0] ```
+concatOffset :: Tensor v1 Int32 -> [Tensor v2 Int32] -> [Tensor Value Int32]
+
+-- | Creates or finds a child frame, and makes `data` available to the
+--   child frame.
+--   
+--   The unique <tt>frame_name</tt> is used by the <tt>Executor</tt> to
+--   identify frames. If <tt>is_constant</tt> is true, <tt>output</tt> is a
+--   constant in the child frame; otherwise it may be changed in the child
+--   frame. At most <tt>parallel_iterations</tt> iterations are run in
+--   parallel in the child frame.
+refEnter :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes softsign: `features / (abs(features) + 1)`.
+softsign :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Push an element onto the tensor_array.
+tensorArrayWrite :: (TensorType t) => Tensor v1 ByteString -> Tensor v2 Int32 -> Tensor v3 t -> Tensor v4 Float -> Tensor Value Float
+
+-- | Returns a diagonal tensor with a given diagonal values.
+--   
+--   Given a <tt>diagonal</tt>, this operation returns a tensor with the
+--   <tt>diagonal</tt> and everything else padded with zeros. The diagonal
+--   is computed as follows:
+--   
+--   Assume <tt>diagonal</tt> has dimensions [D1,..., Dk], then the output
+--   is a tensor of rank 2k with dimensions [D1,..., Dk, D1,..., Dk] where:
+--   
+--   `output[i1,..., ik, i1,..., ik] = diagonal[i1, ..., ik]` and 0
+--   everywhere else.
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>diagonal</tt> is [1, 2, 3, 4] tf.diag(diagonal)
+--   ==&gt; [[1, 0, 0, 0] [0, 2, 0, 0] [0, 0, 3, 0] [0, 0, 0, 4]] ```
+diag :: (TensorType t, OneOf '[Complex Double, Complex Float, Int32, Int64, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Returns the batched diagonal part of a batched tensor.
+--   
+--   This operation returns a tensor with the <tt>diagonal</tt> part of the
+--   batched <tt>input</tt>. The <tt>diagonal</tt> part is computed as
+--   follows:
+--   
+--   Assume <tt>input</tt> has <tt>k</tt> dimensions `[I, J, K, ..., N,
+--   N]`, then the output is a tensor of rank `k - 1` with dimensions `[I,
+--   J, K, ..., N]` where:
+--   
+--   `diagonal[i, j, k, ..., n] = input[i, j, k, ..., n, n]`.
+--   
+--   The input must be at least a matrix.
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>input</tt> is [[[1, 0, 0, 0] [0, 2, 0, 0] [0, 0,
+--   3, 0] [0, 0, 0, 4]], [[5, 0, 0, 0] [0, 6, 0, 0] [0, 0, 7, 0] [0, 0, 0,
+--   8]]]
+--   
+--   and input.shape = (2, 4, 4)
+--   
+--   tf.matrix_diag_part(input) ==&gt; [[1, 2, 3, 4], [5, 6, 7, 8]]
+--   
+--   which has shape (2, 4) ```
+matrixDiagPart :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes the number of elements in the given queue.
+queueSize :: Tensor v1 ByteString -> Tensor Value Int32
+
+-- | Decode a PNG-encoded image to a uint8 or uint16 tensor.
+--   
+--   The attr <tt>channels</tt> indicates the desired number of color
+--   channels for the decoded image.
+--   
+--   Accepted values are:
+--   
+--   <ul>
+--   <li>0: Use the number of channels in the PNG-encoded image.</li>
+--   <li>1: output a grayscale image.</li>
+--   <li>3: output an RGB image.</li>
+--   <li>4: output an RGBA image.</li>
+--   </ul>
+--   
+--   If needed, the PNG-encoded image is transformed to match the requested
+--   number of color channels.
+decodePng :: (TensorType dtype, OneOf '[Word16, Word8] dtype) => Tensor v1 ByteString -> Tensor Value dtype
+
+-- | Returns element-wise smallest integer in not less than x.
+ceil :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | A queue that produces elements sorted by the first component value.
+--   
+--   Note that the PriorityQueue requires the first component of any
+--   element to be a scalar int64, in addition to the other elements
+--   declared by component_types. Therefore calls to Enqueue and
+--   EnqueueMany (resp. Dequeue and DequeueMany) on a PriorityQueue will
+--   all require (resp. output) one extra entry in their input (resp.
+--   output) lists.
+priorityQueue :: Tensor Value ByteString
+
+-- | A placeholder op that passes though <tt>input</tt> when its output is
+--   not fed.
+placeholderWithDefault :: (TensorType dtype) => Tensor v1 dtype -> Tensor Value dtype
+
+-- | Computes the gradient of the crop_and_resize op wrt the input image
+--   tensor.
+cropAndResizeGradImage :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 Float -> Tensor v2 Float -> Tensor v3 Int32 -> Tensor v4 Int32 -> Tensor Value t
+
+-- | Restore a Reader to its initial clean state.
+readerReset :: Tensor v1 ByteString -> ControlNode
+
+-- | Extract <tt>patches</tt> from <tt>images</tt> and put them in the
+--   "depth" output dimension.
+extractImagePatches :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor Value t
+batchMatrixSetDiag :: (TensorType t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Delete the stack from its resource container.
+stackClose :: Tensor v1 ByteString -> ControlNode
+
+-- | Quantizes then dequantizes a tensor.
+--   
+--   This op simulates the precision loss from the quantized forward pass
+--   by: 1. Quantizing the tensor to fixed point numbers, which should
+--   match the target quantization method when it is used in inference. 2.
+--   Dequantizing it back to floating point numbers for the following ops,
+--   most likely matmul.
+--   
+--   There are different ways to quantize. This version does not use the
+--   full range of the output type, choosing to elide the lowest possible
+--   value for symmetry (e.g., output range is -127 to 127, not -128 to 127
+--   for signed 8 bit quantization), so that 0.0 maps to 0.
+--   
+--   To perform this op, we first find the range of values in our tensor.
+--   The range we use is always centered on 0, so we find m such that
+--   
+--   <ol>
+--   <li>m = max(abs(input_min), abs(input_max)) if range_given is
+--   true,</li>
+--   <li>m = max(max(abs(min_elem(input)), abs(max_elem(input)))
+--   otherwise.</li>
+--   </ol>
+--   
+--   Our input tensor range is then [-m, m].
+--   
+--   Next, we choose our fixed-point quantization buckets, [min_fixed,
+--   max_fixed]. If signed_input is true, this is
+--   
+--   <ul>
+--   <li><i>min_fixed, max_fixed </i> =</li>
+--   <li><i>-(1 &lt;&lt; (num_bits - 1) - 1), (1 &lt;&lt; (num_bits - 1)) -
+--   1</i> .</li>
+--   </ul>
+--   
+--   Otherwise, if signed_input is false, the fixed-point range is
+--   
+--   <ul>
+--   <li><i>min_fixed, max_fixed</i> = [0, (1 &lt;&lt; num_bits) - 1].</li>
+--   </ul>
+--   
+--   From this we compute our scaling factor, s:
+--   
+--   s = (max_fixed - min_fixed) / (2 * m).
+--   
+--   Now we can quantize and dequantize the elements of our tensor. An
+--   element e is transformed into e':
+--   
+--   e' = (e * s).round_to_nearest() / s.
+--   
+--   Note that we have a different number of buckets in the signed vs.
+--   unsigned cases. For example, if num_bits == 8, we get 254 buckets in
+--   the signed case vs. 255 in the unsigned case.
+--   
+--   For example, suppose num_bits = 8 and m = 1. Then
+--   
+--   <ul>
+--   <li><i>min_fixed, max_fixed</i> = [-127, 127], and s = (127 + 127) / 2
+--   = 127.</li>
+--   </ul>
+--   
+--   Given the vector {-1, -0.5, 0, 0.3}, this is quantized to {-127, -63,
+--   0, 38}, and dequantized to {-1, -63.0<i>127, 0, 38.0</i>127}.
+quantizeAndDequantize :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Returns which elements of x are NaN.
+isNan :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor Value Bool
+
+-- | Returns locations of true values in a boolean tensor.
+--   
+--   This operation returns the coordinates of true elements in
+--   <tt>input</tt>. The coordinates are returned in a 2-D tensor where the
+--   first dimension (rows) represents the number of true elements, and the
+--   second dimension (columns) represents the coordinates of the true
+--   elements. Keep in mind, the shape of the output tensor can vary
+--   depending on how many true values there are in <tt>input</tt>. Indices
+--   are output in row-major order.
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>input</tt> tensor is [[True, False] # [True,
+--   False]] # <tt>input</tt> has two true values, so output has two
+--   coordinates. # <tt>input</tt> has rank of 2, so coordinates have two
+--   indices. where(input) ==&gt; [[0, 0], [1, 0]]
+--   
+--   # <tt>input</tt> tensor is [[[True, False] # [True, False]] # [[False,
+--   True] # [False, True]] # [[False, False] # [False, True]]] #
+--   <tt>input</tt> has 5 true values, so output has 5 coordinates. #
+--   <tt>input</tt> has rank of 3, so coordinates have three indices.
+--   where(input) ==&gt; [[0, 0, 0], [0, 1, 0], [1, 0, 1], [1, 1, 1], [2,
+--   1, 1]] ```
+where' :: Tensor v1 Bool -> Tensor Value Int64
+
+-- | Computes the difference between two lists of numbers or strings.
+--   
+--   Given a list <tt>x</tt> and a list <tt>y</tt>, this operation returns
+--   a list <tt>out</tt> that represents all values that are in <tt>x</tt>
+--   but not in <tt>y</tt>. The returned list <tt>out</tt> is sorted in the
+--   same order that the numbers appear in <tt>x</tt> (duplicates are
+--   preserved). This operation also returns a list <tt>idx</tt> that
+--   represents the position of each <tt>out</tt> element in <tt>x</tt>. In
+--   other words:
+--   
+--   `out[i] = x[idx[i]] for i in [0, 1, ..., len(out) - 1]`
+--   
+--   For example, given this input:
+--   
+--   ```prettyprint x = [1, 2, 3, 4, 5, 6] y = [1, 3, 5] ```
+--   
+--   This operation would return:
+--   
+--   ```prettyprint out ==&gt; [2, 4, 6] idx ==&gt; [1, 3, 5] ```
+listDiff :: (TensorType t, TensorType out_idx, OneOf '[Int32, Int64] out_idx) => Tensor v1 t -> Tensor v2 t -> (Tensor Value t, Tensor Value out_idx)
+
+-- | Return a strided slice from <tt>input</tt>.
+--   
+--   The output tensor is a tensor with dimensions implied by
+--   <tt>begin</tt>, <tt>end</tt>, and <tt>strides</tt>, whose values are
+--   extracted from <tt>begin</tt>.
+--   
+--   Specifically, the result tensor at index `(i[0], i[1], ..., i[n-1])`
+--   will obtain the value `input[begin[0] + i[0] * stride[0], ..., `
+--   `begin[n-1] + i[n-1] * stride[n-1])]`.
+--   
+--   <ul>
+--   <li>Requirements*: `0 != strides[i] for i in [0, n)`</li>
+--   </ul>
+stridedSlice :: (TensorType index, OneOf '[Int32, Int64] index, TensorType t) => Tensor v1 t -> Tensor v2 index -> Tensor v3 index -> Tensor v4 index -> Tensor Value t
+
+-- | A queue that randomizes the order of elements.
+randomShuffleQueue :: Tensor Value ByteString
+
+-- | Returns the gradient of <tt>Tile</tt>.
+--   
+--   Since <tt>Tile</tt> takes an input and repeats the input
+--   <tt>multiples</tt> times along each dimension, <tt>TileGrad</tt> takes
+--   in <tt>multiples</tt> and aggregates each repeated tile of
+--   <tt>input</tt> into <tt>output</tt>.
+tileGrad :: (TensorType t) => Tensor v1 t -> Tensor v2 Int32 -> Tensor Value t
+
+-- | Assign <a>value</a> to the sliced l-value reference of <tt>ref</tt>.
+--   
+--   The values of <a>value</a> are assigned to the positions in the
+--   variable <tt>ref</tt> that are selected by the slice parameters. The
+--   slice parameters `begin, <tt>end</tt>, <tt>strides</tt>, etc. work
+--   exactly as in <tt>StridedSlice</tt>.
+--   
+--   NOTE this op currently does not support broadcasting and so
+--   <a>value</a>'s shape must be exactly the shape produced by the slice
+--   of <tt>ref</tt>.
+stridedSliceAssign :: (TensorType index, OneOf '[Int32, Int64] index, TensorType t) => Tensor v1 t -> Tensor v2 index -> Tensor v3 index -> Tensor v4 index -> Tensor v5 t -> Tensor Value t
+
+-- | Reshapes a tensor.
+--   
+--   Given <tt>tensor</tt>, this operation returns a tensor that has the
+--   same values as <tt>tensor</tt> with shape <a>shape</a>.
+--   
+--   If one component of <a>shape</a> is the special value -1, the size of
+--   that dimension is computed so that the total size remains constant. In
+--   particular, a <a>shape</a> of `[-1]` flattens into 1-D. At most one
+--   component of <a>shape</a> can be -1.
+--   
+--   If <a>shape</a> is 1-D or higher, then the operation returns a tensor
+--   with shape <a>shape</a> filled with the values of <tt>tensor</tt>. In
+--   this case, the number of elements implied by <a>shape</a> must be the
+--   same as the number of elements in <tt>tensor</tt>.
+--   
+--   For example:
+--   
+--   ```prettyprint # tensor <tt>t</tt> is [1, 2, 3, 4, 5, 6, 7, 8, 9] #
+--   tensor <tt>t</tt> has shape [9] reshape(t, [3, 3]) ==&gt; [[1, 2, 3],
+--   [4, 5, 6], [7, 8, 9]]
+--   
+--   # tensor <tt>t</tt> is [[[1, 1], [2, 2]], # [[3, 3], [4, 4]]] # tensor
+--   <tt>t</tt> has shape [2, 2, 2] reshape(t, [2, 4]) ==&gt; [[1, 1, 2,
+--   2], [3, 3, 4, 4]]
+--   
+--   # tensor <tt>t</tt> is [[[1, 1, 1], # [2, 2, 2]], # [[3, 3, 3], # [4,
+--   4, 4]], # [[5, 5, 5], # [6, 6, 6]]] # tensor <tt>t</tt> has shape [3,
+--   2, 3] # pass '[-1]' to flatten <tt>t</tt> reshape(t, [-1]) ==&gt; [1,
+--   1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6, 6, 6]
+--   
+--   # -1 can also be used to infer the shape
+--   
+--   # -1 is inferred to be 9: reshape(t, [2, -1]) ==&gt; [[1, 1, 1, 2, 2,
+--   2, 3, 3, 3], [4, 4, 4, 5, 5, 5, 6, 6, 6]] # -1 is inferred to be 2:
+--   reshape(t, [-1, 9]) ==&gt; [[1, 1, 1, 2, 2, 2, 3, 3, 3], [4, 4, 4, 5,
+--   5, 5, 6, 6, 6]] # -1 is inferred to be 3: reshape(t, [ 2, -1, 3])
+--   ==&gt; [[[1, 1, 1], [2, 2, 2], [3, 3, 3]], [[4, 4, 4], [5, 5, 5], [6,
+--   6, 6]]]
+--   
+--   # tensor <tt>t</tt> is [7] # shape `[]` reshapes to a scalar
+--   reshape(t, []) ==&gt; 7 ```
+reshape :: (TensorType t, TensorType tshape, OneOf '[Int32, Int64] tshape) => Tensor v1 t -> Tensor v2 tshape -> Tensor Value t
+
+-- | A queue that produces elements in first-in first-out order.
+fIFOQueue :: Tensor Value ByteString
+
+-- | Generates labels for candidate sampling with a learned unigram
+--   distribution.
+--   
+--   See explanations of candidate sampling and the data formats at
+--   go/candidate-sampling.
+--   
+--   For each batch, this op picks a single set of sampled candidate
+--   labels.
+--   
+--   The advantages of sampling candidates per-batch are simplicity and the
+--   possibility of efficient dense matrix multiplication. The disadvantage
+--   is that the sampled candidates must be chosen independently of the
+--   context and of the true labels.
+learnedUnigramCandidateSampler :: Int64 -> Int64 -> Int64 -> Bool -> Tensor v1 Int64 -> (Tensor Value Int64, Tensor Value Float, Tensor Value Float)
+
+-- | Performs fractional average pooling on the input.
+--   
+--   Fractional average pooling is similar to Fractional max pooling in the
+--   pooling region generation step. The only difference is that after
+--   pooling regions are generated, a mean operation is performed instead
+--   of a max operation in each pooling region.
+fractionalAvgPool :: (TensorType t, OneOf '[Int32, Int64, Double, Float] t) => Tensor v1 t -> (Tensor Value t, Tensor Value Int64, Tensor Value Int64)
+
+-- | Randomly crop <tt>image</tt>.
+--   
+--   <a>size</a> is a 1-D int64 tensor with 2 elements representing the
+--   crop height and width. The values must be non negative.
+--   
+--   This Op picks a random location in <tt>image</tt> and crops a
+--   <tt>height</tt> by <tt>width</tt> rectangle from that location. The
+--   random location is picked so the cropped area will fit inside the
+--   original image.
+randomCrop :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 Int64 -> Tensor Value t
+
+-- | Cast x of type SrcT to y of DstT.
+--   
+--   _HostCast requires its input and produces its output in host memory.
+_HostCast :: (TensorType dstT, TensorType srcT) => Tensor v1 srcT -> Tensor Value dstT
+
+-- | Closes the given queue.
+--   
+--   This operation signals that no more elements will be enqueued in the
+--   given queue. Subsequent Enqueue(Many) operations will fail. Subsequent
+--   Dequeue(Many) operations will continue to succeed if sufficient
+--   elements remain in the queue. Subsequent Dequeue(Many) operations that
+--   would block will fail immediately.
+queueClose :: Tensor v1 ByteString -> ControlNode
+
+-- | Return a slice from <tt>input</tt>.
+--   
+--   The output tensor is a tensor with dimensions described by <a>size</a>
+--   whose values are extracted from <tt>input</tt> starting at the offsets
+--   in <tt>begin</tt>.
+--   
+--   <ul>
+--   <li>Requirements*: 0 &lt;= begin[i] &lt;= begin[i] + size[i] &lt;= Di
+--   for i in [0, n)</li>
+--   </ul>
+slice :: (TensorType index, OneOf '[Int32, Int64] index, TensorType t) => Tensor v1 t -> Tensor v2 index -> Tensor v3 index -> Tensor Value t
+
+-- | Returns the gradient of <tt>StridedSlice</tt>.
+--   
+--   Since <tt>StridedSlice</tt> cuts out pieces of its <tt>input</tt>
+--   which is size <a>shape</a>, its gradient will have the same shape
+--   (which is passed here as <a>shape</a>). The gradient will be zero in
+--   any element that the slice does not select.
+--   
+--   Arguments are the same as StridedSliceGrad with the exception that
+--   <tt>dy</tt> is the input gradient to be propagated and <a>shape</a> is
+--   the shape of <tt>StridedSlice</tt>'s <tt>input</tt>.
+stridedSliceGrad :: (TensorType index, OneOf '[Int32, Int64] index, TensorType t) => Tensor v1 index -> Tensor v2 index -> Tensor v3 index -> Tensor v4 index -> Tensor v5 t -> Tensor Value t
+
+-- | Adds up a <tt>SparseTensor</tt> and a dense <a>Tensor</a>, producing a
+--   dense <a>Tensor</a>.
+--   
+--   This Op does not require <tt>a_indices</tt> be sorted in standard
+--   lexicographic order.
+sparseTensorDenseAdd :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 tindices -> Tensor v2 t -> Tensor v3 tindices -> Tensor v4 t -> Tensor Value t
+
+-- | Returns the size of a tensor.
+--   
+--   This operation returns an integer representing the number of elements
+--   in <tt>input</tt>.
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>t</tt> is [[[1, 1,, 1], [2, 2, 2]], [[3, 3, 3],
+--   [4, 4, 4]]]] size(t) ==&gt; 12 ```
+size :: (TensorType t, TensorType out_type, OneOf '[Int32, Int64] out_type) => Tensor v1 t -> Tensor Value out_type
+
+-- | Defines a barrier that persists across different graph executions.
+--   
+--   A barrier represents a key-value map, where each key is a string, and
+--   each value is a tuple of tensors.
+--   
+--   At runtime, the barrier contains <tt>complete</tt> and
+--   <tt>incomplete</tt> elements. A complete element has defined tensors
+--   for all components of its value tuple, and may be accessed using
+--   BarrierTakeMany. An incomplete element has some undefined components
+--   in its value tuple, and may be updated using BarrierInsertMany.
+barrier :: Tensor Value ByteString
+
+-- | Computes the log of the absolute value of `Gamma(x)` element-wise.
+lgamma :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Decode a JPEG-encoded image to a uint8 tensor.
+--   
+--   The attr <tt>channels</tt> indicates the desired number of color
+--   channels for the decoded image.
+--   
+--   Accepted values are:
+--   
+--   <ul>
+--   <li>0: Use the number of channels in the JPEG-encoded image.</li>
+--   <li>1: output a grayscale image.</li>
+--   <li>3: output an RGB image.</li>
+--   </ul>
+--   
+--   If needed, the JPEG-encoded image is transformed to match the
+--   requested number of color channels.
+--   
+--   The attr <tt>ratio</tt> allows downscaling the image by an integer
+--   factor during decoding. Allowed values are: 1, 2, 4, and 8. This is
+--   much faster than downscaling the image later.
+decodeJpeg :: Tensor v1 ByteString -> Tensor Value Word8
+
+-- | Returns shape of tensors.
+--   
+--   This operation returns N 1-D integer tensors representing shape of
+--   `input[i]s`.
+shapeN :: (TensorType t, TensorType out_type, OneOf '[Int32, Int64] out_type) => [Tensor v1 t] -> [Tensor Value out_type]
+
+-- | Generates labels for candidate sampling with a uniform distribution.
+--   
+--   See explanations of candidate sampling and the data formats at
+--   go/candidate-sampling.
+--   
+--   For each batch, this op picks a single set of sampled candidate
+--   labels.
+--   
+--   The advantages of sampling candidates per-batch are simplicity and the
+--   possibility of efficient dense matrix multiplication. The disadvantage
+--   is that the sampled candidates must be chosen independently of the
+--   context and of the true labels.
+uniformCandidateSampler :: Int64 -> Int64 -> Int64 -> Bool -> Tensor v1 Int64 -> (Tensor Value Int64, Tensor Value Float, Tensor Value Float)
+
+-- | Finds unique elements in a 1-D tensor.
+--   
+--   This operation returns a tensor <tt>y</tt> containing all of the
+--   unique elements of <tt>x</tt> sorted in the same order that they occur
+--   in <tt>x</tt>. This operation also returns a tensor <tt>idx</tt> the
+--   same size as <tt>x</tt> that contains the index of each value of
+--   <tt>x</tt> in the unique output <tt>y</tt>. In other words:
+--   
+--   `y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`
+--   
+--   For example:
+--   
+--   ```prettyprint # tensor <tt>x</tt> is [1, 1, 2, 4, 4, 4, 7, 8, 8] y,
+--   idx = unique(x) y ==&gt; [1, 2, 4, 7, 8] idx ==&gt; [0, 0, 1, 2, 2, 2,
+--   3, 4, 4] ```
+unique :: (TensorType t, TensorType out_idx, OneOf '[Int32, Int64] out_idx) => Tensor v1 t -> (Tensor Value t, Tensor Value out_idx)
+
+-- | Draw bounding boxes on a batch of images.
+--   
+--   Outputs a copy of <tt>images</tt> but draws on top of the pixels zero
+--   or more bounding boxes specified by the locations in <tt>boxes</tt>.
+--   The coordinates of the each bounding box in <tt>boxes</tt> are encoded
+--   as `[y_min, x_min, y_max, x_max]`. The bounding box coordinates are
+--   floats in `[0.0, 1.0]` relative to the width and height of the
+--   underlying image.
+--   
+--   For example, if an image is 100 x 200 pixels and the bounding box is
+--   `[0.1, 0.2, 0.5, 0.9]`, the bottom-left and upper-right coordinates of
+--   the bounding box will be `(10, 40)` to `(50, 180)`.
+--   
+--   Parts of the bounding box may fall outside the image.
+drawBoundingBoxes :: (TensorType t, OneOf '[Word16, Float] t) => Tensor v1 t -> Tensor v2 Float -> Tensor Value t
+
+-- | Split the data from the input value into TensorArray elements.
+--   
+--   Assuming that <tt>lengths</tt> takes on values
+--   
+--   ```(n0, n1, ..., n(T-1))```
+--   
+--   and that <a>value</a> has shape
+--   
+--   ```(n0 + n1 + ... + n(T-1) x d0 x d1 x ...)```,
+--   
+--   this splits values into a TensorArray with T tensors.
+--   
+--   TensorArray index t will be the subtensor of values with starting
+--   position
+--   
+--   ```(n0 + n1 + ... + n(t-1), 0, 0, ...)```
+--   
+--   and having size
+--   
+--   ```nt x d0 x d1 x ...```
+tensorArraySplit :: (TensorType t) => Tensor v1 ByteString -> Tensor v2 t -> Tensor v3 Int64 -> Tensor v4 Float -> Tensor Value Float
+
+-- | Splits a tensor into <tt>num_split</tt> tensors along one dimension.
+split :: (TensorType t) => Int64 -> Tensor v1 Int32 -> Tensor v2 t -> [Tensor Value t]
+
+-- | Computes the maximum along segments of a tensor.
+--   
+--   Read <a>the section on Segmentation</a> for an explanation of
+--   segments.
+--   
+--   Computes a tensor such that \(output_i = max_j(data_j)\) where
+--   <a>max</a> is over <tt>j</tt> such that `segment_ids[j] == i`.
+--   
+--   <a>style="width:70%; margin:auto; margin-bottom:10px;
+--   margin-top:20px;"</a> <a>style="width:100%"
+--   src="../../images/SegmentMax.png" alt</a> <a>/div</a>
+segmentMax :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 tindices -> Tensor Value t
+
+-- | Raise a exception to abort the process when called.
+--   
+--   Returns nothing but an exception.
+abort :: ControlNode
+
+-- | Reorders a SparseTensor into the canonical, row-major ordering.
+--   
+--   Note that by convention, all sparse ops preserve the canonical
+--   ordering along increasing dimension number. The only time ordering can
+--   be violated is during manual manipulation of the indices and values
+--   vectors to add entries.
+--   
+--   Reordering does not affect the shape of the SparseTensor.
+--   
+--   If the tensor has rank <tt>R</tt> and <tt>N</tt> non-empty values,
+--   <tt>input_indices</tt> has shape `[N, R]`, input_values has length
+--   <tt>N</tt>, and input_shape has length <tt>R</tt>.
+sparseReorder :: (TensorType t) => Tensor v1 Int64 -> Tensor v2 t -> Tensor v3 Int64 -> (Tensor Value Int64, Tensor Value t)
+
+-- | Computes the gradient for the rsqrt of <tt>x</tt> wrt its input.
+--   
+--   Specifically, `grad = dy * -0.5 * y^3`, where `y = rsqrt(x)`, and
+--   <tt>dy</tt> is the corresponding input gradient.
+rsqrtGrad :: (TensorType t, OneOf '[Complex Double, Complex Float, Word16, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Reverses variable length slices.
+--   
+--   This op first slices <tt>input</tt> along the dimension
+--   <tt>batch_dim</tt>, and for each slice <tt>i</tt>, reverses the first
+--   `seq_lengths[i]` elements along the dimension <tt>seq_dim</tt>.
+--   
+--   The elements of <tt>seq_lengths</tt> must obey `seq_lengths[i] &lt;
+--   input.dims[seq_dim]`, and <tt>seq_lengths</tt> must be a vector of
+--   length `input.dims[batch_dim]`.
+--   
+--   The output slice <tt>i</tt> along dimension <tt>batch_dim</tt> is then
+--   given by input slice <tt>i</tt>, with the first `seq_lengths[i]`
+--   slices along dimension <tt>seq_dim</tt> reversed.
+--   
+--   For example:
+--   
+--   ```prettyprint # Given this: batch_dim = 0 seq_dim = 1 input.dims =
+--   (4, 8, ...) seq_lengths = [7, 2, 3, 5]
+--   
+--   # then slices of input are reversed on seq_dim, but only up to
+--   seq_lengths: output[0, 0:7, :, ...] = input[0, 7:0:-1, :, ...]
+--   output[1, 0:2, :, ...] = input[1, 2:0:-1, :, ...] output[2, 0:3, :,
+--   ...] = input[2, 3:0:-1, :, ...] output[3, 0:5, :, ...] = input[3,
+--   5:0:-1, :, ...]
+--   
+--   # while entries past seq_lens are copied through: output[0, 7:, :,
+--   ...] = input[0, 7:, :, ...] output[1, 2:, :, ...] = input[1, 2:, :,
+--   ...] output[2, 3:, :, ...] = input[2, 3:, :, ...] output[3, 2:, :,
+--   ...] = input[3, 2:, :, ...] ```
+--   
+--   In contrast, if:
+--   
+--   ```prettyprint # Given this: batch_dim = 2 seq_dim = 0 input.dims =
+--   (8, ?, 4, ...) seq_lengths = [7, 2, 3, 5]
+--   
+--   # then slices of input are reversed on seq_dim, but only up to
+--   seq_lengths: output[0:7, :, 0, :, ...] = input[7:0:-1, :, 0, :, ...]
+--   output[0:2, :, 1, :, ...] = input[2:0:-1, :, 1, :, ...] output[0:3, :,
+--   2, :, ...] = input[3:0:-1, :, 2, :, ...] output[0:5, :, 3, :, ...] =
+--   input[5:0:-1, :, 3, :, ...]
+--   
+--   # while entries past seq_lens are copied through: output[7:, :, 0, :,
+--   ...] = input[7:, :, 0, :, ...] output[2:, :, 1, :, ...] = input[2:, :,
+--   1, :, ...] output[3:, :, 2, :, ...] = input[3:, :, 2, :, ...]
+--   output[2:, :, 3, :, ...] = input[2:, :, 3, :, ...] ```
+reverseSequence :: (TensorType t, TensorType tlen, OneOf '[Int32, Int64] tlen) => Int64 -> Tensor v1 t -> Tensor v2 tlen -> Tensor Value t
+
+-- | Returns the number of records this Reader has produced.
+--   
+--   This is the same as the number of ReaderRead executions that have
+--   succeeded.
+readerNumRecordsProduced :: Tensor v1 ByteString -> Tensor Value Int64
+
+-- | Deserialize and concatenate <tt>SparseTensors</tt> from a serialized
+--   minibatch.
+--   
+--   The input <tt>serialized_sparse</tt> must be a string matrix of shape
+--   `[N x 3]` where <tt>N</tt> is the minibatch size and the rows
+--   correspond to packed outputs of <tt>SerializeSparse</tt>. The ranks of
+--   the original <tt>SparseTensor</tt> objects must all match. When the
+--   final <tt>SparseTensor</tt> is created, it has rank one higher than
+--   the ranks of the incoming <tt>SparseTensor</tt> objects (they have
+--   been concatenated along a new row dimension).
+--   
+--   The output <tt>SparseTensor</tt> object's shape values for all
+--   dimensions but the first are the max across the input
+--   <tt>SparseTensor</tt> objects' shape values for the corresponding
+--   dimensions. Its first shape value is <tt>N</tt>, the minibatch size.
+--   
+--   The input <tt>SparseTensor</tt> objects' indices are assumed ordered
+--   in standard lexicographic order. If this is not the case, after this
+--   step run <tt>SparseReorder</tt> to restore index ordering.
+--   
+--   For example, if the serialized input is a `[2 x 3]` matrix
+--   representing two original <tt>SparseTensor</tt> objects:
+--   
+--   index = [ 0] [10] [20] values = [1, 2, 3] shape = [50]
+--   
+--   and
+--   
+--   index = [ 2] [10] values = [4, 5] shape = [30]
+--   
+--   then the final deserialized <tt>SparseTensor</tt> will be:
+--   
+--   index = [0 0] [0 10] [0 20] [1 2] [1 10] values = [1, 2, 3, 4, 5]
+--   shape = [2 50]
+deserializeManySparse :: (TensorType dtype) => Tensor v1 ByteString -> (Tensor Value Int64, Tensor Value dtype, Tensor Value Int64)
+
+-- | Returns immutable tensor from memory region.
+--   
+--   The current implementation memmaps the tensor from a file.
+immutableConst :: (TensorType dtype) => Tensor Value dtype
+
+-- | Returns the min of x and y (i.e. x &lt; y ? x : y) element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>Minimum</tt> supports broadcasting. More about
+--   broadcasting <a>here</a></li>
+--   </ul>
+minimum :: (TensorType t, OneOf '[Int32, Int64, Word16, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Initializes a table from a text file.
+--   
+--   It inserts one key-value pair into the table for each line of the
+--   file. The key and value is extracted from the whole line content,
+--   elements from the split line based on <tt>delimiter</tt> or the line
+--   number (starting from zero). Where to extract the key and value from a
+--   line is specified by <tt>key_index</tt> and <tt>value_index</tt>.
+--   
+--   <ul>
+--   <li>A value of -1 means use the line number(starting from zero),
+--   expects <tt>int64</tt>.</li>
+--   <li>A value of -2 means use the whole line content, expects
+--   <tt>string</tt>.</li>
+--   <li>A value &gt;= 0 means use the index (starting at zero) of the
+--   split line based on <tt>delimiter</tt>.</li>
+--   </ul>
+initializeTableFromTextFile :: Int64 -> Int64 -> Tensor v1 ByteString -> Tensor v2 ByteString -> ControlNode
+
+-- | Returns the diagonal part of the tensor.
+--   
+--   This operation returns a tensor with the <tt>diagonal</tt> part of the
+--   <tt>input</tt>. The <tt>diagonal</tt> part is computed as follows:
+--   
+--   Assume <tt>input</tt> has dimensions `[D1,..., Dk, D1,..., Dk]`, then
+--   the output is a tensor of rank <tt>k</tt> with dimensions `[D1,...,
+--   Dk]` where:
+--   
+--   `diagonal[i1,..., ik] = input[i1, ..., ik, i1,..., ik]`.
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>input</tt> is [[1, 0, 0, 0] [0, 2, 0, 0] [0, 0,
+--   3, 0] [0, 0, 0, 4]]
+--   
+--   tf.diag_part(input) ==&gt; [1, 2, 3, 4] ```
+diagPart :: (TensorType t, OneOf '[Complex Double, Complex Float, Int32, Int64, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes natural logarithm of x element-wise.
+--   
+--   I.e., \(y = log_e x\).
+log :: (TensorType t, OneOf '[Complex Double, Complex Float, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Scatter the data from the input value into specific TensorArray
+--   elements.
+--   
+--   <tt>indices</tt> must be a vector, its length must match the first dim
+--   of <a>value</a>.
+tensorArrayScatter :: (TensorType t) => Tensor v1 ByteString -> Tensor v2 Int32 -> Tensor v3 t -> Tensor v4 Float -> Tensor Value Float
+
+-- | Returns the rank of a tensor.
+--   
+--   This operation returns an integer representing the rank of
+--   <tt>input</tt>.
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>t</tt> is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3],
+--   [4, 4, 4]]] # shape of tensor <tt>t</tt> is [2, 2, 3] rank(t) ==&gt; 3
+--   ```
+--   
+--   <ul>
+--   <li>*Note**: The rank of a tensor is not the same as the rank of a
+--   matrix. The rank of a tensor is the number of indices required to
+--   uniquely select each element of the tensor. Rank is also known as
+--   "order", "degree", or "ndims."</li>
+--   </ul>
+rank :: (TensorType t) => Tensor v1 t -> Tensor Value Int32
+
+-- | Return a tensor with the same shape and contents as the input tensor
+--   or value.
+identity :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Adjust the contrast of one or more images.
+--   
+--   <tt>images</tt> is a tensor of at least 3 dimensions. The last 3
+--   dimensions are interpreted as `[height, width, channels]`. The other
+--   dimensions only represent a collection of images, such as `[batch,
+--   height, width, channels].`
+--   
+--   Contrast is adjusted independently for each channel of each image.
+--   
+--   For each channel, the Op first computes the mean of the image pixels
+--   in the channel and then adjusts each component of each pixel to `(x -
+--   mean) * contrast_factor + mean`.
+adjustContrastv2 :: Tensor v1 Float -> Tensor v2 Float -> Tensor Value Float
+
+-- | Sparse update entries in '*var' and '*accum' according to FOBOS
+--   algorithm.
+--   
+--   That is for rows we have grad for, we update var and accum as follows:
+--   accum += grad * grad prox_v = var prox_v -= lr * grad * (1 /
+--   sqrt(accum)) var = sign(prox_v)/(1+lr*l2) * max{|prox_v|-lr*l1,0}
+sparseApplyProximalAdagrad :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 t -> Tensor v6 t -> Tensor v7 tindices -> Tensor Value t
+
+-- | Gather slices from <tt>params</tt> according to <tt>indices</tt>.
+--   
+--   <tt>indices</tt> must be an integer tensor of any dimension (usually
+--   0-D or 1-D). Produces an output tensor with shape `indices.shape +
+--   params.shape[1:]` where:
+--   
+--   # Scalar indices output[:, ..., :] = params[indices, :, ... :]
+--   
+--   # Vector indices output[i, :, ..., :] = params[indices[i], :, ... :]
+--   
+--   # Higher rank indices output[i, ..., j, :, ... :] = params[indices[i,
+--   ..., j], :, ..., :]
+--   
+--   If <tt>indices</tt> is a permutation and `len(indices) ==
+--   params.shape[0]` then this operation will permute <tt>params</tt>
+--   accordingly.
+--   
+--   <a>style="width:70%; margin:auto; margin-bottom:10px;
+--   margin-top:20px;"</a> <a>style="width:100%"
+--   src="../../images/Gather.png" alt</a> <a>/div</a>
+gather :: (TensorType tindices, OneOf '[Int32, Int64] tindices, TensorType tparams) => Tensor v1 tparams -> Tensor v2 tindices -> Tensor Value tparams
+
+-- | Checks whether a tensor has been initialized.
+--   
+--   Outputs boolean scalar indicating whether the tensor has been
+--   initialized.
+isVariableInitialized :: (TensorType dtype) => Tensor v1 dtype -> Tensor Value Bool
+
+-- | Concatenates tensors along one dimension.
+concat :: (TensorType t) => Tensor v1 Int32 -> [Tensor v2 t] -> Tensor Value t
+
+-- | Outputs random integers from a uniform distribution.
+--   
+--   The generated values are uniform integers in the range `[minval,
+--   maxval)`. The lower bound <tt>minval</tt> is included in the range,
+--   while the upper bound <tt>maxval</tt> is excluded.
+--   
+--   The random integers are slightly biased unless `maxval - minval` is an
+--   exact power of two. The bias is small for values of `maxval - minval`
+--   significantly smaller than the range of the output (either `2^32` or
+--   `2^64`).
+randomUniformInt :: (TensorType t, OneOf '[Int32, Int64] t, TensorType tout, OneOf '[Int32, Int64] tout) => Tensor v1 t -> Tensor v2 tout -> Tensor v3 tout -> Tensor Value tout
+
+-- | Stops gradient computation.
+--   
+--   When executed in a graph, this op outputs its input tensor as-is.
+--   
+--   When building ops to compute gradients, this op prevents the
+--   contribution of its inputs to be taken into account. Normally, the
+--   gradient generator adds ops to a graph to compute the derivatives of a
+--   specified <tt>loss</tt> by recursively finding out inputs that
+--   contributed to its computation. If you insert this op in the graph it
+--   inputs are masked from the gradient generator. They are not taken into
+--   account for computing gradients.
+--   
+--   This is useful any time you want to compute a value with TensorFlow
+--   but need to pretend that the value was a constant. Some examples
+--   include:
+--   
+--   <ul>
+--   <li>The *EM* algorithm where the *M-step* should not involve
+--   backpropagation through the output of the *E-step*.</li>
+--   <li>Contrastive divergence training of Boltzmann machines where, when
+--   differentiating the energy function, the training must not
+--   backpropagate through the graph that generated the samples from the
+--   model.</li>
+--   <li>Adversarial training, where no backprop should happen through the
+--   adversarial example generation process.</li>
+--   </ul>
+stopGradient :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Performs average pooling on the input.
+--   
+--   Each entry in <tt>output</tt> is the mean of the corresponding size
+--   <tt>ksize</tt> window in <a>value</a>.
+avgPool :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | A Reader that outputs the entire contents of a file as a value.
+--   
+--   To use, enqueue filenames in a Queue. The output of ReaderRead will be
+--   a filename (key) and the contents of that file (value).
+wholeFileReader :: Tensor Value ByteString
+
+-- | Forwards `data` to the output port determined by <a>pred</a>.
+--   
+--   If <a>pred</a> is true, the `data` input is forwarded to
+--   <tt>output_true</tt>. Otherwise, the data goes to
+--   <tt>output_false</tt>.
+--   
+--   See also <tt>RefSwitch</tt> and <tt>Merge</tt>.
+switch :: (TensorType t) => Tensor v1 t -> Tensor v2 Bool -> (Tensor Value t, Tensor Value t)
+
+-- | Outputs random values from a normal distribution.
+--   
+--   The generated values will have mean 0 and standard deviation 1.
+randomStandardNormal :: (TensorType t, OneOf '[Int32, Int64] t, TensorType dtype, OneOf '[Word16, Double, Float] dtype) => Tensor v1 t -> Tensor Value dtype
+
+-- | Computes sigmoid of <tt>x</tt> element-wise.
+--   
+--   Specifically, `y = 1 / (1 + exp(-x))`.
+sigmoid :: (TensorType t, OneOf '[Complex Double, Complex Float, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Generate a single randomly distorted bounding box for an image.
+--   
+--   Bounding box annotations are often supplied in addition to
+--   ground-truth labels in image recognition or object localization tasks.
+--   A common technique for training such a system is to randomly distort
+--   an image while preserving its content, i.e. *data augmentation*. This
+--   Op outputs a randomly distorted localization of an object, i.e.
+--   bounding box, given an <tt>image_size</tt>, <tt>bounding_boxes</tt>
+--   and a series of constraints.
+--   
+--   The output of this Op is a single bounding box that may be used to
+--   crop the original image. The output is returned as 3 tensors:
+--   <tt>begin</tt>, <a>size</a> and <tt>bboxes</tt>. The first 2 tensors
+--   can be fed directly into `tf.slice` to crop the image. The latter may
+--   be supplied to `tf.image.draw_bounding_box` to visualize what the
+--   bounding box looks like.
+--   
+--   Bounding boxes are supplied and returned as `[y_min, x_min, y_max,
+--   x_max]`. The bounding box coordinates are floats in `[0.0, 1.0]`
+--   relative to the width and height of the underlying image.
+--   
+--   For example,
+--   
+--   # Generate a single distorted bounding box. begin, size, bbox_for_draw
+--   = tf.image.sample_distorted_bounding_box( tf.shape(image),
+--   bounding_boxes=bounding_boxes)
+--   
+--   # Draw the bounding box in an image summary. image_with_box =
+--   tf.image.draw_bounding_boxes(tf.expand_dims(image, 0), bbox_for_draw)
+--   tf.image_summary(<tt>images_with_box</tt>, image_with_box)
+--   
+--   # Employ the bounding box to distort the image. distorted_image =
+--   tf.slice(image, begin, size)
+--   
+--   Note that if no bounding box information is available, setting
+--   `use_image_if_no_bounding_boxes = true` will assume there is a single
+--   implicit bounding box covering the whole image. If
+--   <tt>use_image_if_no_bounding_boxes</tt> is false and no bounding boxes
+--   are supplied, an error is raised.
+sampleDistortedBoundingBox :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word8] t) => Tensor v1 t -> Tensor v2 Float -> (Tensor Value t, Tensor Value t, Tensor Value Float)
+
+-- | Returns the truth value of (x &gt; y) element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>Greater</tt> supports broadcasting. More about
+--   broadcasting <a>here</a></li>
+--   </ul>
+greater :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value Bool
+
+-- | Makes its input available to the next iteration.
+refNextIteration :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | SpaceToDepth for tensors of type T.
+--   
+--   Rearranges blocks of spatial data, into depth. More specifically, this
+--   op outputs a copy of the input tensor where values from the
+--   <tt>height</tt> and <tt>width</tt> dimensions are moved to the
+--   <tt>depth</tt> dimension. The attr <tt>block_size</tt> indicates the
+--   input block size and how the data is moved.
+--   
+--   <ul>
+--   <li>Non-overlapping blocks of size `block_size x block size` are
+--   rearranged into depth at each location.</li>
+--   <li>The depth of the output tensor is `input_depth * block_size *
+--   block_size`.</li>
+--   <li>The input tensor's height and width must be divisible by
+--   block_size.</li>
+--   </ul>
+--   
+--   That is, assuming the input is in the shape: `[batch, height, width,
+--   depth]`, the shape of the output will be: `[batch,
+--   height<i>block_size, width</i>block_size,
+--   depth*block_size*block_size]`
+--   
+--   This operation requires that the input tensor be of rank 4, and that
+--   <tt>block_size</tt> be &gt;=1 and a divisor of both the input
+--   <tt>height</tt> and <tt>width</tt>.
+--   
+--   This operation is useful for resizing the activations between
+--   convolutions (but keeping all data), e.g. instead of pooling. It is
+--   also useful for training purely convolutional models.
+--   
+--   For example, given this input of shape `[1, 2, 2, 1]`, and block_size
+--   of 2:
+--   
+--   ```prettyprint x = [[[[1], [2]], [[3], [4]]]] ```
+--   
+--   This operation will output a tensor of shape `[1, 1, 1, 4]`:
+--   
+--   ```prettyprint [[[[1, 2, 3, 4]]]] ```
+--   
+--   Here, the input has a batch of 1 and each batch element has shape `[2,
+--   2, 1]`, the corresponding output will have a single element (i.e.
+--   width and height are both 1) and will have a depth of 4 channels (1 *
+--   block_size * block_size). The output element shape is `[1, 1, 4]`.
+--   
+--   For an input tensor with larger depth, here of shape `[1, 2, 2, 3]`,
+--   e.g.
+--   
+--   ```prettyprint x = [[[[1, 2, 3], [4, 5, 6]], [[7, 8, 9], [10, 11,
+--   12]]]] ```
+--   
+--   This operation, for block_size of 2, will return the following tensor
+--   of shape `[1, 1, 1, 12]`
+--   
+--   ```prettyprint [[[[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]]]] ```
+--   
+--   Similarly, for the following input of shape `[1 4 4 1]`, and a block
+--   size of 2:
+--   
+--   ```prettyprint x = [[[[1], [2], [5], [6]], [[3], [4], [7], [8]], [[9],
+--   [10], [13], [14]], [[11], [12], [15], [16]]]] ```
+--   
+--   the operator will return the following tensor of shape `[1 2 2 4]`:
+--   
+--   ```prettyprint x = [[[[1, 2, 3, 4], [5, 6, 7, 8]], [[9, 10, 11, 12],
+--   [13, 14, 15, 16]]]] ```
+spaceToDepth :: (TensorType t) => Int64 -> Tensor v1 t -> Tensor Value t
+
+-- | Does nothing. Serves as a control trigger for scheduling.
+--   
+--   Only useful as a placeholder for control edges.
+controlTrigger :: ControlNode
+
+-- | Divides a variable reference by sparse updates.
+--   
+--   This operation computes
+--   
+--   # Scalar indices ref[indices, ...] /= updates[...]
+--   
+--   # Vector indices (for each i) ref[indices[i], ...] /= updates[i, ...]
+--   
+--   # High rank indices (for each i, ..., j) ref[indices[i, ..., j], ...]
+--   /= updates[i, ..., j, ...]
+--   
+--   This operation outputs <tt>ref</tt> after the update is done. This
+--   makes it easier to chain operations that need to use the reset value.
+--   
+--   Duplicate entries are handled correctly: if multiple <tt>indices</tt>
+--   reference the same location, their contributions divide.
+--   
+--   Requires `updates.shape = indices.shape + ref.shape[1:]`.
+scatterDiv :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 tindices -> Tensor v3 t -> Tensor Value t
+
+-- | Copy Op.
+--   
+--   Performs CPU-to-CPU or GPU-to-GPU deep-copying of tensor, depending on
+--   the device on which the tensor is allocated.
+--   
+--   Unlike the CopyHost Op, this op does not have HostMemory constraint on
+--   its input or output.
+copy :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes the gradient of the crop_and_resize op wrt the input boxes
+--   tensor.
+cropAndResizeGradBoxes :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 Float -> Tensor v2 t -> Tensor v3 Float -> Tensor v4 Int32 -> Tensor Value Float
+
+-- | Computes the mean along sparse segments of a tensor.
+--   
+--   Read <a>the section on Segmentation</a> for an explanation of
+--   segments.
+--   
+--   Like <tt>SegmentMean</tt>, but <tt>segment_ids</tt> can have rank less
+--   than `data`'s first dimension, selecting a subset of dimension 0,
+--   specified by <tt>indices</tt>.
+sparseSegmentMean :: (TensorType t, OneOf '[Double, Float] t, TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 t -> Tensor v2 tidx -> Tensor v3 Int32 -> Tensor Value t
+
+-- | Update <tt>ref</tt> by assigning <a>value</a> to it.
+--   
+--   This operation outputs "ref" after the assignment is done. This makes
+--   it easier to chain operations that need to use the reset value.
+assign :: (TensorType t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Restores a tensor from checkpoint files.
+--   
+--   Reads a tensor stored in one or several files. If there are several
+--   files (for instance because a tensor was saved as slices),
+--   <tt>file_pattern</tt> may contain wildcard symbols (<a>*</a> and
+--   <tt>?</tt>) in the filename portion only, not in the directory
+--   portion.
+--   
+--   If a <tt>file_pattern</tt> matches several files,
+--   <tt>preferred_shard</tt> can be used to hint in which file the
+--   requested tensor is likely to be found. This op will first open the
+--   file at index <tt>preferred_shard</tt> in the list of matching files
+--   and try to restore tensors from that file. Only if some tensors or
+--   tensor slices are not found in that first file, then the Op opens all
+--   the files. Setting <tt>preferred_shard</tt> to match the value passed
+--   as the <tt>shard</tt> input of a matching <tt>Save</tt> Op may speed
+--   up Restore. This attribute only affects performance, not correctness.
+--   The default value -1 means files are processed in order.
+--   
+--   See also <tt>RestoreSlice</tt>.
+restore :: (TensorType dt) => Tensor v1 ByteString -> Tensor v2 ByteString -> Tensor Value dt
+
+-- | Computes gradients of the maxpooling function.
+maxPoolGradWithArgmax :: (TensorType t, OneOf '[Word16, Float] t, TensorType targmax, OneOf '[Int32, Int64] targmax) => Tensor v1 t -> Tensor v2 t -> Tensor v3 targmax -> Tensor Value t
+
+-- | Checks a tensor for NaN and Inf values.
+--   
+--   When run, reports an <tt>InvalidArgument</tt> error if <tt>tensor</tt>
+--   has any values that are not a number (NaN) or infinity (Inf).
+--   Otherwise, passes <tt>tensor</tt> as-is.
+checkNumerics :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Returns a tensor of zeros with the same shape and type as x.
+zerosLike :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Reads and outputs the entire contents of the input filename.
+readFile :: Tensor v1 ByteString -> Tensor Value ByteString
+
+-- | Shuffle dimensions of x according to a permutation.
+--   
+--   The output <tt>y</tt> has the same rank as <tt>x</tt>. The shapes of
+--   <tt>x</tt> and <tt>y</tt> satisfy: `y.shape[i] == x.shape[perm[i]] for
+--   i in [0, 1, ..., rank(x) - 1]`
+transpose :: (TensorType t, TensorType tperm, OneOf '[Int32, Int64] tperm) => Tensor v1 t -> Tensor v2 tperm -> Tensor Value t
+
+-- | Transforms a serialized tensorflow.TensorProto proto into a Tensor.
+parseTensor :: (TensorType out_type) => Tensor v1 ByteString -> Tensor Value out_type
+
+-- | Computes acos of x element-wise.
+acos :: (TensorType t, OneOf '[Complex Double, Complex Float, Int32, Int64, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Bitcasts a tensor from one type to another without copying data.
+--   
+--   Given a tensor <tt>input</tt>, this operation returns a tensor that
+--   has the same buffer data as <tt>input</tt> with datatype `type`.
+--   
+--   If the input datatype <tt>T</tt> is larger than the output datatype
+--   `type` then the shape changes from [...] to [...,
+--   sizeof(<tt>T</tt>)/sizeof(`type`)].
+--   
+--   If <tt>T</tt> is smaller than `type`, the operator requires that the
+--   rightmost dimension be equal to sizeof(`type`)/sizeof(<tt>T</tt>). The
+--   shape then goes from [..., sizeof(`type`)/sizeof(<tt>T</tt>)] to
+--   [...].
+--   
+--   <ul>
+--   <li>NOTE*: Bitcast is implemented as a low-level cast, so machines
+--   with different endian orderings will give different results.</li>
+--   </ul>
+bitcast :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType type', OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] type') => Tensor v1 t -> Tensor Value type'
+
+-- | Replaces the contents of the table with the specified keys and values.
+--   
+--   The tensor <tt>keys</tt> must be of the same type as the keys of the
+--   table. The tensor <tt>values</tt> must be of the type of the table
+--   values.
+lookupTableImport :: (TensorType tin, TensorType tout) => Tensor v1 ByteString -> Tensor v2 tin -> Tensor v3 tout -> ControlNode
+
+-- | The backward operation for <a>BiasAdd</a> on the "bias" tensor.
+--   
+--   It accumulates all the values from out_backprop into the feature
+--   dimension. For NHWC data format, the feature dimension is the last.
+--   For NCHW data format, the feature dimension is the third-to-last.
+biasAddGrad :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor Value t
+batchSelfAdjointEig :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes the product of elements across dimensions of a tensor.
+--   
+--   Reduces <tt>input</tt> along the dimensions given in
+--   <tt>reduction_indices</tt>. Unless <tt>keep_dims</tt> is true, the
+--   rank of the tensor is reduced by 1 for each entry in
+--   <tt>reduction_indices</tt>. If <tt>keep_dims</tt> is true, the reduced
+--   dimensions are retained with length 1.
+prod :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 t -> Tensor v2 tidx -> Tensor Value t
+
+-- | Resize <tt>images</tt> to <a>size</a> using bilinear interpolation.
+--   
+--   Input images can be of different types but output images are always
+--   float.
+resizeBilinear :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 Int32 -> Tensor Value Float
+
+-- | Unpack the data from the input value into TensorArray elements.
+--   
+--   <ul>
+--   <li>*WARNING: This op is deprecated.**</li>
+--   </ul>
+--   
+--   Instead of this op, use <tt>TensorArrayScatter</tt> with `indices =
+--   RangeOp(0, SizeOp(value)[0])`.
+tensorArrayUnpack :: (TensorType t) => Tensor v1 ByteString -> Tensor v2 t -> Tensor v3 Float -> Tensor Value Float
+batchMatrixDeterminant :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes the sum of elements across dimensions of a tensor.
+--   
+--   Reduces <tt>input</tt> along the dimensions given in
+--   <tt>reduction_indices</tt>. Unless <tt>keep_dims</tt> is true, the
+--   rank of the tensor is reduced by 1 for each entry in
+--   <tt>reduction_indices</tt>. If <tt>keep_dims</tt> is true, the reduced
+--   dimensions are retained with length 1.
+sum :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 t -> Tensor v2 tidx -> Tensor Value t
+
+-- | Compute the inverse 2-dimensional discrete Fourier Transform over the
+--   inner-most
+--   
+--   2 dimensions of <tt>input</tt>.
+iFFT2D :: Tensor v1 (Complex Float) -> Tensor Value (Complex Float)
+
+-- | Creates a tensor filled with a scalar value.
+--   
+--   This operation creates a tensor of shape <tt>dims</tt> and fills it
+--   with <a>value</a>.
+--   
+--   For example:
+--   
+--   ```prettyprint # Output tensor has shape [2, 3]. fill([2, 3], 9)
+--   ==&gt; [[9, 9, 9] [9, 9, 9]] ```
+fill :: (TensorType t) => Tensor v1 Int32 -> Tensor v2 t -> Tensor Value t
+
+-- | Generates labels for candidate sampling with a learned unigram
+--   distribution.
+--   
+--   A unigram sampler could use a fixed unigram distribution read from a
+--   file or passed in as an in-memory array instead of building up the
+--   distribution from data on the fly. There is also an option to skew the
+--   distribution by applying a distortion power to the weights.
+--   
+--   The vocabulary file should be in CSV-like format, with the last field
+--   being the weight associated with the word.
+--   
+--   For each batch, this op picks a single set of sampled candidate
+--   labels.
+--   
+--   The advantages of sampling candidates per-batch are simplicity and the
+--   possibility of efficient dense matrix multiplication. The disadvantage
+--   is that the sampled candidates must be chosen independently of the
+--   context and of the true labels.
+fixedUnigramCandidateSampler :: Int64 -> Int64 -> Int64 -> Bool -> Tensor v1 Int64 -> (Tensor Value Int64, Tensor Value Float, Tensor Value Float)
+
+-- | Computes the grayscale dilation of 4-D <tt>input</tt> and 3-D
+--   <a>filter</a> tensors.
+--   
+--   The <tt>input</tt> tensor has shape `[batch, in_height, in_width,
+--   depth]` and the <a>filter</a> tensor has shape `[filter_height,
+--   filter_width, depth]`, i.e., each input channel is processed
+--   independently of the others with its own structuring function. The
+--   <tt>output</tt> tensor has shape `[batch, out_height, out_width,
+--   depth]`. The spatial dimensions of the output tensor depend on the
+--   <tt>padding</tt> algorithm. We currently only support the default
+--   <a>NHWC</a> <tt>data_format</tt>.
+--   
+--   In detail, the grayscale morphological 2-D dilation is the max-sum
+--   correlation (for consistency with <tt>conv2d</tt>, we use unmirrored
+--   filters):
+--   
+--   output[b, y, x, c] = max_{dy, dx} input[b, strides[1] * y + rates[1] *
+--   dy, strides[2] * x + rates[2] * dx, c] + filter[dy, dx, c]
+--   
+--   Max-pooling is a special case when the filter has size equal to the
+--   pooling kernel size and contains all zeros.
+--   
+--   Note on duality: The dilation of <tt>input</tt> by the <a>filter</a>
+--   is equal to the negation of the erosion of `-input` by the reflected
+--   <a>filter</a>.
+dilation2D :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Compute the polygamma function \(psi^{(n)}(x)\).
+--   
+--   The polygamma function is defined as:
+--   
+--   ``` psi^{(n)}(x) = frac{d^n}{dx^n} psi(x) ``` where \(psi(x)\) is the
+--   digamma function.
+polygamma :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Return the same ref tensor as the input ref tensor.
+refIdentity :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | PNG-encode an image.
+--   
+--   <tt>image</tt> is a 3-D uint8 or uint16 Tensor of shape `[height,
+--   width, channels]` where <tt>channels</tt> is:
+--   
+--   <ul>
+--   <li>1: for grayscale.</li>
+--   <li>2: for grayscale + alpha.</li>
+--   <li>3: for RGB.</li>
+--   <li>4: for RGBA.</li>
+--   </ul>
+--   
+--   The ZLIB compression level, <tt>compression</tt>, can be -1 for the
+--   PNG-encoder default or a value from 0 to 9. 9 is the highest
+--   compression level, generating the smallest output, but is slower.
+encodePng :: (TensorType t, OneOf '[Word16, Word8] t) => Tensor v1 t -> Tensor Value ByteString
+
+-- | Updates the table to associates keys with values.
+--   
+--   The tensor <tt>keys</tt> must be of the same type as the keys of the
+--   table. The tensor <tt>values</tt> must be of the type of the table
+--   values.
+lookupTableInsert :: (TensorType tin, TensorType tout) => Tensor v1 ByteString -> Tensor v2 tin -> Tensor v3 tout -> ControlNode
+batchIFFT2D :: Tensor v1 (Complex Float) -> Tensor Value (Complex Float)
+
+-- | Finds unique elements in a 1-D tensor.
+--   
+--   This operation returns a tensor <tt>y</tt> containing all of the
+--   unique elements of <tt>x</tt> sorted in the same order that they occur
+--   in <tt>x</tt>. This operation also returns a tensor <tt>idx</tt> the
+--   same size as <tt>x</tt> that contains the index of each value of
+--   <tt>x</tt> in the unique output <tt>y</tt>. Finally, it returns a
+--   third tensor <tt>count</tt> that contains the count of each element of
+--   <tt>y</tt> in <tt>x</tt>. In other words:
+--   
+--   `y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`
+--   
+--   For example:
+--   
+--   ```prettyprint # tensor <tt>x</tt> is [1, 1, 2, 4, 4, 4, 7, 8, 8] y,
+--   idx, count = unique_with_counts(x) y ==&gt; [1, 2, 4, 7, 8] idx ==&gt;
+--   [0, 0, 1, 2, 2, 2, 3, 4, 4] count ==&gt; [2, 1, 3, 1, 2] ```
+uniqueWithCounts :: (TensorType t, TensorType out_idx, OneOf '[Int32, Int64] out_idx) => Tensor v1 t -> (Tensor Value t, Tensor Value out_idx, Tensor Value out_idx)
+
+-- | Gather values or slices from <tt>params</tt> according to
+--   <tt>indices</tt>.
+--   
+--   <tt>params</tt> is a Tensor of rank <tt>R</tt> and <tt>indices</tt> is
+--   a Tensor of rank <tt>M</tt>.
+--   
+--   <tt>indices</tt> must be integer tensor, containing indices into
+--   <tt>params</tt>. It must be shape `[d_0, ..., d_N, R]` where `0 &lt; R
+--   &lt;= M`.
+--   
+--   The innermost dimension of <tt>indices</tt> (with length <tt>R</tt>)
+--   corresponds to indices into elements (if `R = M`) or slices (if `R
+--   &lt; M`) along the <tt>N</tt>th dimension of <tt>params</tt>.
+--   
+--   Produces an output tensor with shape
+--   
+--   <ul>
+--   <li><i>d_0, ..., d_{n-1}, params.shape[R</i> , ...,
+--   params.shape[M-1]].</li>
+--   </ul>
+--   
+--   Some examples below.
+--   
+--   Simple indexing into a matrix:
+--   
+--   indices = [[0, 0], [1, 1]] params = [[<tt>a</tt>, <tt>b</tt>],
+--   [<tt>c</tt>, <tt>d</tt>]] output = [<tt>a</tt>, <tt>d</tt>]
+--   
+--   Slice indexing into a matrix:
+--   
+--   indices = [[1], [0]] params = [[<tt>a</tt>, <tt>b</tt>], [<tt>c</tt>,
+--   <tt>d</tt>]] output = [[<tt>c</tt>, <tt>d</tt>], [<tt>a</tt>,
+--   <tt>b</tt>]]
+--   
+--   Indexing into a 3-tensor:
+--   
+--   indices = [[1]] params = [[[<tt>a0</tt>, <tt>b0</tt>], [<tt>c0</tt>,
+--   <tt>d0</tt>]], [[<tt>a1</tt>, <tt>b1</tt>], [<tt>c1</tt>,
+--   <tt>d1</tt>]]] output = [[[<tt>a1</tt>, <tt>b1</tt>], [<tt>c1</tt>,
+--   <tt>d1</tt>]]]
+--   
+--   indices = [[0, 1], [1, 0]] params = [[[<tt>a0</tt>, <tt>b0</tt>],
+--   [<tt>c0</tt>, <tt>d0</tt>]], [[<tt>a1</tt>, <tt>b1</tt>],
+--   [<tt>c1</tt>, <tt>d1</tt>]]] output = [[<tt>c0</tt>, <tt>d0</tt>],
+--   [<tt>a1</tt>, <tt>b1</tt>]]
+--   
+--   indices = [[0, 0, 1], [1, 0, 1]] params = [[[<tt>a0</tt>,
+--   <tt>b0</tt>], [<tt>c0</tt>, <tt>d0</tt>]], [[<tt>a1</tt>,
+--   <tt>b1</tt>], [<tt>c1</tt>, <tt>d1</tt>]]] output = [<tt>b0</tt>,
+--   <tt>b1</tt>]
+--   
+--   Batched indexing into a matrix:
+--   
+--   indices = [[[0, 0]], [[0, 1]]] params = [[<tt>a</tt>, <tt>b</tt>],
+--   [<tt>c</tt>, <tt>d</tt>]] output = [[<tt>a</tt>], [<tt>b</tt>]]
+--   
+--   Batched slice indexing into a matrix:
+--   
+--   indices = [[[1]], [[0]]] params = [[<tt>a</tt>, <tt>b</tt>],
+--   [<tt>c</tt>, <tt>d</tt>]] output = [[[<tt>c</tt>, <tt>d</tt>]],
+--   [[<tt>a</tt>, <tt>b</tt>]]]
+--   
+--   Batched indexing into a 3-tensor:
+--   
+--   indices = [[[1]], [[0]]] params = [[[<tt>a0</tt>, <tt>b0</tt>],
+--   [<tt>c0</tt>, <tt>d0</tt>]], [[<tt>a1</tt>, <tt>b1</tt>],
+--   [<tt>c1</tt>, <tt>d1</tt>]]] output = [[[[<tt>a1</tt>, <tt>b1</tt>],
+--   [<tt>c1</tt>, <tt>d1</tt>]]], [[[<tt>a0</tt>, <tt>b0</tt>],
+--   [<tt>c0</tt>, <tt>d0</tt>]]]]
+--   
+--   indices = [[[0, 1], [1, 0]], [[0, 0], [1, 1]]] params =
+--   [[[<tt>a0</tt>, <tt>b0</tt>], [<tt>c0</tt>, <tt>d0</tt>]],
+--   [[<tt>a1</tt>, <tt>b1</tt>], [<tt>c1</tt>, <tt>d1</tt>]]] output =
+--   [[[<tt>c0</tt>, <tt>d0</tt>], [<tt>a1</tt>, <tt>b1</tt>]],
+--   [[<tt>a0</tt>, <tt>b0</tt>], [<tt>c1</tt>, <tt>d1</tt>]]]
+--   
+--   indices = [[[0, 0, 1], [1, 0, 1]], [[0, 1, 1], [1, 1, 0]]] params =
+--   [[[<tt>a0</tt>, <tt>b0</tt>], [<tt>c0</tt>, <tt>d0</tt>]],
+--   [[<tt>a1</tt>, <tt>b1</tt>], [<tt>c1</tt>, <tt>d1</tt>]]] output =
+--   [[<tt>b0</tt>, <tt>b1</tt>], [<tt>d0</tt>, <tt>c1</tt>]]
+gatherNd :: (TensorType tindices, OneOf '[Int32, Int64] tindices, TensorType tparams) => Tensor v1 tparams -> Tensor v2 tindices -> Tensor Value tparams
+
+-- | Read an element from the TensorArray into output <a>value</a>.
+tensorArrayRead :: (TensorType dtype) => Tensor v1 ByteString -> Tensor v2 Int32 -> Tensor v3 Float -> Tensor Value dtype
+
+-- | Returns up to <tt>num_records</tt> (key, value) pairs produced by a
+--   Reader.
+--   
+--   Will dequeue from the input queue if necessary (e.g. when the Reader
+--   needs to start reading from a new file since it has finished with the
+--   previous file). It may return less than <tt>num_records</tt> even
+--   before the last batch.
+readerReadUpTo :: Tensor v1 ByteString -> Tensor v2 ByteString -> Tensor v3 Int64 -> (Tensor Value ByteString, Tensor Value ByteString)
+
+-- | Compute the regularized incomplete beta integral \(I_x(a, b)\).
+--   
+--   The regularized incomplete beta integral is defined as:
+--   
+--   ``` I_x(a, b) = frac{B(x; a, b)}{B(a, b)} ``` where
+--   
+--   ``` B(x; a, b) = int_0^x t^{a-1} (1 - t)^{b-1} dt ```
+--   
+--   is the incomplete beta function and \(B(a, b)\) is the *complete* beta
+--   function.
+betainc :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor Value t
+batchMatrixBandPart :: (TensorType t) => Tensor v1 t -> Tensor v2 Int64 -> Tensor v3 Int64 -> Tensor Value t
+
+-- | Computes the gradients of depthwise convolution with respect to the
+--   input.
+depthwiseConv2dNativeBackpropInput :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 Int32 -> Tensor v2 t -> Tensor v3 t -> Tensor Value t
+
+-- | Forwards the <tt>index</tt>th element of <tt>inputs</tt> to
+--   <tt>output</tt>.
+refSelect :: (TensorType t) => Tensor v1 Int32 -> [Tensor v2 t] -> Tensor Value t
+
+-- | Exits the current frame to its parent frame.
+--   
+--   Exit makes its input `data` available to the parent frame.
+exit :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Looks up keys in a table, outputs the corresponding values.
+--   
+--   The tensor <tt>keys</tt> must of the same type as the keys of the
+--   table. The output <tt>values</tt> is of the type of the table values.
+--   
+--   The scalar <tt>default_value</tt> is the value output for keys not
+--   present in the table. It must also be of the same type as the table
+--   values.
+lookupTableFind :: (TensorType tin, TensorType tout) => Tensor v1 ByteString -> Tensor v2 tin -> Tensor v3 tout -> Tensor Value tout
+
+-- | Removes dimensions of size 1 from the shape of a tensor.
+--   
+--   Given a tensor <tt>input</tt>, this operation returns a tensor of the
+--   same type with all dimensions of size 1 removed. If you don't want to
+--   remove all size 1 dimensions, you can remove specific size 1
+--   dimensions by specifying <tt>squeeze_dims</tt>.
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>t</tt> is a tensor of shape [1, 2, 1, 3, 1, 1]
+--   shape(squeeze(t)) ==&gt; [2, 3] ```
+--   
+--   Or, to remove specific size 1 dimensions:
+--   
+--   ```prettyprint # <tt>t</tt> is a tensor of shape [1, 2, 1, 3, 1, 1]
+--   shape(squeeze(t, [2, 4])) ==&gt; [1, 2, 3, 1] ```
+squeeze :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes the mean of elements across dimensions of a tensor.
+--   
+--   Reduces <tt>input</tt> along the dimensions given in
+--   <tt>reduction_indices</tt>. Unless <tt>keep_dims</tt> is true, the
+--   rank of the tensor is reduced by 1 for each entry in
+--   <tt>reduction_indices</tt>. If <tt>keep_dims</tt> is true, the reduced
+--   dimensions are retained with length 1.
+mean :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 t -> Tensor v2 tidx -> Tensor Value t
+
+-- | SpaceToBatch for N-D tensors of type T.
+--   
+--   This operation divides "spatial" dimensions `[1, ..., M]` of the input
+--   into a grid of blocks of shape <tt>block_shape</tt>, and interleaves
+--   these blocks with the "batch" dimension (0) such that in the output,
+--   the spatial dimensions `[1, ..., M]` correspond to the position within
+--   the grid, and the batch dimension combines both the position within a
+--   spatial block and the original batch position. Prior to division into
+--   blocks, the spatial dimensions of the input are optionally zero padded
+--   according to <tt>paddings</tt>. See below for a precise description.
+spaceToBatchND :: (TensorType t, TensorType tblock_shape, OneOf '[Int32, Int64] tblock_shape, TensorType tpaddings, OneOf '[Int32, Int64] tpaddings) => Tensor v1 t -> Tensor v2 tblock_shape -> Tensor v3 tpaddings -> Tensor Value t
+
+-- | SpaceToBatch for 4-D tensors of type T.
+--   
+--   This is a legacy version of the more general SpaceToBatchND.
+--   
+--   Zero-pads and then rearranges (permutes) blocks of spatial data into
+--   batch. More specifically, this op outputs a copy of the input tensor
+--   where values from the <tt>height</tt> and <tt>width</tt> dimensions
+--   are moved to the <tt>batch</tt> dimension. After the zero-padding,
+--   both <tt>height</tt> and <tt>width</tt> of the input must be divisible
+--   by the block size.
+spaceToBatch :: (TensorType t, TensorType tpaddings, OneOf '[Int32, Int64] tpaddings) => Int64 -> Tensor v1 t -> Tensor v2 tpaddings -> Tensor Value t
+
+-- | Performs greedy decoding on the logits given in inputs.
+--   
+--   A note about the attribute merge_repeated: if enabled, when
+--   consecutive logits' maximum indices are the same, only the first of
+--   these is emitted. Labeling the blank <a>*</a>, the sequence "A B B * B
+--   B" becomes "A B" if merge_repeated = True and "A B B B B" if
+--   merge_repeated = False.
+--   
+--   Regardless of the value of merge_repeated, if the maximum index of a
+--   given time and batch corresponds to the blank, index `(num_classes -
+--   1)`, no new element is emitted.
+cTCGreedyDecoder :: Tensor v1 Float -> Tensor v2 Int32 -> (Tensor Value Int64, Tensor Value Int64, Tensor Value Int64, Tensor Value Float)
+
+-- | BatchToSpace for N-D tensors of type T.
+--   
+--   This operation reshapes the "batch" dimension 0 into `M + 1`
+--   dimensions of shape `block_shape + [batch]`, interleaves these blocks
+--   back into the grid defined by the spatial dimensions `[1, ..., M]`, to
+--   obtain a result with the same rank as the input. The spatial
+--   dimensions of this intermediate result are then optionally cropped
+--   according to <tt>crops</tt> to produce the output. This is the reverse
+--   of SpaceToBatch. See below for a precise description.
+batchToSpaceND :: (TensorType t, TensorType tblock_shape, OneOf '[Int32, Int64] tblock_shape, TensorType tcrops, OneOf '[Int32, Int64] tcrops) => Tensor v1 t -> Tensor v2 tblock_shape -> Tensor v3 tcrops -> Tensor Value t
+
+-- | Packs a list of <tt>N</tt> rank-<tt>R</tt> tensors into one
+--   rank-`(R+1)` tensor.
+--   
+--   Packs the <tt>N</tt> tensors in <tt>values</tt> into a tensor with
+--   rank one higher than each tensor in <tt>values</tt>, by packing them
+--   along the <tt>axis</tt> dimension. Given a list of tensors of shape
+--   `(A, B, C)`;
+--   
+--   if `axis == 0` then the <tt>output</tt> tensor will have the shape
+--   `(N, A, B, C)`. if `axis == 1` then the <tt>output</tt> tensor will
+--   have the shape `(A, N, B, C)`. Etc.
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>x</tt> is [1, 4] # <tt>y</tt> is [2, 5] #
+--   <tt>z</tt> is [3, 6] pack([x, y, z]) =&gt; [[1, 4], [2, 5], [3, 6]] #
+--   Pack along first dim. pack([x, y, z], axis=1) =&gt; [[1, 2, 3], [4, 5,
+--   6]] ```
+--   
+--   This is the opposite of <a>unpack</a>.
+pack :: (TensorType t) => [Tensor v1 t] -> Tensor Value t
+
+-- | Returns a one-hot tensor.
+--   
+--   The locations represented by indices in <tt>indices</tt> take value
+--   <tt>on_value</tt>, while all other locations take value
+--   <tt>off_value</tt>.
+--   
+--   If the input <tt>indices</tt> is rank <tt>N</tt>, the output will have
+--   rank `N+1`, The new axis is created at dimension <tt>axis</tt>
+--   (default: the new axis is appended at the end).
+--   
+--   If <tt>indices</tt> is a scalar the output shape will be a vector of
+--   length <tt>depth</tt>.
+--   
+--   If <tt>indices</tt> is a vector of length <tt>features</tt>, the
+--   output shape will be: ``` features x depth if axis == -1 depth x
+--   features if axis == 0 ```
+--   
+--   If <tt>indices</tt> is a matrix (batch) with shape `[batch,
+--   features]`, the output shape will be: ``` batch x features x depth if
+--   axis == -1 batch x depth x features if axis == 1 depth x batch x
+--   features if axis == 0 ```
+--   
+--   Examples =========
+--   
+--   Suppose that
+--   
+--   ``` indices = [0, 2, -1, 1] depth = 3 on_value = 5.0 off_value = 0.0
+--   axis = -1 ```
+--   
+--   Then output is `[4 x 3]`:
+--   
+--   ```output = [5.0 0.0 0.0] // one_hot(0) [0.0 0.0 5.0] // one_hot(2)
+--   [0.0 0.0 0.0] // one_hot(-1) [0.0 5.0 0.0] // one_hot(1) ```
+--   
+--   Suppose that
+--   
+--   ``` indices = [0, 2, -1, 1] depth = 3 on_value = 0.0 off_value = 3.0
+--   axis = 0 ```
+--   
+--   Then output is `[3 x 4]`:
+--   
+--   ```output = [0.0 3.0 3.0 3.0] [3.0 3.0 3.0 0.0] [3.0 3.0 3.0 3.0] [3.0
+--   0.0 3.0 3.0] // ^ one_hot(0) // ^ one_hot(2) // ^ one_hot(-1) // ^
+--   one_hot(1) ``` Suppose that
+--   
+--   ``` indices = [[0, 2], [1, -1]] depth = 3 on_value = 1.0 off_value =
+--   0.0 axis = -1 ```
+--   
+--   Then output is `[2 x 2 x 3]`:
+--   
+--   ```output = [ [1.0, 0.0, 0.0] // one_hot(0) [0.0, 0.0, 1.0] //
+--   one_hot(2) ][ [0.0, 1.0, 0.0] // one_hot(1) [0.0, 0.0, 0.0] //
+--   one_hot(-1) ]```
+oneHot :: (TensorType t, TensorType tI, OneOf '[Int32, Int64, Word8] tI) => Tensor v1 tI -> Tensor v2 Int32 -> Tensor v3 t -> Tensor v4 t -> Tensor Value t
+
+-- | Return the reduction indices for computing gradients of s0 op s1 with
+--   broadcast.
+--   
+--   This is typically used by gradient computations for a broadcasting
+--   operation.
+broadcastGradientArgs :: (TensorType t, OneOf '[Int32, Int64] t) => Tensor v1 t -> Tensor v2 t -> (Tensor Value t, Tensor Value t)
+
+-- | Returns a batched matrix tensor with new batched diagonal values.
+--   
+--   Given <tt>input</tt> and <tt>diagonal</tt>, this operation returns a
+--   tensor with the same shape and values as <tt>input</tt>, except for
+--   the diagonals of the innermost matrices. These will be overwritten by
+--   the values in <tt>diagonal</tt>. The batched matrices must be square.
+--   
+--   The output is computed as follows:
+--   
+--   Assume <tt>input</tt> has `k+1` dimensions `[I, J, K, ..., N, N]` and
+--   <tt>diagonal</tt> has <tt>k</tt> dimensions `[I, J, K, ..., N]`. Then
+--   the output is a tensor of rank `k+1` with dimensions [I, J, K, ..., N,
+--   N]` where:
+--   
+--   <ul>
+--   <li>`output[i, j, k, ..., m, n] = diagonal[i, j, k, ..., n]` for `m ==
+--   n`.</li>
+--   <li>`output[i, j, k, ..., m, n] = input[i, j, k, ..., m, n]` for `m !=
+--   n`.</li>
+--   </ul>
+matrixSetDiag :: (TensorType t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Update '*var' according to the RMSProp algorithm.
+--   
+--   Note that in dense implement of this algorithm, ms and mom will update
+--   even if the grad is zero, but in this sparse implement, ms and mom
+--   will not update in iterations the grad is zero.
+--   
+--   mean_square = decay * mean_square + (1-decay) * gradient ** 2 Delta =
+--   learning_rate * gradient / sqrt(mean_square + epsilon)
+--   
+--   ms &lt;- rho * ms_{t-1} + (1-rho) * grad * grad mom &lt;- momentum *
+--   mom_{t-1} + lr * grad / sqrt(ms + epsilon) var &lt;- var - mom
+applyRMSProp :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 t -> Tensor v6 t -> Tensor v7 t -> Tensor v8 t -> Tensor Value t
+
+-- | Returns a constant tensor.
+const :: (TensorType dtype) => Tensor Value dtype
+
+-- | Creates or finds a child frame, and makes `data` available to the
+--   child frame.
+--   
+--   This op is used together with <tt>Exit</tt> to create loops in the
+--   graph. The unique <tt>frame_name</tt> is used by the <tt>Executor</tt>
+--   to identify frames. If <tt>is_constant</tt> is true, <tt>output</tt>
+--   is a constant in the child frame; otherwise it may be changed in the
+--   child frame. At most <tt>parallel_iterations</tt> iterations are run
+--   in parallel in the child frame.
+enter :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Debug Identity Op.
+--   
+--   Provides an identity mapping of the non-Ref type input tensor for
+--   debugging.
+debugIdentity :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Debug NaN Value Counter Op
+--   
+--   Counts number of NaNs in the input tensor, for debugging.
+debugNanCount :: (TensorType t) => Tensor v1 t -> Tensor Value Int64
+
+-- | Batch normalization.
+--   
+--   This op is deprecated. Prefer `tf.nn.batch_normalization`.
+batchNormWithGlobalNormalization :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Bool -> Float -> Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 t -> Tensor Value t
+batchMatrixDiag :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Unpacks a given dimension of a rank-<tt>R</tt> tensor into
+--   <tt>num</tt> rank-`(R-1)` tensors.
+--   
+--   Unpacks <tt>num</tt> tensors from <a>value</a> by chipping it along
+--   the <tt>axis</tt> dimension. For example, given a tensor of shape `(A,
+--   B, C, D)`;
+--   
+--   If `axis == 0` then the i'th tensor in <tt>output</tt> is the slice
+--   `value[i, :, :, :]` and each tensor in <tt>output</tt> will have shape
+--   `(B, C, D)`. (Note that the dimension unpacked along is gone, unlike
+--   <a>split</a>).
+--   
+--   If `axis == 1` then the i'th tensor in <tt>output</tt> is the slice
+--   `value[:, i, :, :]` and each tensor in <tt>output</tt> will have shape
+--   `(A, C, D)`. Etc.
+--   
+--   This is the opposite of <a>pack</a>.
+unpack :: (TensorType t) => Int64 -> Tensor v1 t -> [Tensor Value t]
+
+-- | Split a <tt>SparseTensor</tt> into <tt>num_split</tt> tensors along
+--   one dimension.
+--   
+--   If the `shape[split_dim]` is not an integer multiple of
+--   <tt>num_split</tt>. Slices `[0 : shape[split_dim] % num_split]` gets
+--   one extra dimension. For example, if `split_dim = 1` and `num_split =
+--   2` and the input is
+--   
+--   input_tensor = shape = [2, 7] [ a d e ] [b c ]
+--   
+--   Graphically the output tensors are:
+--   
+--   output_tensor[0] = shape = [2, 4] [ a ] [b c ]
+--   
+--   output_tensor[1] = shape = [2, 3] [ d e ] [ ]
+sparseSplit :: (TensorType t) => Int64 -> Tensor v1 Int64 -> Tensor v2 Int64 -> Tensor v3 t -> Tensor v4 Int64 -> ([Tensor Value Int64], [Tensor Value t], [Tensor Value Int64])
+
+-- | Pads a tensor with mirrored values.
+--   
+--   This operation pads a <tt>input</tt> with mirrored values according to
+--   the <tt>paddings</tt> you specify. <tt>paddings</tt> is an integer
+--   tensor with shape `[n, 2]`, where n is the rank of <tt>input</tt>. For
+--   each dimension D of <tt>input</tt>, `paddings[D, 0]` indicates how
+--   many values to add before the contents of <tt>input</tt> in that
+--   dimension, and `paddings[D, 1]` indicates how many values to add after
+--   the contents of <tt>input</tt> in that dimension. Both `paddings[D,
+--   0]` and `paddings[D, 1]` must be no greater than `input.dim_size(D)`
+--   (or `input.dim_size(D) - 1`) if <tt>copy_border</tt> is true (if
+--   false, respectively).
+--   
+--   The padded size of each dimension D of the output is:
+--   
+--   `paddings(D, 0) + input.dim_size(D) + paddings(D, 1)`
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>t</tt> is [[1, 2, 3], [4, 5, 6]]. #
+--   <tt>paddings</tt> is [[1, 1]], [2, 2]]. # <tt>mode</tt> is SYMMETRIC.
+--   # rank of <tt>t</tt> is 2. pad(t, paddings) ==&gt; [[2, 1, 1, 2, 3, 3,
+--   2] [2, 1, 1, 2, 3, 3, 2] [5, 4, 4, 5, 6, 6, 5] [5, 4, 4, 5, 6, 6, 5]]
+--   ```
+mirrorPad :: (TensorType t, TensorType tpaddings, OneOf '[Int32, Int64] tpaddings) => Tensor v1 t -> Tensor v2 tpaddings -> Tensor Value t
+batchMatrixDiagPart :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes gradient of the FractionalMaxPool function.
+fractionalMaxPoolGrad :: (TensorType t, OneOf '[Int32, Int64, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 Int64 -> Tensor v5 Int64 -> Tensor Value t
+
+-- | Returns the set of files matching a pattern.
+--   
+--   Note that this routine only supports wildcard characters in the
+--   basename portion of the pattern, not in the directory portion.
+matchingFiles :: Tensor v1 ByteString -> Tensor Value ByteString
+
+-- | Constructs a tensor by tiling a given tensor.
+--   
+--   This operation creates a new tensor by replicating <tt>input</tt>
+--   <tt>multiples</tt> times. The output tensor's i'th dimension has
+--   `input.dims(i) * multiples[i]` elements, and the values of
+--   <tt>input</tt> are replicated `multiples[i]` times along the
+--   <tt>i</tt>th dimension. For example, tiling `[a b c d]` by `[2]`
+--   produces `[a b c d a b c d]`.
+tile :: (TensorType t, TensorType tmultiples, OneOf '[Int32, Int64] tmultiples) => Tensor v1 t -> Tensor v2 tmultiples -> Tensor Value t
+
+-- | Returns the element-wise min of two SparseTensors.
+--   
+--   Assumes the two SparseTensors have the same shape, i.e., no
+--   broadcasting.
+sparseSparseMinimum :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 Int64 -> Tensor v2 t -> Tensor v3 Int64 -> Tensor v4 Int64 -> Tensor v5 t -> Tensor v6 Int64 -> (Tensor Value Int64, Tensor Value t)
+
+-- | Generates labels for candidate sampling with a learned unigram
+--   distribution.
+--   
+--   See explanations of candidate sampling and the data formats at
+--   go/candidate-sampling.
+--   
+--   For each batch, this op picks a single set of sampled candidate
+--   labels.
+--   
+--   The advantages of sampling candidates per-batch are simplicity and the
+--   possibility of efficient dense matrix multiplication. The disadvantage
+--   is that the sampled candidates must be chosen independently of the
+--   context and of the true labels.
+allCandidateSampler :: Int64 -> Int64 -> Bool -> Tensor v1 Int64 -> (Tensor Value Int64, Tensor Value Float, Tensor Value Float)
+
+-- | Forwards the ref tensor `data` to the output port determined by
+--   <a>pred</a>.
+--   
+--   If <a>pred</a> is true, the `data` input is forwarded to
+--   <tt>output_true</tt>. Otherwise, the data goes to
+--   <tt>output_false</tt>.
+--   
+--   See also <tt>Switch</tt> and <tt>Merge</tt>.
+refSwitch :: (TensorType t) => Tensor v1 t -> Tensor v2 Bool -> (Tensor Value t, Tensor Value t)
+
+-- | Merges summaries.
+--   
+--   This op creates a <a>`Summary`</a> protocol buffer that contains the
+--   union of all the values in the input summaries.
+--   
+--   When the Op is run, it reports an <tt>InvalidArgument</tt> error if
+--   multiple values in the summaries to merge use the same tag.
+mergeSummary :: [Tensor v1 ByteString] -> Tensor Value ByteString
+
+-- | Returns the truth value of NOT x element-wise.
+logicalNot :: Tensor v1 Bool -> Tensor Value Bool
+
+-- | Gradients for Local Response Normalization.
+lRNGrad :: (TensorType t, OneOf '[Word16, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor Value t
+
+-- | Converts each string in the input Tensor to the specified numeric
+--   type.
+--   
+--   (Note that int32 overflow results in an error while float overflow
+--   results in a rounded value.)
+stringToNumber :: (TensorType out_type, OneOf '[Int32, Float] out_type) => Tensor v1 ByteString -> Tensor Value out_type
+
+-- | Multiply matrix "a" by matrix "b".
+--   
+--   The inputs must be two-dimensional matrices and the inner dimension of
+--   "a" must match the outer dimension of "b". This op is optimized for
+--   the case where at least one of "a" or "b" is sparse. The breakeven for
+--   using this versus a dense matrix multiply on one platform was 30% zero
+--   values in the sparse matrix.
+sparseMatMul :: (TensorType ta, OneOf '[Word16, Float] ta, TensorType tb, OneOf '[Word16, Float] tb) => Tensor v1 ta -> Tensor v2 tb -> Tensor Value Float
+
+-- | Forwards the value of an available tensor from <tt>inputs</tt> to
+--   <tt>output</tt>.
+--   
+--   <tt>Merge</tt> waits for at least one of the tensors in
+--   <tt>inputs</tt> to become available. It is usually combined with
+--   <tt>Switch</tt> to implement branching.
+--   
+--   <tt>Merge</tt> forwards the first tensor for become available to
+--   <tt>output</tt>, and sets <tt>value_index</tt> to its index in
+--   <tt>inputs</tt>.
+merge :: (TensorType t) => [Tensor v1 t] -> (Tensor Value t, Tensor Value Int32)
+
+-- | Computes the reverse mode backpropagated gradient of the Cholesky
+--   algorithm.
+--   
+--   For an explanation see "Differentiation of the Cholesky algorithm" by
+--   Iain Murray <a>http://arxiv.org/abs/1602.07527</a>.
+choleskyGrad :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+batchCholeskyGrad :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Gather specific elements from the TensorArray into output
+--   <a>value</a>.
+--   
+--   All elements selected by <tt>indices</tt> must have the same shape.
+tensorArrayGather :: (TensorType dtype) => Tensor v1 ByteString -> Tensor v2 Int32 -> Tensor v3 Float -> Tensor Value dtype
+
+-- | Resize <tt>images</tt> to <a>size</a> using nearest neighbor
+--   interpolation.
+resizeNearestNeighbor :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 Int32 -> Tensor Value t
+
+-- | Training via negative sampling.
+negTrain :: Int64 -> Tensor v1 Float -> Tensor v2 Float -> Tensor v3 Int32 -> Tensor v4 Int32 -> Tensor v5 Float -> ControlNode
+
+-- | Creates a TensorArray for storing the gradients of values in the given
+--   handle.
+--   
+--   If the given TensorArray gradient already exists, returns a reference
+--   to it.
+--   
+--   Locks the size of the original TensorArray by disabling its dynamic
+--   size flag.
+--   
+--   <ul>
+--   <li>*A note about the input flow_in:**</li>
+--   </ul>
+--   
+--   The handle flow_in forces the execution of the gradient lookup to
+--   occur only after certain other operations have occurred. For example,
+--   when the forward TensorArray is dynamically sized, writes to this
+--   TensorArray may resize the object. The gradient TensorArray is
+--   statically sized based on the size of the forward TensorArray when
+--   this operation executes. Furthermore, the size of the forward
+--   TensorArray is frozen by this call. As a result, the flow is used to
+--   ensure that the call to generate the gradient TensorArray only happens
+--   after all writes are executed.
+--   
+--   In the case of dynamically sized TensorArrays, gradient computation
+--   should only be performed on read operations that have themselves been
+--   chained via flow to occur only after all writes have executed. That
+--   way the final size of the forward TensorArray is known when this
+--   operation is called.
+--   
+--   <ul>
+--   <li>*A note about the source attribute:**</li>
+--   </ul>
+--   
+--   TensorArray gradient calls use an accumulator TensorArray object. If
+--   multiple gradients are calculated and run in the same session, the
+--   multiple gradient nodes may accidentally flow throuth the same
+--   accumulator TensorArray. This double counts and generally breaks the
+--   TensorArray gradient flow.
+--   
+--   The solution is to identify which gradient call this particular
+--   TensorArray gradient is being called in. This is performed by
+--   identifying a unique string (e.g. "gradients", "gradients_1", ...)
+--   from the input gradient Tensor's name. This string is used as a suffix
+--   when creating the TensorArray gradient object here (the attribute
+--   <tt>source</tt>).
+--   
+--   The attribute <tt>source</tt> is added as a suffix to the forward
+--   TensorArray's name when performing the creation / lookup, so that each
+--   separate gradient calculation gets its own TensorArray accumulator.
+tensorArrayGrad :: Tensor v1 ByteString -> Tensor v2 Float -> Tensor Value ByteString
+
+-- | Outputs a <tt>Summary</tt> protocol buffer with audio.
+--   
+--   The summary has up to <tt>max_outputs</tt> summary values containing
+--   audio. The audio is built from <tt>tensor</tt> which must be 3-D with
+--   shape `[batch_size, frames, channels]` or 2-D with shape `[batch_size,
+--   frames]`. The values are assumed to be in the range of `[-1.0, 1.0]`
+--   with a sample rate of <tt>sample_rate</tt>.
+--   
+--   The <tt>tag</tt> argument is a scalar <a>Tensor</a> of type
+--   <tt>string</tt>. It is used to build the <tt>tag</tt> of the summary
+--   values:
+--   
+--   <ul>
+--   <li>If <tt>max_outputs</tt> is 1, the summary value tag is
+--   '*tag*/audio'.</li>
+--   <li>If <tt>max_outputs</tt> is greater than 1, the summary value tags
+--   are generated sequentially as '*tag*/audio/0', '*tag*/audio/1',
+--   etc.</li>
+--   </ul>
+audioSummary :: Float -> Tensor v1 ByteString -> Tensor v2 Float -> Tensor Value ByteString
+
+-- | Does nothing. Only useful as a placeholder for control edges.
+noOp :: ControlNode
+
+-- | Makes its input available to the next iteration.
+nextIteration :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes softplus gradients for a softplus operation.
+softplusGrad :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Computes the singular value decompositions of one or more matrices.
+--   
+--   Computes the SVD of each inner matrix in <tt>input</tt> such that
+--   `input[..., :, :] = u[..., :, :] * diag(s[..., :, :]) *
+--   transpose(v[..., :, :])`
+--   
+--   ```prettyprint # a is a tensor containing a batch of matrices. # s is
+--   a tensor of singular values for each matrix. # u is the tensor
+--   containing of left singular vectors for each matrix. # v is the tensor
+--   containing of right singular vectors for each matrix. s, u, v = svd(a)
+--   s, _, _ = svd(a, compute_uv=False) ```
+svd :: (TensorType t, OneOf '[Complex Double, Complex Float, Double, Float] t) => Tensor v1 t -> (Tensor Value t, Tensor Value t, Tensor Value t)
+
+-- | Convert one or more images from HSV to RGB.
+--   
+--   Outputs a tensor of the same shape as the <tt>images</tt> tensor,
+--   containing the RGB value of the pixels. The output is only well
+--   defined if the value in <tt>images</tt> are in `[0,1]`.
+--   
+--   See <tt>rgb_to_hsv</tt> for a description of the HSV encoding.
+hSVToRGB :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Outputs random values from a normal distribution. The parameters may
+--   each be a
+--   
+--   scalar which applies to the entire output, or a vector of length
+--   shape[0] which stores the parameters for each batch.
+parameterizedTruncatedNormal :: (TensorType t, OneOf '[Int32, Int64] t, TensorType dtype, OneOf '[Word16, Double, Float] dtype) => Tensor v1 t -> Tensor v2 dtype -> Tensor v3 dtype -> Tensor v4 dtype -> Tensor v5 dtype -> Tensor Value dtype
+
+-- | Computes square of x element-wise.
+--   
+--   I.e., \(y = x * x = x^2\).
+square :: (TensorType t, OneOf '[Complex Double, Complex Float, Int32, Int64, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes exponential linear: `exp(features) - 1` if &lt; 0,
+--   <tt>features</tt> otherwise.
+--   
+--   See <a>Fast and Accurate Deep Network Learning by Exponential Linear
+--   Units (ELUs)</a>
+elu :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Outputs all keys and values in the table.
+lookupTableExport :: (TensorType tkeys, TensorType tvalues) => Tensor v1 ByteString -> (Tensor Value tkeys, Tensor Value tvalues)
+
+-- | Computes the number of elements in the given table.
+lookupTableSize :: Tensor v1 ByteString -> Tensor Value Int64
+
+-- | Computes gradients of the average pooling function.
+avgPoolGrad :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 Int32 -> Tensor v2 t -> Tensor Value t
+
+-- | Computes the ids of the positions in sampled_candidates that match
+--   true_labels.
+--   
+--   When doing log-odds NCE, the result of this op should be passed
+--   through a SparseToDense op, then added to the logits of the sampled
+--   candidates. This has the effect of <tt>removing</tt> the sampled
+--   labels that match the true labels by making the classifier sure that
+--   they are sampled labels.
+computeAccidentalHits :: Int64 -> Tensor v1 Int64 -> Tensor v2 Int64 -> (Tensor Value Int32, Tensor Value Int64, Tensor Value Float)
+
+-- | Calculates the CTC Loss (log probability) for each batch entry. Also
+--   calculates
+--   
+--   the gradient. This class performs the softmax operation for you, so
+--   inputs should be e.g. linear projections of outputs by an LSTM.
+cTCLoss :: Tensor v1 Float -> Tensor v2 Int64 -> Tensor v3 Int32 -> Tensor v4 Int32 -> (Tensor Value Float, Tensor Value Float)
+
+-- | Performs 3D average pooling on the input.
+avgPool3D :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes the reciprocal of x element-wise.
+--   
+--   I.e., \(y = 1 / x\).
+inv :: (TensorType t, OneOf '[Complex Double, Complex Float, Int32, Int64, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Pop the element at the top of the stack.
+stackPop :: (TensorType elem_type) => Tensor v1 ByteString -> Tensor Value elem_type
+
+-- | A queue that produces elements in first-in first-out order.
+--   
+--   Variable-size shapes are allowed by setting the corresponding shape
+--   dimensions to 0 in the shape attr. In this case DequeueMany will pad
+--   up to the maximum size of any given element in the minibatch. See
+--   below for details.
+paddingFIFOQueue :: Tensor Value ByteString
+batchSelfAdjointEigV2 :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> (Tensor Value t, Tensor Value t)
+batchMatrixTriangularSolve :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+batchMatrixSolveLs :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 Double -> Tensor Value t
+batchSvd :: (TensorType t, OneOf '[Complex Double, Complex Float, Double, Float] t) => Tensor v1 t -> (Tensor Value t, Tensor Value t, Tensor Value t)
+
+-- | Outputs a <tt>Summary</tt> protocol buffer with a tensor.
+tensorSummary :: (TensorType t) => Tensor v1 t -> Tensor Value ByteString
+
+-- | Computes softmax cross entropy cost and gradients to backpropagate.
+--   
+--   Unlike <tt>SoftmaxCrossEntropyWithLogits</tt>, this operation does not
+--   accept a matrix of label probabilities, but rather a single label per
+--   row of features. This label is considered to have probability 1.0 for
+--   the given row.
+--   
+--   Inputs are the logits, not probabilities.
+sparseSoftmaxCrossEntropyWithLogits :: (TensorType t, OneOf '[Word16, Double, Float] t, TensorType tlabels, OneOf '[Int32, Int64] tlabels) => Tensor v1 t -> Tensor v2 tlabels -> (Tensor Value t, Tensor Value t)
+
+-- | Performs max pooling on the input and outputs both max values and
+--   indices.
+--   
+--   The indices in <tt>argmax</tt> are flattened, so that a maximum value
+--   at position `[b, y, x, c]` becomes flattened index `((b * height + y)
+--   * width + x) * channels + c`.
+maxPoolWithArgmax :: (TensorType t, OneOf '[Word16, Float] t, TensorType targmax, OneOf '[Int32, Int64] targmax) => Tensor v1 t -> (Tensor Value t, Tensor Value targmax)
+
+-- | Compute the 1-dimensional discrete Fourier Transform over the
+--   inner-most
+--   
+--   dimension of <tt>input</tt>.
+fFT :: Tensor v1 (Complex Float) -> Tensor Value (Complex Float)
+
+-- | Outputs a <tt>Summary</tt> protocol buffer with a histogram.
+--   
+--   The generated <a>`Summary`</a> has one summary value containing a
+--   histogram for <tt>values</tt>.
+--   
+--   This op reports an <tt>InvalidArgument</tt> error if any value is not
+--   finite.
+histogramSummary :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 ByteString -> Tensor v2 t -> Tensor Value ByteString
+
+-- | Pads a tensor with zeros.
+--   
+--   This operation pads a <tt>input</tt> with zeros according to the
+--   <tt>paddings</tt> you specify. <tt>paddings</tt> is an integer tensor
+--   with shape `[Dn, 2]`, where n is the rank of <tt>input</tt>. For each
+--   dimension D of <tt>input</tt>, `paddings[D, 0]` indicates how many
+--   zeros to add before the contents of <tt>input</tt> in that dimension,
+--   and `paddings[D, 1]` indicates how many zeros to add after the
+--   contents of <tt>input</tt> in that dimension.
+--   
+--   The padded size of each dimension D of the output is:
+--   
+--   `paddings(D, 0) + input.dim_size(D) + paddings(D, 1)`
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>t</tt> is [[1, 1], [2, 2]] # <tt>paddings</tt> is
+--   [[1, 1], [2, 2]] # rank of <tt>t</tt> is 2 pad(t, paddings) ==&gt;
+--   [[0, 0, 0, 0, 0, 0] [0, 0, 1, 1, 0, 0] [0, 0, 2, 2, 0, 0] [0, 0, 0, 0,
+--   0, 0]] ```
+pad :: (TensorType t, TensorType tpaddings, OneOf '[Int32, Int64] tpaddings) => Tensor v1 t -> Tensor v2 tpaddings -> Tensor Value t
+batchIFFT3D :: Tensor v1 (Complex Float) -> Tensor Value (Complex Float)
+
+-- | Outputs a <tt>Summary</tt> protocol buffer with images.
+--   
+--   The summary has up to <tt>max_images</tt> summary values containing
+--   images. The images are built from <tt>tensor</tt> which must be 4-D
+--   with shape `[batch_size, height, width, channels]` and where
+--   <tt>channels</tt> can be:
+--   
+--   <ul>
+--   <li>1: <tt>tensor</tt> is interpreted as Grayscale.</li>
+--   <li>3: <tt>tensor</tt> is interpreted as RGB.</li>
+--   <li>4: <tt>tensor</tt> is interpreted as RGBA.</li>
+--   </ul>
+--   
+--   The images have the same number of channels as the input tensor. For
+--   float input, the values are normalized one image at a time to fit in
+--   the range `[0, 255]`. <tt>uint8</tt> values are unchanged. The op uses
+--   two different normalization algorithms:
+--   
+--   <ul>
+--   <li>If the input values are all positive, they are rescaled so the
+--   largest one is 255.</li>
+--   <li>If any input value is negative, the values are shifted so input
+--   value 0.0 is at 127. They are then rescaled so that either the
+--   smallest value is 0, or the largest one is 255.</li>
+--   </ul>
+--   
+--   The <tt>tag</tt> argument is a scalar <a>Tensor</a> of type
+--   <tt>string</tt>. It is used to build the <tt>tag</tt> of the summary
+--   values:
+--   
+--   <ul>
+--   <li>If <tt>max_images</tt> is 1, the summary value tag is
+--   '*tag*/image'.</li>
+--   <li>If <tt>max_images</tt> is greater than 1, the summary value tags
+--   are generated sequentially as '*tag*/image/0', '*tag*/image/1',
+--   etc.</li>
+--   </ul>
+--   
+--   The <tt>bad_color</tt> argument is the color to use in the generated
+--   images for non-finite input values. It is a <tt>unit8</tt> 1-D tensor
+--   of length <tt>channels</tt>. Each element must be in the range `[0,
+--   255]` (It represents the value of a pixel in the output image).
+--   Non-finite values in the input tensor are replaced by this tensor in
+--   the output image. The default value is the color red.
+imageSummary :: (TensorType t, OneOf '[Word16, Word8, Float] t) => Tensor v1 ByteString -> Tensor v2 t -> Tensor Value ByteString
+
+-- | Computes the sum along segments of a tensor.
+--   
+--   Read <a>the section on Segmentation</a> for an explanation of
+--   segments.
+--   
+--   Computes a tensor such that \(output_i = sum_j data_j\) where sum is
+--   over <tt>j</tt> such that `segment_ids[j] == i`.
+--   
+--   <a>style="width:70%; margin:auto; margin-bottom:10px;
+--   margin-top:20px;"</a> <a>style="width:100%"
+--   src="../../images/SegmentSum.png" alt</a> <a>/div</a>
+segmentSum :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 tindices -> Tensor Value t
+
+-- | JPEG-encode an image.
+--   
+--   <tt>image</tt> is a 3-D uint8 Tensor of shape `[height, width,
+--   channels]`.
+--   
+--   The attr <tt>format</tt> can be used to override the color format of
+--   the encoded output. Values can be:
+--   
+--   <ul>
+--   <li>`''`: Use a default format based on the number of channels in the
+--   image.</li>
+--   <li><tt>grayscale</tt>: Output a grayscale JPEG image. The
+--   <tt>channels</tt> dimension of <tt>image</tt> must be 1.</li>
+--   <li><tt>rgb</tt>: Output an RGB JPEG image. The <tt>channels</tt>
+--   dimension of <tt>image</tt> must be 3.</li>
+--   </ul>
+--   
+--   If <tt>format</tt> is not specified or is the empty string, a default
+--   format is picked in function of the number of channels in
+--   <tt>image</tt>:
+--   
+--   <ul>
+--   <li>1: Output a grayscale image.</li>
+--   <li>3: Output an RGB image.</li>
+--   </ul>
+encodeJpeg :: Tensor v1 Word8 -> Tensor Value ByteString
+
+-- | Gradients for batch normalization.
+--   
+--   This op is deprecated. See `tf.nn.batch_normalization`.
+batchNormWithGlobalNormalizationGrad :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Bool -> Float -> Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 t -> (Tensor Value t, Tensor Value t, Tensor Value t, Tensor Value t, Tensor Value t)
+
+-- | Adds <tt>bias</tt> to <a>value</a>.
+--   
+--   This is a deprecated version of BiasAdd and will be soon removed.
+--   
+--   This is a special case of `tf.add` where <tt>bias</tt> is restricted
+--   to be 1-D. Broadcasting is supported, so <a>value</a> may have any
+--   number of dimensions.
+biasAddV1 :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Computes the inverse permutation of a tensor.
+--   
+--   This operation computes the inverse of an index permutation. It takes
+--   a 1-D integer tensor <tt>x</tt>, which represents the indices of a
+--   zero-based array, and swaps each value with its index position. In
+--   other words, for an output tensor <tt>y</tt> and an input tensor
+--   <tt>x</tt>, this operation computes the following:
+--   
+--   `y[x[i]] = i for i in [0, 1, ..., len(x) - 1]`
+--   
+--   The values must include 0. There can be no duplicate values or
+--   negative values.
+--   
+--   For example:
+--   
+--   ```prettyprint # tensor <tt>x</tt> is [3, 4, 0, 2, 1]
+--   invert_permutation(x) ==&gt; [2, 4, 3, 0, 1] ```
+invertPermutation :: (TensorType t, OneOf '[Int32, Int64] t) => Tensor v1 t -> Tensor Value t
+
+-- | Gradient op for <tt>MirrorPad</tt> op. This op folds a mirror-padded
+--   tensor.
+--   
+--   This operation folds the padded areas of <tt>input</tt> by
+--   <tt>MirrorPad</tt> according to the <tt>paddings</tt> you specify.
+--   <tt>paddings</tt> must be the same as <tt>paddings</tt> argument given
+--   to the corresponding <tt>MirrorPad</tt> op.
+--   
+--   The folded size of each dimension D of the output is:
+--   
+--   `input.dim_size(D) - paddings(D, 0) - paddings(D, 1)`
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>t</tt> is [[1, 2, 3], [4, 5, 6], [7, 8, 9]]. #
+--   <tt>paddings</tt> is [[0, 1]], [0, 1]]. # <tt>mode</tt> is SYMMETRIC.
+--   # rank of <tt>t</tt> is 2. pad(t, paddings) ==&gt; [[ 1, 5] [11, 28]]
+--   ```
+mirrorPadGrad :: (TensorType t, TensorType tpaddings, OneOf '[Int32, Int64] tpaddings) => Tensor v1 t -> Tensor v2 tpaddings -> Tensor Value t
+
+-- | Reverses specific dimensions of a tensor.
+--   
+--   Given a <tt>tensor</tt>, and a <tt>bool</tt> tensor <tt>dims</tt>
+--   representing the dimensions of <tt>tensor</tt>, this operation
+--   reverses each dimension i of <tt>tensor</tt> where `dims[i]` is
+--   <a>True</a>.
+--   
+--   <tt>tensor</tt> can have up to 8 dimensions. The number of dimensions
+--   of <tt>tensor</tt> must equal the number of elements in <tt>dims</tt>.
+--   In other words:
+--   
+--   `rank(tensor) = size(dims)`
+--   
+--   For example:
+--   
+--   ```prettyprint # tensor <tt>t</tt> is [[[[ 0, 1, 2, 3], # [ 4, 5, 6,
+--   7], # [ 8, 9, 10, 11]], # [[12, 13, 14, 15], # [16, 17, 18, 19], #
+--   [20, 21, 22, 23]]]] # tensor <tt>t</tt> shape is [1, 2, 3, 4]
+--   
+--   # <tt>dims</tt> is [False, False, False, True] reverse(t, dims) ==&gt;
+--   [[[[ 3, 2, 1, 0], [ 7, 6, 5, 4], [ 11, 10, 9, 8]], [[15, 14, 13, 12],
+--   [19, 18, 17, 16], [23, 22, 21, 20]]]]
+--   
+--   # <tt>dims</tt> is [False, True, False, False] reverse(t, dims) ==&gt;
+--   [[[[12, 13, 14, 15], [16, 17, 18, 19], [20, 21, 22, 23] [[ 0, 1, 2,
+--   3], [ 4, 5, 6, 7], [ 8, 9, 10, 11]]]]
+--   
+--   # <tt>dims</tt> is [False, False, True, False] reverse(t, dims) ==&gt;
+--   [[[[8, 9, 10, 11], [4, 5, 6, 7], [0, 1, 2, 3]] [[20, 21, 22, 23], [16,
+--   17, 18, 19], [12, 13, 14, 15]]]] ```
+reverse :: (TensorType t, OneOf '[Complex Double, Complex Float, Bool, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 Bool -> Tensor Value t
+
+-- | Computes a 2-D convolution given 4-D <tt>input</tt> and <a>filter</a>
+--   tensors.
+--   
+--   Given an input tensor of shape `[batch, in_height, in_width,
+--   in_channels]` and a filter / kernel tensor of shape `[filter_height,
+--   filter_width, in_channels, out_channels]`, this op performs the
+--   following:
+--   
+--   <ol>
+--   <li>Flattens the filter to a 2-D matrix with shape `[filter_height *
+--   filter_width * in_channels, output_channels]`.</li>
+--   <li>Extracts image patches from the input tensor to form a *virtual*
+--   tensor of shape `[batch, out_height, out_width, filter_height *
+--   filter_width * in_channels]`.</li>
+--   <li>For each patch, right-multiplies the filter matrix and the image
+--   patch vector.</li>
+--   </ol>
+--   
+--   In detail, with the default NHWC format,
+--   
+--   output[b, i, j, k] = sum_{di, dj, q} input[b, strides[1] * i + di,
+--   strides[2] * j + dj, q] * filter[di, dj, q, k]
+--   
+--   Must have `strides[0] = strides[3] = 1`. For the most common case of
+--   the same horizontal and vertices strides, `strides = [1, stride,
+--   stride, 1]`.
+conv2D :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Computes the gradients of convolution with respect to the input.
+conv2DBackpropInput :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 Int32 -> Tensor v2 t -> Tensor v3 t -> Tensor Value t
+
+-- | Produce a string tensor that encodes the state of a Reader.
+--   
+--   Not all Readers support being serialized, so this can produce an
+--   Unimplemented error.
+readerSerializeState :: Tensor v1 ByteString -> Tensor Value ByteString
+
+-- | Returns a tensor that may be mutated, but only persists within a
+--   single step.
+--   
+--   This is an experimental op for internal use only and it is possible to
+--   use this op in unsafe ways. DO NOT USE unless you fully understand the
+--   risks.
+--   
+--   It is the caller's responsibility to ensure that <tt>ref</tt> is
+--   eventually passed to a matching <tt>DestroyTemporaryVariable</tt> op
+--   after all other uses have completed.
+--   
+--   Outputs a ref to the tensor state so it may be read or modified.
+--   
+--   E.g. var = state_ops._temporary_variable([1, 2], types.float_)
+--   var_name = var.op.name var = state_ops.assign(var, [[4.0, 5.0]]) var =
+--   state_ops.assign_add(var, [[6.0, 7.0]]) final =
+--   state_ops._destroy_temporary_variable(var, var_name=var_name)
+temporaryVariable :: (TensorType dtype) => Tensor Value dtype
+
+-- | Extracts crops from the input image tensor and bilinearly resizes them
+--   (possibly
+--   
+--   with aspect ratio change) to a common output size specified by
+--   <tt>crop_size</tt>. This is more general than the
+--   <tt>crop_to_bounding_box</tt> op which extracts a fixed size slice
+--   from the input image and does not allow resizing or aspect ratio
+--   change.
+--   
+--   Returns a tensor with <tt>crops</tt> from the input <tt>image</tt> at
+--   positions defined at the bounding box locations in <tt>boxes</tt>. The
+--   cropped boxes are all resized (with bilinear interpolation) to a fixed
+--   `size = [crop_height, crop_width]`. The result is a 4-D tensor
+--   `[num_boxes, crop_height, crop_width, depth]`.
+cropAndResize :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 Float -> Tensor v3 Int32 -> Tensor v4 Int32 -> Tensor Value Float
+
+-- | Computes gradients of the maxpooling function.
+maxPoolGrad :: (TensorType t, OneOf '[Word16, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor Value t
+
+-- | Performs a resize and padding as a preprocess during a convolution.
+--   
+--   It's often possible to do spatial transformations more efficiently as
+--   part of the packing stage of a convolution, so this op allows for an
+--   optimized implementation where these stages are fused together. This
+--   prevents the need to write out the intermediate results as whole
+--   tensors, reducing memory pressure, and we can get some latency gains
+--   by merging the transformation calculations. The data_format attribute
+--   for Conv2D isn't supported by this op, and defaults to <tt>NHWC</tt>
+--   order. Internally this op uses a single per-graph scratch buffer,
+--   which means that it will block if multiple versions are being run in
+--   parallel. This is because this operator is primarily an optimization
+--   to minimize memory usage.
+fusedResizeAndPadConv2D :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor v2 Int32 -> Tensor v3 Int32 -> Tensor v4 t -> Tensor Value t
+
+-- | Outputs random values from a uniform distribution.
+--   
+--   The generated values follow a uniform distribution in the range `[0,
+--   1)`. The lower bound 0 is included in the range, while the upper bound
+--   1 is excluded.
+randomUniform :: (TensorType t, OneOf '[Int32, Int64] t, TensorType dtype, OneOf '[Word16, Double, Float] dtype) => Tensor v1 t -> Tensor Value dtype
+
+-- | Computes a 2-D depthwise convolution given 4-D <tt>input</tt> and
+--   <a>filter</a> tensors.
+--   
+--   Given an input tensor of shape `[batch, in_height, in_width,
+--   in_channels]` and a filter / kernel tensor of shape `[filter_height,
+--   filter_width, in_channels, channel_multiplier]`, containing
+--   <tt>in_channels</tt> convolutional filters of depth 1,
+--   <tt>depthwise_conv2d</tt> applies a different filter to each input
+--   channel (expanding from 1 channel to <tt>channel_multiplier</tt>
+--   channels for each), then concatenates the results together. Thus, the
+--   output has `in_channels * channel_multiplier` channels.
+--   
+--   for k in 0..in_channels-1 for q in 0..channel_multiplier-1 output[b,
+--   i, j, k * channel_multiplier + q] = sum_{di, dj} input[b, strides[1] *
+--   i + di, strides[2] * j + dj, k] * filter[di, dj, k, q]
+--   
+--   Must have `strides[0] = strides[3] = 1`. For the most common case of
+--   the same horizontal and vertices strides, `strides = [1, stride,
+--   stride, 1]`.
+depthwiseConv2dNative :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | var: Should be from a Variable().
+sparseApplyAdadelta :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 t -> Tensor v6 t -> Tensor v7 t -> Tensor v8 tindices -> Tensor Value t
+
+-- | Computes the gradients of depthwise convolution with respect to the
+--   filter.
+depthwiseConv2dNativeBackpropFilter :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor v2 Int32 -> Tensor v3 t -> Tensor Value t
+
+-- | Computes a 3-D convolution given 5-D <tt>input</tt> and <a>filter</a>
+--   tensors.
+--   
+--   In signal processing, cross-correlation is a measure of similarity of
+--   two waveforms as a function of a time-lag applied to one of them. This
+--   is also known as a sliding dot product or sliding inner-product.
+--   
+--   Our Conv3D implements a form of cross-correlation.
+conv3D :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Returns the truth value of (x &gt;= y) element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>GreaterEqual</tt> supports broadcasting. More about
+--   broadcasting <a>here</a></li>
+--   </ul>
+greaterEqual :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value Bool
+
+-- | Adds up a SparseTensor and a dense Tensor, using these special rules:
+--   
+--   <ol>
+--   <li>Broadcasts the dense side to have the same shape as the sparse
+--   side, if eligible;</li>
+--   <li>Then, only the dense values pointed to by the indices of the
+--   SparseTensor participate in the cwise addition.</li>
+--   </ol>
+--   
+--   By these rules, the result is a logical SparseTensor with exactly the
+--   same indices and shape, but possibly with different non-zero values.
+--   The output of this Op is the resultant non-zero values.
+sparseDenseCwiseAdd :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 Int64 -> Tensor v2 t -> Tensor v3 Int64 -> Tensor v4 t -> Tensor Value t
+
+-- | Computes the gradients of 3-D convolution with respect to the filter.
+conv3DBackpropFilter :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor Value t
+
+-- | Computes the gradients of 3-D convolution with respect to the input.
+conv3DBackpropInputV2 :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 Int32 -> Tensor v2 t -> Tensor v3 t -> Tensor Value t
+
+-- | Returns element-wise remainder of division.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>Mod</tt> supports broadcasting. More about broadcasting
+--   <a>here</a></li>
+--   </ul>
+mod :: (TensorType t, OneOf '[Int32, Int64, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Forwards the value of an available tensor from <tt>inputs</tt> to
+--   <tt>output</tt>.
+--   
+--   <tt>Merge</tt> waits for at least one of the tensors in
+--   <tt>inputs</tt> to become available. It is usually combined with
+--   <tt>Switch</tt> to implement branching.
+--   
+--   <tt>Merge</tt> forwards the first tensor for become available to
+--   <tt>output</tt>, and sets <tt>value_index</tt> to its index in
+--   <tt>inputs</tt>.
+refMerge :: (TensorType t) => [Tensor v1 t] -> (Tensor Value t, Tensor Value Int32)
+
+-- | Computes the gradients of 3-D convolution with respect to the filter.
+conv3DBackpropFilterV2 :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 Int32 -> Tensor v3 t -> Tensor Value t
+
+-- | Serialize an <tt>N</tt>-minibatch <tt>SparseTensor</tt> into an `[N,
+--   3]` string <a>Tensor</a>.
+--   
+--   The <tt>SparseTensor</tt> must have rank <tt>R</tt> greater than 1,
+--   and the first dimension is treated as the minibatch dimension.
+--   Elements of the <tt>SparseTensor</tt> must be sorted in increasing
+--   order of this first dimension. The serialized <tt>SparseTensor</tt>
+--   objects going into each row of <tt>serialized_sparse</tt> will have
+--   rank `R-1`.
+--   
+--   The minibatch size <tt>N</tt> is extracted from `sparse_shape[0]`.
+serializeManySparse :: (TensorType t) => Tensor v1 Int64 -> Tensor v2 t -> Tensor v3 Int64 -> Tensor Value ByteString
+
+-- | Computes gradients of average pooling function.
+avgPool3DGrad :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 Int32 -> Tensor v2 t -> Tensor Value t
+
+-- | Computes gradients of max pooling function.
+maxPool3DGrad :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 Float -> Tensor v2 Float -> Tensor v3 t -> Tensor Value t
+
+-- | Computes the sum of elements across dimensions of a SparseTensor.
+--   
+--   This Op takes a SparseTensor and is the sparse counterpart to
+--   `tf.reduce_sum()`. In particular, this Op also returns a dense
+--   <a>Tensor</a> instead of a sparse one.
+--   
+--   Reduces <tt>sp_input</tt> along the dimensions given in
+--   <tt>reduction_axes</tt>. Unless <tt>keep_dims</tt> is true, the rank
+--   of the tensor is reduced by 1 for each entry in
+--   <tt>reduction_axes</tt>. If <tt>keep_dims</tt> is true, the reduced
+--   dimensions are retained with length 1.
+--   
+--   If <tt>reduction_axes</tt> has no entries, all dimensions are reduced,
+--   and a tensor with a single element is returned. Additionally, the axes
+--   can be negative, which are interpreted according to the indexing rules
+--   in Python.
+sparseReduceSum :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 Int64 -> Tensor v2 t -> Tensor v3 Int64 -> Tensor v4 Int32 -> Tensor Value t
+
+-- | Computes rectified linear: `max(features, 0)`.
+relu :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | L2 Loss.
+--   
+--   Computes half the L2 norm of a tensor without the <a>sqrt</a>:
+--   
+--   output = sum(t ** 2) / 2
+l2Loss :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Restore a reader to a previously saved state.
+--   
+--   Not all Readers support being restored, so this can produce an
+--   Unimplemented error.
+readerRestoreState :: Tensor v1 ByteString -> Tensor v2 ByteString -> ControlNode
+
+-- | Returns the shape of a tensor.
+--   
+--   This operation returns a 1-D integer tensor representing the shape of
+--   <tt>input</tt>.
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>t</tt> is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3],
+--   [4, 4, 4]]] shape(t) ==&gt; [2, 2, 3] ```
+shape :: (TensorType t, TensorType out_type, OneOf '[Int32, Int64] out_type) => Tensor v1 t -> Tensor Value out_type
+
+-- | Computes softmax cross entropy cost and gradients to backpropagate.
+--   
+--   Inputs are the logits, not probabilities.
+softmaxCrossEntropyWithLogits :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> (Tensor Value t, Tensor Value t)
+
+-- | Performs max pooling on the input.
+maxPool :: (TensorType t, OneOf '[Word16, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes the gradient of morphological 2-D dilation with respect to
+--   the input.
+dilation2DBackpropInput :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor Value t
+
+-- | Returns the truth value of (x == y) element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>Equal</tt> supports broadcasting. More about
+--   broadcasting <a>here</a></li>
+--   </ul>
+equal :: (TensorType t, OneOf '[Complex Double, Complex Float, Bool, ByteString, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value Bool
+
+-- | Computes the gradient of morphological 2-D dilation with respect to
+--   the filter.
+dilation2DBackpropFilter :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor Value t
+
+-- | Computes rectified linear gradients for a Relu operation.
+reluGrad :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Computes rectified linear 6: `min(max(features, 0), 6)`.
+relu6 :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Resize <tt>images</tt> to <a>size</a> using bicubic interpolation.
+--   
+--   Input images can be of different types but output images are always
+--   float.
+resizeBicubic :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 Int32 -> Tensor Value Float
+
+-- | Computes rectified linear 6 gradients for a Relu6 operation.
+relu6Grad :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Multiply SparseTensor (of rank 2) <a>A</a> by dense matrix <a>B</a>.
+--   
+--   No validity checking is performed on the indices of A. However, the
+--   following input format is recommended for optimal behavior:
+--   
+--   if adjoint_a == false: A should be sorted in lexicographically
+--   increasing order. Use SparseReorder if you're not sure. if adjoint_a
+--   == true: A should be sorted in order of increasing dimension 1 (i.e.,
+--   "column major" order instead of "row major" order).
+sparseTensorDenseMatMul :: (TensorType t) => Tensor v1 Int64 -> Tensor v2 t -> Tensor v3 Int64 -> Tensor v4 t -> Tensor Value t
+
+-- | Computes softplus: `log(exp(features) + 1)`.
+softplus :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Multiplies slices of two tensors in batches.
+--   
+--   Multiplies all slices of <a>Tensor</a> <tt>x</tt> and <tt>y</tt> (each
+--   slice can be viewed as an element of a batch), and arranges the
+--   individual results in a single output tensor of the same batch size.
+--   Each of the individual slices can optionally be adjointed (to adjoint
+--   a matrix means to transpose and conjugate it) before multiplication by
+--   setting the <tt>adj_x</tt> or <tt>adj_y</tt> flag to <a>True</a>,
+--   which are by default <a>False</a>.
+--   
+--   The input tensors <tt>x</tt> and <tt>y</tt> are 3-D or higher with
+--   shape `[..., r_x, c_x]` and `[..., r_y, c_y]`.
+--   
+--   The output tensor is 3-D or higher with shape `[..., r_o, c_o]`,
+--   where:
+--   
+--   r_o = c_x if adj_x else r_x c_o = r_y if adj_y else c_y
+--   
+--   It is computed as:
+--   
+--   output[..., :, :] = matrix(x[..., :, :]) * matrix(y[..., :, :])
+batchMatMul :: (TensorType t, OneOf '[Complex Double, Complex Float, Int32, Word16, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Computes softsign gradients for a softsign operation.
+softsignGrad :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Returns the truth value of (x &lt;= y) element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>LessEqual</tt> supports broadcasting. More about
+--   broadcasting <a>here</a></li>
+--   </ul>
+lessEqual :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value Bool
+
+-- | Computes log softmax activations.
+--   
+--   For each batch <tt>i</tt> and class <tt>j</tt> we have
+--   
+--   logsoftmax[i, j] = logits[i, j] - log(sum(exp(logits[i])))
+logSoftmax :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Says whether the targets are in the top <tt>K</tt> predictions.
+--   
+--   This outputs a <tt>batch_size</tt> bool array, an entry `out[i]` is
+--   <tt>true</tt> if the prediction for the target class is among the top
+--   <tt>k</tt> predictions among all predictions for example <tt>i</tt>.
+--   Note that the behavior of <tt>InTopK</tt> differs from the
+--   <tt>TopK</tt> op in its handling of ties; if multiple classes have the
+--   same prediction value and straddle the top-<tt>k</tt> boundary, all of
+--   those classes are considered to be in the top <tt>k</tt>.
+--   
+--   More formally, let
+--   
+--   \(predictions_i\) be the predictions for all classes for example
+--   <tt>i</tt>, \(targets_i\) be the target class for example <tt>i</tt>,
+--   \(out_i\) be the output for example <tt>i</tt>,
+--   
+--   $$out_i = predictions_{i, targets_i} in
+--   TopKIncludingTies(predictions_i)$$
+inTopK :: (TensorType t, OneOf '[Int32, Int64] t) => Int64 -> Tensor v1 Float -> Tensor v2 t -> Tensor Value Bool
+
+-- | Returns a batched diagonal tensor with a given batched diagonal
+--   values.
+--   
+--   Given a <tt>diagonal</tt>, this operation returns a tensor with the
+--   <tt>diagonal</tt> and everything else padded with zeros. The diagonal
+--   is computed as follows:
+--   
+--   Assume <tt>diagonal</tt> has <tt>k</tt> dimensions `[I, J, K, ...,
+--   N]`, then the output is a tensor of rank `k+1` with dimensions [I, J,
+--   K, ..., N, N]` where:
+--   
+--   `output[i, j, k, ..., m, n] = 1{m=n} * diagonal[i, j, k, ..., n]`.
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>diagonal</tt> is [[1, 2, 3, 4], [5, 6, 7, 8]]
+--   
+--   and diagonal.shape = (2, 4)
+--   
+--   tf.matrix_diag(diagonal) ==&gt; [[[1, 0, 0, 0] [0, 2, 0, 0] [0, 0, 3,
+--   0] [0, 0, 0, 4]], [[5, 0, 0, 0] [0, 6, 0, 0] [0, 0, 7, 0] [0, 0, 0,
+--   8]]]
+--   
+--   which has shape (2, 4, 4) ```
+matrixDiag :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Performs 3D max pooling on the input.
+maxPool3D :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Finds values and indices of the <tt>k</tt> largest elements for the
+--   last dimension.
+--   
+--   If the input is a vector (rank-1), finds the <tt>k</tt> largest
+--   entries in the vector and outputs their values and indices as vectors.
+--   Thus `values[j]` is the <tt>j</tt>-th largest entry in <tt>input</tt>,
+--   and its index is `indices[j]`.
+--   
+--   For matrices (resp. higher rank input), computes the top <tt>k</tt>
+--   entries in each row (resp. vector along the last dimension). Thus,
+--   
+--   values.shape = indices.shape = input.shape[:-1] + [k]
+--   
+--   If two elements are equal, the lower-index element appears first.
+--   
+--   If <tt>k</tt> varies dynamically, use <tt>TopKV2</tt> below.
+topK :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Int64 -> Tensor v1 t -> (Tensor Value t, Tensor Value Int32)
+
+-- | Finds values and indices of the <tt>k</tt> largest elements for the
+--   last dimension.
+--   
+--   If the input is a vector (rank-1), finds the <tt>k</tt> largest
+--   entries in the vector and outputs their values and indices as vectors.
+--   Thus `values[j]` is the <tt>j</tt>-th largest entry in <tt>input</tt>,
+--   and its index is `indices[j]`.
+--   
+--   For matrices (resp. higher rank input), computes the top <tt>k</tt>
+--   entries in each row (resp. vector along the last dimension). Thus,
+--   
+--   values.shape = indices.shape = input.shape[:-1] + [k]
+--   
+--   If two elements are equal, the lower-index element appears first.
+--   
+--   This is the same as <tt>TopK</tt>, but takes <tt>k</tt> as in input
+--   rather than an attr.
+topKV2 :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 Int32 -> (Tensor Value t, Tensor Value Int32)
+
+-- | Performs fractional max pooling on the input.
+--   
+--   Fractional max pooling is slightly different than regular max pooling.
+--   In regular max pooling, you downsize an input set by taking the
+--   maximum value of smaller N x N subsections of the set (often 2x2), and
+--   try to reduce the set by a factor of N, where N is an integer.
+--   Fractional max pooling, as you might expect from the word
+--   "fractional", means that the overall reduction ratio N does not have
+--   to be an integer.
+--   
+--   The sizes of the pooling regions are generated randomly but are fairly
+--   uniform. For example, let's look at the height dimension, and the
+--   constraints on the list of rows that will be pool boundaries.
+--   
+--   First we define the following:
+--   
+--   <ol>
+--   <li>input_row_length : the number of rows from the input set</li>
+--   <li>output_row_length : which will be smaller than the input</li>
+--   <li>alpha = input_row_length / output_row_length : our reduction
+--   ratio</li>
+--   <li>K = floor(alpha)</li>
+--   <li>row_pooling_sequence : this is the result list of pool boundary
+--   rows</li>
+--   </ol>
+--   
+--   Then, row_pooling_sequence should satisfy:
+--   
+--   <ol>
+--   <li>a[0] = 0 : the first value of the sequence is 0</li>
+--   <li>a[end] = input_row_length : the last value of the sequence is the
+--   size</li>
+--   <li>K &lt;= (a[i+1] - a[i]) &lt;= K+1 : all intervals are K or K+1
+--   size</li>
+--   <li>length(row_pooling_sequence) = output_row_length+1</li>
+--   </ol>
+--   
+--   For more details on fractional max pooling, see this paper:
+--   <a>Benjamin Graham, Fractional Max-Pooling</a>
+fractionalMaxPool :: (TensorType t, OneOf '[Int32, Int64, Double, Float] t) => Tensor v1 t -> (Tensor Value t, Tensor Value Int64, Tensor Value Int64)
+
+-- | Copy a tensor setting everything outside a central band in each
+--   innermost matrix
+--   
+--   to zero.
+--   
+--   The <tt>band</tt> part is computed as follows: Assume <tt>input</tt>
+--   has <tt>k</tt> dimensions `[I, J, K, ..., M, N]`, then the output is a
+--   tensor with the same shape where
+--   
+--   `band[i, j, k, ..., m, n] = in_band(m, n) * input[i, j, k, ..., m,
+--   n]`.
+--   
+--   The indicator function 'in_band(m, n)` is one if `(num_lower &lt; 0 ||
+--   (m-n) &lt;= num_lower)) &amp;&amp; (num_upper &lt; 0 || (n-m) &lt;=
+--   num_upper)`, and zero otherwise.
+--   
+--   For example:
+--   
+--   ```prettyprint # if <tt>input</tt> is [[ 0, 1, 2, 3] [-1, 0, 1, 2]
+--   [-2, -1, 0, 1] [-3, -2, -1, 0]],
+--   
+--   tf.matrix_band_part(input, 1, -1) ==&gt; [[ 0, 1, 2, 3] [-1, 0, 1, 2]
+--   [ 0, -1, 0, 1] [ 0, 0, -1, 0]],
+--   
+--   tf.matrix_band_part(input, 2, 1) ==&gt; [[ 0, 1, 0, 0] [-1, 0, 1, 0]
+--   [-2, -1, 0, 1] [ 0, -2, -1, 0]] ```
+--   
+--   Useful special cases:
+--   
+--   ```prettyprint tf.matrix_band_part(input, 0, -1) ==&gt; Upper
+--   triangular part. tf.matrix_band_part(input, -1, 0) ==&gt; Lower
+--   triangular part. tf.matrix_band_part(input, 0, 0) ==&gt; Diagonal. ```
+matrixBandPart :: (TensorType t) => Tensor v1 t -> Tensor v2 Int64 -> Tensor v3 Int64 -> Tensor Value t
+
+-- | Reinterpret the bytes of a string as a vector of numbers.
+decodeRaw :: (TensorType out_type, OneOf '[Int16, Int32, Int64, Int8, Word8, Double, Float] out_type) => Tensor v1 ByteString -> Tensor Value out_type
+
+-- | Convert JSON-encoded Example records to binary protocol buffer
+--   strings.
+--   
+--   This op translates a tensor containing Example records, encoded using
+--   the <a>standard JSON mapping</a>, into a tensor containing the same
+--   records encoded as binary protocol buffers. The resulting tensor can
+--   then be fed to any of the other Example-parsing ops.
+decodeJSONExample :: Tensor v1 ByteString -> Tensor Value ByteString
+
+-- | Outputs random values from a truncated normal distribution.
+--   
+--   The generated values follow a normal distribution with mean 0 and
+--   standard deviation 1, except that values whose magnitude is more than
+--   2 standard deviations from the mean are dropped and re-picked.
+truncatedNormal :: (TensorType t, OneOf '[Int32, Int64] t, TensorType dtype, OneOf '[Word16, Double, Float] dtype) => Tensor v1 t -> Tensor Value dtype
+
+-- | Randomly shuffles a tensor along its first dimension.
+--   
+--   The tensor is shuffled along dimension 0, such that each `value[j]` is
+--   mapped to one and only one `output[i]`. For example, a mapping that
+--   might occur for a 3x2 tensor is:
+--   
+--   ```prettyprint [[1, 2], [[5, 6], [3, 4], ==&gt; [1, 2], [5, 6]] [3,
+--   4]] ```
+randomShuffle :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Draws samples from a multinomial distribution.
+multinomial :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 Int32 -> Tensor Value Int64
+
+-- | Outputs random values from the Gamma distribution(s) described by
+--   alpha.
+--   
+--   This op uses the algorithm by Marsaglia et al. to acquire samples via
+--   transformation-rejection from pairs of uniform and normal random
+--   variables. See <a>http://dl.acm.org/citation.cfm?id=358414</a>
+randomGamma :: (TensorType s, OneOf '[Int32, Int64] s, TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 s -> Tensor v2 t -> Tensor Value t
+
+-- | Add all input tensors element wise.
+addN :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => [Tensor v1 t] -> Tensor Value t
+
+-- | Computes the maximum of elements across dimensions of a tensor.
+--   
+--   Reduces <tt>input</tt> along the dimensions given in
+--   <tt>reduction_indices</tt>. Unless <tt>keep_dims</tt> is true, the
+--   rank of the tensor is reduced by 1 for each entry in
+--   <tt>reduction_indices</tt>. If <tt>keep_dims</tt> is true, the reduced
+--   dimensions are retained with length 1.
+max :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 t -> Tensor v2 tidx -> Tensor Value t
+
+-- | A graph node which represents a return value of a function.
+_Retval :: (TensorType t) => Int64 -> Tensor v1 t -> ControlNode
+
+-- | Destroys the temporary variable and returns its final value.
+--   
+--   Sets output to the value of the Tensor pointed to by <tt>ref</tt>,
+--   then destroys the temporary variable called <tt>var_name</tt>. All
+--   other uses of <tt>ref</tt> *must* have executed before this op. This
+--   is typically achieved by chaining the ref through each assign op, or
+--   by using control dependencies.
+--   
+--   Outputs the final value of the tensor pointed to by <tt>ref</tt>.
+destroyTemporaryVariable :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Cast x of type SrcT to y of DstT.
+cast :: (TensorType dstT, TensorType srcT) => Tensor v1 srcT -> Tensor Value dstT
+
+-- | Increments <tt>ref</tt> until it reaches <tt>limit</tt>.
+--   
+--   This operation outputs "ref" after the update is done. This makes it
+--   easier to chain operations that need to use the updated value.
+countUpTo :: (TensorType t, OneOf '[Int32, Int64] t) => Int64 -> Tensor v1 t -> Tensor Value t
+
+-- | Computes the absolute value of a tensor.
+--   
+--   Given a tensor <tt>x</tt>, this operation returns a tensor containing
+--   the absolute value of each element in <tt>x</tt>. For example, if x is
+--   an input element and y is an output element, this operation computes
+--   \(y = |x|\).
+abs :: (TensorType t, OneOf '[Int32, Int64, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes numerical negative value element-wise.
+--   
+--   I.e., \(y = -x\).
+neg :: (TensorType t, OneOf '[Complex Double, Complex Float, Int32, Int64, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Returns the element-wise max of two SparseTensors.
+--   
+--   Assumes the two SparseTensors have the same shape, i.e., no
+--   broadcasting.
+sparseSparseMaximum :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 Int64 -> Tensor v2 t -> Tensor v3 Int64 -> Tensor v4 Int64 -> Tensor v5 t -> Tensor v6 Int64 -> (Tensor Value Int64, Tensor Value t)
+
+-- | Computes the gradient for the inverse of <tt>x</tt> wrt its input.
+--   
+--   Specifically, `grad = -dy * y*y`, where `y = 1/x`, and <tt>dy</tt> is
+--   the corresponding input gradient.
+invGrad :: (TensorType t, OneOf '[Complex Double, Complex Float, Word16, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Computes square root of x element-wise.
+--   
+--   I.e., \(y = sqrt{x} = x^{1/2}\).
+sqrt :: (TensorType t, OneOf '[Complex Double, Complex Float, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes the inverse of one or more square invertible matrices or
+--   their
+--   
+--   adjoints (conjugate transposes).
+--   
+--   The input is a tensor of shape `[..., M, M]` whose inner-most 2
+--   dimensions form square matrices. The output is a tensor of the same
+--   shape as the input containing the inverse for all input submatrices
+--   `[..., :, :]`.
+--   
+--   The op uses LU decomposition with partial pivoting to compute the
+--   inverses.
+--   
+--   If a matrix is not invertible there is no guarantee what the op does.
+--   It may detect the condition and raise an exception or it may simply
+--   return a garbage result.
+matrixInverse :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes the gradient for the sqrt of <tt>x</tt> wrt its input.
+--   
+--   Specifically, `grad = dy * 0.5 / y`, where `y = sqrt(x)`, and
+--   <tt>dy</tt> is the corresponding input gradient.
+sqrtGrad :: (TensorType t, OneOf '[Complex Double, Complex Float, Word16, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Inserts a dimension of 1 into a tensor's shape.
+--   
+--   Given a tensor <tt>input</tt>, this operation inserts a dimension of 1
+--   at the dimension index <tt>dim</tt> of <tt>input</tt>'s shape. The
+--   dimension index <tt>dim</tt> starts at zero; if you specify a negative
+--   number for <tt>dim</tt> it is counted backward from the end.
+--   
+--   This operation is useful if you want to add a batch dimension to a
+--   single element. For example, if you have a single image of shape
+--   `[height, width, channels]`, you can make it a batch of 1 image with
+--   `expand_dims(image, 0)`, which will make the shape `[1, height, width,
+--   channels]`.
+--   
+--   Other examples:
+--   
+--   ```prettyprint # <tt>t</tt> is a tensor of shape [2]
+--   shape(expand_dims(t, 0)) ==&gt; [1, 2] shape(expand_dims(t, 1)) ==&gt;
+--   [2, 1] shape(expand_dims(t, -1)) ==&gt; [2, 1]
+--   
+--   # <tt>t2</tt> is a tensor of shape [2, 3, 5] shape(expand_dims(t2, 0))
+--   ==&gt; [1, 2, 3, 5] shape(expand_dims(t2, 2)) ==&gt; [2, 3, 1, 5]
+--   shape(expand_dims(t2, 3)) ==&gt; [2, 3, 5, 1] ```
+--   
+--   This operation requires that:
+--   
+--   `-1-input.dims() &lt;= dim &lt;= input.dims()`
+--   
+--   This operation is related to `squeeze()`, which removes dimensions of
+--   size 1.
+expandDims :: (TensorType t, TensorType tdim, OneOf '[Int32, Int64] tdim) => Tensor v1 t -> Tensor v2 tdim -> Tensor Value t
+
+-- | Computes the "logical and" of elements across dimensions of a tensor.
+--   
+--   Reduces <tt>input</tt> along the dimensions given in
+--   <tt>reduction_indices</tt>. Unless <tt>keep_dims</tt> is true, the
+--   rank of the tensor is reduced by 1 for each entry in
+--   <tt>reduction_indices</tt>. If <tt>keep_dims</tt> is true, the reduced
+--   dimensions are retained with length 1.
+all :: (TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 Bool -> Tensor v2 tidx -> Tensor Value Bool
+
+-- | Performs beam search decoding on the logits given in input.
+--   
+--   A note about the attribute merge_repeated: For the beam search
+--   decoder, this means that if consecutive entries in a beam are the
+--   same, only the first of these is emitted. That is, when the top path
+--   is "A B B B B", "A B" is returned if merge_repeated = True but "A B B
+--   B B" is returned if merge_repeated = False.
+cTCBeamSearchDecoder :: Int64 -> Int64 -> Tensor v1 Float -> Tensor v2 Int32 -> ([Tensor Value Int64], [Tensor Value Int64], [Tensor Value Int64], Tensor Value Float)
+
+-- | Computes reciprocal of square root of x element-wise.
+--   
+--   I.e., \(y = 1 / sqrt{x}\).
+rsqrt :: (TensorType t, OneOf '[Complex Double, Complex Float, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes the gradient for the tanh of <tt>x</tt> wrt its input.
+--   
+--   Specifically, `grad = dy * (1 - y*y)`, where `y = tanh(x)`, and
+--   <tt>dy</tt> is the corresponding input gradient.
+tanhGrad :: (TensorType t, OneOf '[Complex Double, Complex Float, Word16, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Computes sin of x element-wise.
+sin :: (TensorType t, OneOf '[Complex Double, Complex Float, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes the determinant of one ore more square matrices.
+--   
+--   The input is a tensor of shape `[..., M, M]` whose inner-most 2
+--   dimensions form square matrices. The output is a tensor containing the
+--   determinants for all input submatrices `[..., :, :]`.
+matrixDeterminant :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes cos of x element-wise.
+cos :: (TensorType t, OneOf '[Complex Double, Complex Float, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | BatchToSpace for 4-D tensors of type T.
+--   
+--   This is a legacy version of the more general BatchToSpaceND.
+--   
+--   Rearranges (permutes) data from batch into blocks of spatial data,
+--   followed by cropping. This is the reverse transformation of
+--   SpaceToBatch. More specifically, this op outputs a copy of the input
+--   tensor where values from the <tt>batch</tt> dimension are moved in
+--   spatial blocks to the <tt>height</tt> and <tt>width</tt> dimensions,
+--   followed by cropping along the <tt>height</tt> and <tt>width</tt>
+--   dimensions.
+batchToSpace :: (TensorType t, TensorType tidx, OneOf '[Int32, Int64] tidx) => Int64 -> Tensor v1 t -> Tensor v2 tidx -> Tensor Value t
+
+-- | Converts a sparse representation into a dense tensor.
+--   
+--   Builds an array <tt>dense</tt> with shape <tt>output_shape</tt> such
+--   that
+--   
+--   ```prettyprint # If sparse_indices is scalar dense[i] = (i ==
+--   sparse_indices ? sparse_values : default_value)
+--   
+--   # If sparse_indices is a vector, then for each i
+--   dense[sparse_indices[i]] = sparse_values[i]
+--   
+--   # If sparse_indices is an n by d matrix, then for each i in [0, n)
+--   dense[sparse_indices[i][0], ..., sparse_indices[i][d-1]] =
+--   sparse_values[i] ```
+--   
+--   All other values in <tt>dense</tt> are set to <tt>default_value</tt>.
+--   If <tt>sparse_values</tt> is a scalar, all sparse indices are set to
+--   this single value.
+--   
+--   Indices should be sorted in lexicographic order, and indices must not
+--   contain any repeats. If <tt>validate_indices</tt> is true, these
+--   properties are checked during execution.
+sparseToDense :: (TensorType t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 tindices -> Tensor v2 tindices -> Tensor v3 t -> Tensor v4 t -> Tensor Value t
+
+-- | Computes asin of x element-wise.
+asin :: (TensorType t, OneOf '[Complex Double, Complex Float, Int32, Int64, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Returns the index with the smallest value across dimensions of a
+--   tensor.
+argMin :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 t -> Tensor v2 tidx -> Tensor Value Int64
+
+-- | Returns which elements of x are Inf.
+isInf :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor Value Bool
+
+-- | Returns an element-wise indication of the sign of a number.
+--   
+--   `y = sign(x) = -1` if `x <a>0 if `x == 0`; 1 if `x</a> 0`.
+--   
+--   For complex numbers, `y = sign(x) = x / |x|` if `x != 0`, otherwise `y
+--   = 0`.
+sign :: (TensorType t, OneOf '[Complex Double, Complex Float, Int32, Int64, Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Returns x + y element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>Add</tt> supports broadcasting. <tt>AddN</tt> does not.
+--   More about broadcasting <a>here</a></li>
+--   </ul>
+add :: (TensorType t, OneOf '[Complex Double, Complex Float, ByteString, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Update relevant entries in '*var' according to the Ftrl-proximal
+--   scheme.
+--   
+--   That is for rows we have grad for, we update var, accum and linear as
+--   follows: accum_new = accum + grad * grad linear += grad +
+--   (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var quadratic = 1.0
+--   / (accum_new^(lr_power) * lr) + 2 * l2 var = (sign(linear) * l1 -
+--   linear) / quadratic if |linear| &gt; l1 else 0.0 accum = accum_new
+sparseApplyFtrl :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 t -> Tensor v3 t -> Tensor v4 t -> Tensor v5 tindices -> Tensor v6 t -> Tensor v7 t -> Tensor v8 t -> Tensor v9 t -> Tensor Value t
+
+-- | Returns x - y element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>Sub</tt> supports broadcasting. More about broadcasting
+--   <a>here</a></li>
+--   </ul>
+sub :: (TensorType t, OneOf '[Complex Double, Complex Float, Int32, Int64, Word16, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+batchFFT3D :: Tensor v1 (Complex Float) -> Tensor Value (Complex Float)
+
+-- | Computes the sum of elements across dimensions of a SparseTensor.
+--   
+--   This Op takes a SparseTensor and is the sparse counterpart to
+--   `tf.reduce_sum()`. In contrast to SparseReduceSum, this Op returns a
+--   SparseTensor.
+--   
+--   Reduces <tt>sp_input</tt> along the dimensions given in
+--   <tt>reduction_axes</tt>. Unless <tt>keep_dims</tt> is true, the rank
+--   of the tensor is reduced by 1 for each entry in
+--   <tt>reduction_axes</tt>. If <tt>keep_dims</tt> is true, the reduced
+--   dimensions are retained with length 1.
+--   
+--   If <tt>reduction_axes</tt> has no entries, all dimensions are reduced,
+--   and a tensor with a single element is returned. Additionally, the axes
+--   can be negative, which are interpreted according to the indexing rules
+--   in Python.
+sparseReduceSumSparse :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 Int64 -> Tensor v2 t -> Tensor v3 Int64 -> Tensor v4 Int32 -> (Tensor Value Int64, Tensor Value t, Tensor Value Int64)
+
+-- | Adds <tt>bias</tt> to <a>value</a>.
+--   
+--   This is a special case of `tf.add` where <tt>bias</tt> is restricted
+--   to be 1-D. Broadcasting is supported, so <a>value</a> may have any
+--   number of dimensions.
+biasAdd :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Returns x * y element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>Mul</tt> supports broadcasting. More about broadcasting
+--   <a>here</a></li>
+--   </ul>
+mul :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Returns x / y element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>Div</tt> supports broadcasting. More about broadcasting
+--   <a>here</a></li>
+--   </ul>
+div :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Forwards the input to the output.
+--   
+--   This operator represents the loop termination condition used by the
+--   "pivot" switches of a loop.
+loopCond :: Tensor v1 Bool -> Tensor Value Bool
+
+-- | Returns (x - y)(x - y) element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>SquaredDifference</tt> supports broadcasting. More
+--   about broadcasting <a>here</a></li>
+--   </ul>
+squaredDifference :: (TensorType t, OneOf '[Complex Double, Complex Float, Int32, Int64, Word16, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Returns the max of x and y (i.e. x &gt; y ? x : y) element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>Maximum</tt> supports broadcasting. More about
+--   broadcasting <a>here</a></li>
+--   </ul>
+maximum :: (TensorType t, OneOf '[Int32, Int64, Word16, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Generates labels for candidate sampling with a log-uniform
+--   distribution.
+--   
+--   See explanations of candidate sampling and the data formats at
+--   go/candidate-sampling.
+--   
+--   For each batch, this op picks a single set of sampled candidate
+--   labels.
+--   
+--   The advantages of sampling candidates per-batch are simplicity and the
+--   possibility of efficient dense matrix multiplication. The disadvantage
+--   is that the sampled candidates must be chosen independently of the
+--   context and of the true labels.
+logUniformCandidateSampler :: Int64 -> Int64 -> Int64 -> Bool -> Tensor v1 Int64 -> (Tensor Value Int64, Tensor Value Float, Tensor Value Float)
+
+-- | Returns the truth value of (x &lt; y) element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>Less</tt> supports broadcasting. More about
+--   broadcasting <a>here</a></li>
+--   </ul>
+less :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value Bool
+
+-- | Computes the power of one value to another.
+--   
+--   Given a tensor <tt>x</tt> and a tensor <tt>y</tt>, this operation
+--   computes \(x^y\) for corresponding elements in <tt>x</tt> and
+--   <tt>y</tt>. For example:
+--   
+--   ``` # tensor <tt>x</tt> is [[2, 2]], [3, 3]] # tensor <tt>y</tt> is
+--   [[8, 16], [2, 3]] tf.pow(x, y) ==&gt; [[256, 65536], [9, 27]] ```
+pow :: (TensorType t, OneOf '[Complex Double, Complex Float, Int32, Int64, Word16, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Compute the upper regularized incomplete Gamma function `Q(a, x)`.
+--   
+--   The upper regularized incomplete Gamma function is defined as:
+--   
+--   ``` Q(a, x) = Gamma(a, x) / Gamma(a) = 1 - P(a, x) ``` where ```
+--   Gamma(a, x) = int_{x}^{infty} t^{a-1} exp(-t) dt ``` is the upper
+--   incomplete Gama function.
+--   
+--   Note, above `P(a, x)` (<tt>Igamma</tt>) is the lower regularized
+--   complete Gamma function.
+igammac :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Compute the lower regularized incomplete Gamma function `Q(a, x)`.
+--   
+--   The lower regularized incomplete Gamma function is defined as:
+--   
+--   ``` P(a, x) = gamma(a, x) / Gamma(a) = 1 - Q(a, x) ``` where ```
+--   gamma(a, x) = int_{0}^{x} t^{a-1} exp(-t) dt ``` is the lower
+--   incomplete Gamma function.
+--   
+--   Note, above `Q(a, x)` (<tt>Igammac</tt>) is the upper regularized
+--   complete Gamma function.
+igamma :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Compute the Hurwitz zeta function \(zeta(x, q)\).
+--   
+--   The Hurwitz zeta function is defined as:
+--   
+--   ``` zeta(x, q) = sum_{n=0}^{infty} (q + n)^{-x} ```
+zeta :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Returns the imaginary part of a complex number.
+--   
+--   Given a tensor <tt>input</tt> of complex numbers, this operation
+--   returns a tensor of type <tt>float</tt> that is the imaginary part of
+--   each element in <tt>input</tt>. All elements in <tt>input</tt> must be
+--   complex numbers of the form \(a + bj\), where *a* is the real part and
+--   *b* is the imaginary part returned by this operation.
+--   
+--   For example:
+--   
+--   ``` # tensor <tt>input</tt> is [-2.25 + 4.75j, 3.25 + 5.75j]
+--   tf.imag(input) ==&gt; [4.75, 5.75] ```
+imag :: (TensorType t, OneOf '[Complex Double, Complex Float] t, TensorType tout, OneOf '[Double, Float] tout) => Tensor v1 t -> Tensor Value tout
+
+-- | Converts two real numbers to a complex number.
+--   
+--   Given a tensor <a>real</a> representing the real part of a complex
+--   number, and a tensor <a>imag</a> representing the imaginary part of a
+--   complex number, this operation returns complex numbers elementwise of
+--   the form \(a + bj\), where *a* represents the <a>real</a> part and *b*
+--   represents the <a>imag</a> part.
+--   
+--   The input tensors <a>real</a> and <a>imag</a> must have the same
+--   shape.
+--   
+--   For example:
+--   
+--   ``` # tensor <a>real</a> is [2.25, 3.25] # tensor <a>imag</a> is
+--   [4.75, 5.75] tf.complex(real, imag) ==&gt; [[2.25 + 4.75j], [3.25 +
+--   5.75j]] ```
+complex :: (TensorType t, OneOf '[Double, Float] t, TensorType tout, OneOf '[Complex Double, Complex Float] tout) => Tensor v1 t -> Tensor v2 t -> Tensor Value tout
+
+-- | Returns the truth value of (x != y) element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>NotEqual</tt> supports broadcasting. More about
+--   broadcasting <a>here</a></li>
+--   </ul>
+notEqual :: (TensorType t, OneOf '[Complex Double, Complex Float, Bool, ByteString, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value Bool
+
+-- | Computes the complex absolute value of a tensor.
+--   
+--   Given a tensor <tt>x</tt> of complex numbers, this operation returns a
+--   tensor of type <tt>float</tt> or <tt>double</tt> that is the absolute
+--   value of each element in <tt>x</tt>. All elements in <tt>x</tt> must
+--   be complex numbers of the form \(a + bj\). The absolute value is
+--   computed as \( sqrt{a^2 + b^2}\).
+--   
+--   For example:
+--   
+--   ``` # tensor <tt>x</tt> is [[-2.25 + 4.75j], [-3.25 + 5.75j]]
+--   tf.complex_abs(x) ==&gt; [5.25594902, 6.60492229] ```
+complexAbs :: (TensorType t, OneOf '[Complex Double, Complex Float] t, TensorType tout, OneOf '[Double, Float] tout) => Tensor v1 t -> Tensor Value tout
+
+-- | Returns the truth value of x AND y element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>LogicalAnd</tt> supports broadcasting. More about
+--   broadcasting <a>here</a></li>
+--   </ul>
+logicalAnd :: Tensor v1 Bool -> Tensor v2 Bool -> Tensor Value Bool
+batchFFT :: Tensor v1 (Complex Float) -> Tensor Value (Complex Float)
+
+-- | Selects elements from <tt>t</tt> or <tt>e</tt>, depending on
+--   <tt>condition</tt>.
+--   
+--   The <tt>t</tt>, and <tt>e</tt> tensors must all have the same shape,
+--   and the output will also have that shape. The <tt>condition</tt>
+--   tensor must be a scalar if <tt>t</tt> and <tt>e</tt> are scalars. If
+--   <tt>t</tt> and <tt>e</tt> are vectors or higher rank, then
+--   <tt>condition</tt> must be either a vector with size matching the
+--   first dimension of <tt>t</tt>, or must have the same shape as
+--   <tt>t</tt>.
+--   
+--   The <tt>condition</tt> tensor acts as a mask that chooses, based on
+--   the value at each element, whether the corresponding element / row in
+--   the output should be taken from <tt>t</tt> (if true) or <tt>e</tt> (if
+--   false).
+--   
+--   If <tt>condition</tt> is a vector and <tt>t</tt> and <tt>e</tt> are
+--   higher rank matrices, then it chooses which row (outer dimension) to
+--   copy from <tt>t</tt> and <tt>e</tt>. If <tt>condition</tt> has the
+--   same shape as <tt>t</tt> and <tt>e</tt>, then it chooses which element
+--   to copy from <tt>t</tt> and <tt>e</tt>.
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>condition</tt> tensor is [[True, False] # [False,
+--   True]] # <tt>t</tt> is [[1, 2], # [3, 4]] # <tt>e</tt> is [[5, 6], #
+--   [7, 8]] select(condition, t, e) ==&gt; [[1, 6], [7, 4]]
+--   
+--   # <tt>condition</tt> tensor is [True, False] # <tt>t</tt> is [[1, 2],
+--   # [3, 4]] # <tt>e</tt> is [[5, 6], # [7, 8]] select(condition, t, e)
+--   ==&gt; [[1, 2], [7, 8]]
+--   
+--   ```
+select :: (TensorType t) => Tensor v1 Bool -> Tensor v2 t -> Tensor v3 t -> Tensor Value t
+
+-- | Multiply the matrix "a" by the matrix "b".
+--   
+--   The inputs must be two-dimensional matrices and the inner dimension of
+--   "a" (after being transposed if transpose_a is true) must match the
+--   outer dimension of "b" (after being transposed if transposed_b is
+--   true).
+--   
+--   <ul>
+--   <li>Note*: The default kernel implementation for MatMul on GPUs uses
+--   cublas.</li>
+--   </ul>
+matMul :: (TensorType t, OneOf '[Complex Double, Complex Float, Int32, Word16, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Computes Psi, the derivative of Lgamma (the log of the absolute value
+--   of
+--   
+--   `Gamma(x)`), element-wise.
+digamma :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes the gradients of convolution with respect to the filter.
+conv2DBackpropFilter :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor v2 Int32 -> Tensor v3 t -> Tensor Value t
+
+-- | Computes the minimum of elements across dimensions of a tensor.
+--   
+--   Reduces <tt>input</tt> along the dimensions given in
+--   <tt>reduction_indices</tt>. Unless <tt>keep_dims</tt> is true, the
+--   rank of the tensor is reduced by 1 for each entry in
+--   <tt>reduction_indices</tt>. If <tt>keep_dims</tt> is true, the reduced
+--   dimensions are retained with length 1.
+min :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 t -> Tensor v2 tidx -> Tensor Value t
+
+-- | Returns which elements of x are finite.
+isFinite :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor Value Bool
+
+-- | Returns the index with the largest value across dimensions of a
+--   tensor.
+argMax :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 t -> Tensor v2 tidx -> Tensor Value Int64
+
+-- | Computes the mean along segments of a tensor.
+--   
+--   Read <a>the section on Segmentation</a> for an explanation of
+--   segments.
+--   
+--   Computes a tensor such that \(output_i = frac{sum_j data_j}{N}\) where
+--   <a>mean</a> is over <tt>j</tt> such that `segment_ids[j] == i` and
+--   <tt>N</tt> is the total number of values summed.
+--   
+--   <a>style="width:70%; margin:auto; margin-bottom:10px;
+--   margin-top:20px;"</a> <a>style="width:100%"
+--   src="../../images/SegmentMean.png" alt</a> <a>/div</a>
+segmentMean :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 tindices -> Tensor Value t
+
+-- | Compute the cumulative product of the tensor <tt>x</tt> along
+--   <tt>axis</tt>.
+--   
+--   By default, this op performs an inclusive cumprod, which means that
+--   the first element of the input is identical to the first element of
+--   the output: ```prettyprint tf.cumprod([a, b, c]) ==&gt; [a, a * b, a *
+--   b * c] ```
+--   
+--   By setting the <tt>exclusive</tt> kwarg to <a>True</a>, an exclusive
+--   cumprod is performed instead: ```prettyprint tf.cumprod([a, b, c],
+--   exclusive=True) ==&gt; [0, a, a * b] ```
+--   
+--   By setting the <a>reverse</a> kwarg to <a>True</a>, the cumprod is
+--   performed in the opposite direction: ```prettyprint tf.cumprod([a, b,
+--   c], reverse=True) ==&gt; [a * b * c, b * c, c] ``` This is more
+--   efficient than using separate `tf.reverse` ops.
+--   
+--   The <a>reverse</a> and <tt>exclusive</tt> kwargs can also be combined:
+--   ```prettyprint tf.cumprod([a, b, c], exclusive=True, reverse=True)
+--   ==&gt; [b * c, c, 0] ```
+cumprod :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 t -> Tensor v2 tidx -> Tensor Value t
+
+-- | Computes the minimum along segments of a tensor.
+--   
+--   Read <a>the section on Segmentation</a> for an explanation of
+--   segments.
+--   
+--   Computes a tensor such that \(output_i = min_j(data_j)\) where
+--   <a>min</a> is over <tt>j</tt> such that `segment_ids[j] == i`.
+--   
+--   <a>style="width:70%; margin:auto; margin-bottom:10px;
+--   margin-top:20px;"</a> <a>style="width:100%"
+--   src="../../images/SegmentMin.png" alt</a> <a>/div</a>
+segmentMin :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 tindices -> Tensor Value t
+
+-- | Computes the sum along segments of a tensor.
+--   
+--   Read <a>the section on Segmentation</a> for an explanation of
+--   segments.
+--   
+--   Computes a tensor such that `(output[i] = sum_{j...} data[j...]` where
+--   the sum is over tuples `j...` such that `segment_ids[j...] == i`.
+--   Unlike <tt>SegmentSum</tt>, <tt>segment_ids</tt> need not be sorted
+--   and need not cover all values in the full range of valid values.
+--   
+--   If the sum is empty for a given segment ID <tt>i</tt>, `output[i] =
+--   0`.
+--   
+--   <tt>num_segments</tt> should equal the number of distinct segment IDs.
+--   
+--   <a>style="width:70%; margin:auto; margin-bottom:10px;
+--   margin-top:20px;"</a> <a>style="width:100%"
+--   src="../../images/UnsortedSegmentSum.png" alt</a> <a>/div</a>
+unsortedSegmentSum :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 tindices -> Tensor v3 Int32 -> Tensor Value t
+
+-- | A Reader that outputs the records from a TensorFlow Records file.
+tFRecordReader :: Tensor Value ByteString
+
+-- | Computes the sum along sparse segments of a tensor.
+--   
+--   Read <a>the section on Segmentation</a> for an explanation of
+--   segments.
+--   
+--   Like <tt>SegmentSum</tt>, but <tt>segment_ids</tt> can have rank less
+--   than `data`'s first dimension, selecting a subset of dimension 0,
+--   specified by <tt>indices</tt>.
+--   
+--   For example:
+--   
+--   ```prettyprint c = tf.constant([[1,2,3,4], [-1,-2,-3,-4], [5,6,7,8]])
+--   
+--   # Select two rows, one segment. tf.sparse_segment_sum(c,
+--   tf.constant([0, 1]), tf.constant([0, 0])) ==&gt; [[0 0 0 0]]
+--   
+--   # Select two rows, two segment. tf.sparse_segment_sum(c,
+--   tf.constant([0, 1]), tf.constant([0, 1])) ==&gt; [[ 1 2 3 4] [-1 -2 -3
+--   -4]]
+--   
+--   # Select all rows, two segments. tf.sparse_segment_sum(c,
+--   tf.constant([0, 1, 2]), tf.constant([0, 0, 1])) ==&gt; [[0 0 0 0] [5 6
+--   7 8]]
+--   
+--   # Which is equivalent to: tf.segment_sum(c, tf.constant([0, 0, 1]))
+--   ```
+sparseSegmentSum :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 t -> Tensor v2 tidx -> Tensor v3 Int32 -> Tensor Value t
+
+-- | Computes the sum along sparse segments of a tensor divided by the sqrt
+--   of N.
+--   
+--   N is the size of the segment being reduced.
+--   
+--   Read <a>the section on Segmentation</a> for an explanation of
+--   segments.
+sparseSegmentSqrtN :: (TensorType t, OneOf '[Double, Float] t, TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 t -> Tensor v2 tidx -> Tensor v3 Int32 -> Tensor Value t
+
+-- | Copy Host Op.
+--   
+--   Performs CPU-to-CPU deep-copying of tensor.
+--   
+--   Unlike the Copy Op, this op has HostMemory constraint on its input or
+--   output.
+copyHost :: (TensorType t) => Tensor v1 t -> Tensor Value t
+
+-- | Holds state in the form of a tensor that persists across steps.
+--   
+--   Outputs a ref to the tensor state so it may be read or modified.
+--   TODO(zhifengc/mrry): Adds a pointer to a more detail document about
+--   sharing states in tensorflow.
+variable :: (TensorType dtype) => Tensor Value dtype
+
+-- | Computes gradients for SparseSegmentSqrtN.
+--   
+--   Returns tensor "output" with same shape as grad, except for dimension
+--   0 whose value is output_dim0.
+sparseSegmentSqrtNGrad :: (TensorType t, OneOf '[Double, Float] t, TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 t -> Tensor v2 tidx -> Tensor v3 Int32 -> Tensor v4 Int32 -> Tensor Value t
+
+-- | Creates a sequence of integers.
+--   
+--   This operation creates a sequence of integers that begins at
+--   <tt>start</tt> and extends by increments of <tt>delta</tt> up to but
+--   not including <tt>limit</tt>.
+--   
+--   For example:
+--   
+--   ``` # <tt>start</tt> is 3 # <tt>limit</tt> is 18 # <tt>delta</tt> is 3
+--   tf.range(start, limit, delta) ==&gt; [3, 6, 9, 12, 15] ```
+range :: (TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 tidx -> Tensor v2 tidx -> Tensor v3 tidx -> Tensor Value tidx
+
+-- | Computes the "logical or" of elements across dimensions of a tensor.
+--   
+--   Reduces <tt>input</tt> along the dimensions given in
+--   <tt>reduction_indices</tt>. Unless <tt>keep_dims</tt> is true, the
+--   rank of the tensor is reduced by 1 for each entry in
+--   <tt>reduction_indices</tt>. If <tt>keep_dims</tt> is true, the reduced
+--   dimensions are retained with length 1.
+any :: (TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 Bool -> Tensor v2 tidx -> Tensor Value Bool
+
+-- | Generates values in an interval.
+--   
+--   A sequence of <tt>num</tt> evenly-spaced values are generated
+--   beginning at <tt>start</tt>. If `num &gt; 1`, the values in the
+--   sequence increase by `stop - start / num - 1`, so that the last one is
+--   exactly <tt>stop</tt>.
+--   
+--   For example:
+--   
+--   ``` tf.linspace(10.0, 12.0, 3, name="linspace") =&gt; [ 10.0 11.0
+--   12.0] ```
+linSpace :: (TensorType t, OneOf '[Double, Float] t, TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 t -> Tensor v2 t -> Tensor v3 tidx -> Tensor Value t
+
+-- | Resize <tt>images</tt> to <a>size</a> using area interpolation.
+--   
+--   Input images can be of different types but output images are always
+--   float.
+resizeArea :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 Int32 -> Tensor Value Float
+
+-- | Returns the real part of a complex number.
+--   
+--   Given a tensor <tt>input</tt> of complex numbers, this operation
+--   returns a tensor of type <tt>float</tt> that is the real part of each
+--   element in <tt>input</tt>. All elements in <tt>input</tt> must be
+--   complex numbers of the form \(a + bj\), where *a* is the real part
+--   returned by this operation and *b* is the imaginary part.
+--   
+--   For example:
+--   
+--   ``` # tensor <tt>input</tt> is [-2.25 + 4.75j, 3.25 + 5.75j]
+--   tf.real(input) ==&gt; [-2.25, 3.25] ```
+real :: (TensorType t, OneOf '[Complex Double, Complex Float] t, TensorType tout, OneOf '[Double, Float] tout) => Tensor v1 t -> Tensor Value tout
+
+-- | Compute the inverse 1-dimensional discrete Fourier Transform over the
+--   inner-most
+--   
+--   dimension of <tt>input</tt>.
+iFFT :: Tensor v1 (Complex Float) -> Tensor Value (Complex Float)
+
+-- | Compute the inverse 3-dimensional discrete Fourier Transform over the
+--   inner-most
+--   
+--   3 dimensions of <tt>input</tt>.
+iFFT3D :: Tensor v1 (Complex Float) -> Tensor Value (Complex Float)
+
+-- | Compute the pairwise cross product.
+--   
+--   <tt>a</tt> and <tt>b</tt> must be the same shape; they can either be
+--   simple 3-element vectors, or any shape where the innermost dimension
+--   is 3. In the latter case, each pair of corresponding 3-element vectors
+--   is cross-multiplied independently.
+cross :: (TensorType t, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Compute the cumulative sum of the tensor <tt>x</tt> along
+--   <tt>axis</tt>.
+--   
+--   By default, this op performs an inclusive cumsum, which means that the
+--   first element of the input is identical to the first element of the
+--   output: ```prettyprint tf.cumsum([a, b, c]) ==&gt; [a, a + b, a + b +
+--   c] ```
+--   
+--   By setting the <tt>exclusive</tt> kwarg to <a>True</a>, an exclusive
+--   cumsum is performed instead: ```prettyprint tf.cumsum([a, b, c],
+--   exclusive=True) ==&gt; [0, a, a + b] ```
+--   
+--   By setting the <a>reverse</a> kwarg to <a>True</a>, the cumsum is
+--   performed in the opposite direction: ```prettyprint tf.cumsum([a, b,
+--   c], reverse=True) ==&gt; [a + b + c, b + c, c] ``` This is more
+--   efficient than using separate `tf.reverse` ops.
+--   
+--   The <a>reverse</a> and <tt>exclusive</tt> kwargs can also be combined:
+--   ```prettyprint tf.cumsum([a, b, c], exclusive=True, reverse=True)
+--   ==&gt; [b + c, c, 0] ```
+cumsum :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tidx, OneOf '[Int32, Int64] tidx) => Tensor v1 t -> Tensor v2 tidx -> Tensor Value t
+batchIFFT :: Tensor v1 (Complex Float) -> Tensor Value (Complex Float)
+
+-- | Computes the Gauss error function of <tt>x</tt> element-wise.
+erf :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | For each key, assigns the respective value to the specified component.
+--   
+--   If a key is not found in the barrier, this operation will create a new
+--   incomplete element. If a key is found in the barrier, and the element
+--   already has a value at component_index, this operation will fail with
+--   INVALID_ARGUMENT, and leave the barrier in an undefined state.
+barrierInsertMany :: (TensorType t) => Int64 -> Tensor v1 ByteString -> Tensor v2 ByteString -> Tensor v3 t -> ControlNode
+
+-- | Returns element-wise largest integer not greater than x.
+floor :: (TensorType t, OneOf '[Word16, Double, Float] t) => Tensor v1 t -> Tensor Value t
+batchFFT2D :: Tensor v1 (Complex Float) -> Tensor Value (Complex Float)
+
+-- | The gradient operator for the SparseAdd op.
+--   
+--   The SparseAdd op calculates A + B, where A, B, and the sum are all
+--   represented as <tt>SparseTensor</tt> objects. This op takes in the
+--   upstream gradient w.r.t. non-empty values of the sum, and outputs the
+--   gradients w.r.t. the non-empty values of A and B.
+sparseAddGrad :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 t -> Tensor v2 Int64 -> Tensor v3 Int64 -> Tensor v4 Int64 -> (Tensor Value t, Tensor Value t)
+
+-- | Adds two <tt>SparseTensor</tt> objects to produce another
+--   <tt>SparseTensor</tt>.
+--   
+--   The input <tt>SparseTensor</tt> objects' indices are assumed ordered
+--   in standard lexicographic order. If this is not the case, before this
+--   step run <tt>SparseReorder</tt> to restore index ordering.
+--   
+--   By default, if two values sum to zero at some index, the output
+--   <tt>SparseTensor</tt> would still include that particular location in
+--   its index, storing a zero in the corresponding value slot. To override
+--   this, callers can specify <tt>thresh</tt>, indicating that if the sum
+--   has a magnitude strictly smaller than <tt>thresh</tt>, its
+--   corresponding value and index would then not be included. In
+--   particular, `thresh == 0` (default) means everything is kept and
+--   actual thresholding happens only for a positive value.
+--   
+--   In the following shapes, <tt>nnz</tt> is the count after taking
+--   <tt>thresh</tt> into account.
+sparseAdd :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType treal, OneOf '[Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] treal) => Tensor v1 Int64 -> Tensor v2 t -> Tensor v3 Int64 -> Tensor v4 Int64 -> Tensor v5 t -> Tensor v6 Int64 -> Tensor v7 treal -> (Tensor Value Int64, Tensor Value t, Tensor Value Int64)
+batchCholesky :: (TensorType t, OneOf '[Double, Float] t) => Tensor v1 t -> Tensor Value t
+
+-- | Partitions `data` into <tt>num_partitions</tt> tensors using indices
+--   from <tt>partitions</tt>.
+--   
+--   For each index tuple <tt>js</tt> of size `partitions.ndim`, the slice
+--   `data[js, ...]` becomes part of `outputs[partitions[js]]`. The slices
+--   with `partitions[js] = i` are placed in `outputs[i]` in lexicographic
+--   order of <tt>js</tt>, and the first dimension of `outputs[i]` is the
+--   number of entries in <tt>partitions</tt> equal to <tt>i</tt>. In
+--   detail,
+--   
+--   outputs[i].shape = [sum(partitions == i)] +
+--   data.shape[partitions.ndim:]
+--   
+--   outputs[i] = pack([data[js, ...] for js if partitions[js] == i])
+--   
+--   `data.shape` must start with `partitions.shape`.
+--   
+--   For example:
+--   
+--   # Scalar partitions partitions = 1 num_partitions = 2 data = [10, 20]
+--   outputs[0] = [] # Empty with shape [0, 2] outputs[1] = [[10, 20]]
+--   
+--   # Vector partitions partitions = [0, 0, 1, 1, 0] num_partitions = 2
+--   data = [10, 20, 30, 40, 50] outputs[0] = [10, 20, 50] outputs[1] =
+--   [30, 40]
+--   
+--   <a>style="width:70%; margin:auto; margin-bottom:10px;
+--   margin-top:20px;"</a> <a>style="width:100%"
+--   src="../../images/DynamicPartition.png" alt</a> <a>/div</a>
+dynamicPartition :: (TensorType t) => Int64 -> Tensor v1 t -> Tensor v2 Int32 -> [Tensor Value t]
+
+-- | Serialize a <tt>SparseTensor</tt> into a string 3-vector (1-D
+--   <a>Tensor</a>) object.
+serializeSparse :: (TensorType t) => Tensor v1 Int64 -> Tensor v2 t -> Tensor v3 Int64 -> Tensor Value ByteString
+
+-- | Concatenates a list of <tt>SparseTensor</tt> along the specified
+--   dimension.
+--   
+--   Concatenation is with respect to the dense versions of these sparse
+--   tensors. It is assumed that each input is a <tt>SparseTensor</tt>
+--   whose elements are ordered along increasing dimension number.
+--   
+--   All inputs' shapes must match, except for the concat dimension. The
+--   <tt>indices</tt>, <tt>values</tt>, and <tt>shapes</tt> lists must have
+--   the same length.
+--   
+--   The output shape is identical to the inputs', except along the concat
+--   dimension, where it is the sum of the inputs' sizes along that
+--   dimension.
+--   
+--   The output elements will be resorted to preserve the sort order along
+--   increasing dimension number.
+--   
+--   This op runs in `O(M log M)` time, where <tt>M</tt> is the total
+--   number of non-empty values across all inputs. This is due to the need
+--   for an internal sort in order to concatenate efficiently across an
+--   arbitrary dimension.
+--   
+--   For example, if `concat_dim = 1` and the inputs are
+--   
+--   sp_inputs[0]: shape = [2, 3] [0, 2]: "a" [1, 0]: "b" [1, 1]: "c"
+--   
+--   sp_inputs[1]: shape = [2, 4] [0, 1]: "d" [0, 2]: "e"
+--   
+--   then the output will be
+--   
+--   shape = [2, 7] [0, 2]: "a" [0, 4]: "d" [0, 5]: "e" [1, 0]: "b" [1, 1]:
+--   "c"
+--   
+--   Graphically this is equivalent to doing
+--   
+--   <ul>
+--   <li><i> a</i> concat [ d e ] = [ a d e ]</li>
+--   <li><i>b c </i> [ ] [b c ]</li>
+--   </ul>
+sparseConcat :: (TensorType t) => Int64 -> [Tensor v1 Int64] -> [Tensor v2 t] -> [Tensor v3 Int64] -> (Tensor Value Int64, Tensor Value t, Tensor Value Int64)
+
+-- | Computes the product along segments of a tensor.
+--   
+--   Read <a>the section on Segmentation</a> for an explanation of
+--   segments.
+--   
+--   Computes a tensor such that \(output_i = prod_j data_j\) where the
+--   product is over <tt>j</tt> such that `segment_ids[j] == i`.
+--   
+--   <a>style="width:70%; margin:auto; margin-bottom:10px;
+--   margin-top:20px;"</a> <a>style="width:100%"
+--   src="../../images/SegmentProd.png" alt</a> <a>/div</a>
+segmentProd :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t, TensorType tindices, OneOf '[Int32, Int64] tindices) => Tensor v1 t -> Tensor v2 tindices -> Tensor Value t
+
+-- | Reshapes a SparseTensor to represent values in a new dense shape.
+--   
+--   This operation has the same semantics as reshape on the represented
+--   dense tensor. The <tt>input_indices</tt> are recomputed based on the
+--   requested <tt>new_shape</tt>.
+--   
+--   If one component of <tt>new_shape</tt> is the special value -1, the
+--   size of that dimension is computed so that the total dense size
+--   remains constant. At most one component of <tt>new_shape</tt> can be
+--   -1. The number of dense elements implied by <tt>new_shape</tt> must be
+--   the same as the number of dense elements originally implied by
+--   <tt>input_shape</tt>.
+--   
+--   Reshaping does not affect the order of values in the SparseTensor.
+--   
+--   If the input tensor has rank <tt>R_in</tt> and <tt>N</tt> non-empty
+--   values, and <tt>new_shape</tt> has length <tt>R_out</tt>, then
+--   <tt>input_indices</tt> has shape `[N, R_in]`, <tt>input_shape</tt> has
+--   length <tt>R_in</tt>, <tt>output_indices</tt> has shape `[N, R_out]`,
+--   and <tt>output_shape</tt> has length <tt>R_out</tt>.
+sparseReshape :: Tensor v1 Int64 -> Tensor v2 Int64 -> Tensor v3 Int64 -> (Tensor Value Int64, Tensor Value Int64)
+
+-- | Component-wise multiplies a SparseTensor by a dense Tensor.
+--   
+--   The output locations corresponding to the implicitly zero elements in
+--   the sparse tensor will be zero (i.e., will not take up storage space),
+--   regardless of the contents of the dense tensor (even if it's +/-INF
+--   and that INF*0 == NaN).
+--   
+--   <ul>
+--   <li>Limitation*: this Op only broadcasts the dense side to the sparse
+--   side, but not the other direction.</li>
+--   </ul>
+sparseDenseCwiseMul :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 Int64 -> Tensor v2 t -> Tensor v3 Int64 -> Tensor v4 t -> Tensor Value t
+
+-- | Component-wise divides a SparseTensor by a dense Tensor.
+--   
+--   <ul>
+--   <li>Limitation*: this Op only broadcasts the dense side to the sparse
+--   side, but not the other direction.</li>
+--   </ul>
+sparseDenseCwiseDiv :: (TensorType t, OneOf '[Complex Double, Complex Float, Int16, Int32, Int64, Int8, Word16, Word8, Double, Float] t) => Tensor v1 Int64 -> Tensor v2 t -> Tensor v3 Int64 -> Tensor v4 t -> Tensor Value t
diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html b/docs/haddock/tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html
new file mode 100644
index 0000000..8788b6e
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-Parse.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Examples.MNIST.Parse</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-Examples-MNIST-Parse.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-Examples-MNIST-Parse.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-mnist-0.1.0.0: TensorFlow demo application for learning MNIST model.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.Examples.MNIST.Parse</p></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><span class="keyword">type</span> <a href="#t:MNIST">MNIST</a> = <a href="../vector-0.11.0.0/Data-Vector.html#t:Vector">Vector</a> <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a></li><li class="src short"><a href="#v:drawMNIST">drawMNIST</a> :: <a href="TensorFlow-Examples-MNIST-Parse.html#t:MNIST">MNIST</a> -&gt; <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></li><li class="src short"><a href="#v:checkEndian">checkEndian</a> :: <a href="../binary-0.7.5.0/Data-Binary-Get-Internal.html#t:Get">Get</a> ()</li><li class="src short"><a href="#v:readMNISTSamples">readMNISTSamples</a> :: <a href="../base-4.8.2.0/System-IO.html#t:FilePath">FilePath</a> -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> [<a href="TensorFlow-Examples-MNIST-Parse.html#t:MNIST">MNIST</a>]</li><li class="src short"><a href="#v:readMNISTLabels">readMNISTLabels</a> :: <a href="../base-4.8.2.0/System-IO.html#t:FilePath">FilePath</a> -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> [<a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>]</li><li class="src short"><a href="#v:readMessageFromFileOrDie">readMessageFromFileOrDie</a> :: Message m =&gt; <a href="../base-4.8.2.0/System-IO.html#t:FilePath">FilePath</a> -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> m</li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">type</span> <a name="t:MNIST" class="def">MNIST</a> = <a href="../vector-0.11.0.0/Data-Vector.html#t:Vector">Vector</a> <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> <a href="src/TensorFlow-Examples-MNIST-Parse.html#MNIST" class="link">Source</a></p><div class="doc"><p>Utilities specific to MNIST.</p></div></div><div class="top"><p class="src"><a name="v:drawMNIST" class="def">drawMNIST</a> :: <a href="TensorFlow-Examples-MNIST-Parse.html#t:MNIST">MNIST</a> -&gt; <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> <a href="src/TensorFlow-Examples-MNIST-Parse.html#drawMNIST" class="link">Source</a></p><div class="doc"><p>Produces a unicode rendering of the MNIST digit sample.</p></div></div><div class="top"><p class="src"><a name="v:checkEndian" class="def">checkEndian</a> :: <a href="../binary-0.7.5.0/Data-Binary-Get-Internal.html#t:Get">Get</a> () <a href="src/TensorFlow-Examples-MNIST-Parse.html#checkEndian" class="link">Source</a></p><div class="doc"><p>Check's the file's endianess, throwing an error if it's not as expected.</p></div></div><div class="top"><p class="src"><a name="v:readMNISTSamples" class="def">readMNISTSamples</a> :: <a href="../base-4.8.2.0/System-IO.html#t:FilePath">FilePath</a> -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> [<a href="TensorFlow-Examples-MNIST-Parse.html#t:MNIST">MNIST</a>] <a href="src/TensorFlow-Examples-MNIST-Parse.html#readMNISTSamples" class="link">Source</a></p><div class="doc"><p>Reads an MNIST file and returns a list of samples.</p></div></div><div class="top"><p class="src"><a name="v:readMNISTLabels" class="def">readMNISTLabels</a> :: <a href="../base-4.8.2.0/System-IO.html#t:FilePath">FilePath</a> -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> [<a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a>] <a href="src/TensorFlow-Examples-MNIST-Parse.html#readMNISTLabels" class="link">Source</a></p><div class="doc"><p>Reads a list of MNIST labels from a file and returns them.</p></div></div><div class="top"><p class="src"><a name="v:readMessageFromFileOrDie" class="def">readMessageFromFileOrDie</a> :: Message m =&gt; <a href="../base-4.8.2.0/System-IO.html#t:FilePath">FilePath</a> -&gt; <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> m <a href="src/TensorFlow-Examples-MNIST-Parse.html#readMessageFromFileOrDie" class="link">Source</a></p></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-TrainedGraph.html b/docs/haddock/tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-TrainedGraph.html
new file mode 100644
index 0000000..630e30a
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-0.1.0.0/TensorFlow-Examples-MNIST-TrainedGraph.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Examples.MNIST.TrainedGraph</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-Examples-MNIST-TrainedGraph.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-Examples-MNIST-TrainedGraph.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-mnist-0.1.0.0: TensorFlow demo application for learning MNIST model.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>Safe</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.Examples.MNIST.TrainedGraph</p></div><div id="description"><p class="caption">Description</p><div class="doc"><p>Paths to test helper files.</p></div></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><a href="#v:mnistPb">mnistPb</a> :: <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> <a href="../base-4.8.2.0/System-IO.html#t:FilePath">FilePath</a></li><li class="src short"><a href="#v:wtsCkpt">wtsCkpt</a> :: <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li><li class="src short"><a href="#v:biasCkpt">biasCkpt</a> :: <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><a name="v:mnistPb" class="def">mnistPb</a> :: <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> <a href="../base-4.8.2.0/System-IO.html#t:FilePath">FilePath</a> <a href="src/TensorFlow-Examples-MNIST-TrainedGraph.html#mnistPb" class="link">Source</a></p><div class="doc"><p>File containing a Tensorflow serialized proto of MNIST.</p></div></div><div class="top"><p class="src"><a name="v:wtsCkpt" class="def">wtsCkpt</a> :: <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> <a href="src/TensorFlow-Examples-MNIST-TrainedGraph.html#wtsCkpt" class="link">Source</a></p><div class="doc"><p>Files containing pre-trained weights for MNIST.</p></div></div><div class="top"><p class="src"><a name="v:biasCkpt" class="def">biasCkpt</a> :: <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> <a href="src/TensorFlow-Examples-MNIST-TrainedGraph.html#biasCkpt" class="link">Source</a></p><div class="doc"><p>Files containing pre-trained weights for MNIST.</p></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/doc-index.html b/docs/haddock/tensorflow-mnist-0.1.0.0/doc-index.html
new file mode 100644
index 0000000..5f705d0
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-0.1.0.0/doc-index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-mnist-0.1.0.0: TensorFlow demo application for learning MNIST model. (Index)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-mnist-0.1.0.0: TensorFlow demo application for learning MNIST model.</p></div><div id="content"><div id="index"><p class="caption">Index</p><table><tr><td class="src">biasCkpt</td><td class="module"><a href="TensorFlow-Examples-MNIST-TrainedGraph.html#v:biasCkpt">TensorFlow.Examples.MNIST.TrainedGraph</a></td></tr><tr><td class="src">checkEndian</td><td class="module"><a href="TensorFlow-Examples-MNIST-Parse.html#v:checkEndian">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">drawMNIST</td><td class="module"><a href="TensorFlow-Examples-MNIST-Parse.html#v:drawMNIST">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">MNIST</td><td class="module"><a href="TensorFlow-Examples-MNIST-Parse.html#t:MNIST">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">mnistPb</td><td class="module"><a href="TensorFlow-Examples-MNIST-TrainedGraph.html#v:mnistPb">TensorFlow.Examples.MNIST.TrainedGraph</a></td></tr><tr><td class="src">readMessageFromFileOrDie</td><td class="module"><a href="TensorFlow-Examples-MNIST-Parse.html#v:readMessageFromFileOrDie">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">readMNISTLabels</td><td class="module"><a href="TensorFlow-Examples-MNIST-Parse.html#v:readMNISTLabels">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">readMNISTSamples</td><td class="module"><a href="TensorFlow-Examples-MNIST-Parse.html#v:readMNISTSamples">TensorFlow.Examples.MNIST.Parse</a></td></tr><tr><td class="src">wtsCkpt</td><td class="module"><a href="TensorFlow-Examples-MNIST-TrainedGraph.html#v:wtsCkpt">TensorFlow.Examples.MNIST.TrainedGraph</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/frames.html b/docs/haddock/tensorflow-mnist-0.1.0.0/frames.html
new file mode 100644
index 0000000..1b4e38d
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-0.1.0.0/frames.html
@@ -0,0 +1,30 @@
+<!DOCTYPE html 
+     PUBLIC "-//W3C//DTD XHTML 1.0 Frameset//EN"
+     "http://www.w3.org/TR/xhtml1/DTD/xhtml1-frameset.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<title></title>
+<script src="haddock-util.js" type="text/javascript"></script>
+<script type="text/javascript"><!--
+/*
+
+  The synopsis frame needs to be updated using javascript, so we hide
+  it by default and only show it if javascript is enabled.
+
+  TODO: provide some means to disable it.
+*/
+function load() {
+  var d = document.getElementById("inner-fs");
+  d.rows = "50%,50%";
+  postReframe();
+}
+--></script>
+</head>
+<frameset id="outer-fs" cols="25%,75%" onload="load()">
+  <frameset id="inner-fs" rows="100%,0%">
+    <frame src="index-frames.html" name="modules" />
+    <frame src="" name="synopsis" />
+  </frameset>
+  <frame src="index.html" name="main" />
+</frameset>
+</html>
diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/haddock-util.js b/docs/haddock/tensorflow-mnist-0.1.0.0/haddock-util.js
new file mode 100644
index 0000000..9a6fccf
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-0.1.0.0/haddock-util.js
@@ -0,0 +1,344 @@
+// Haddock JavaScript utilities
+
+var rspace = /\s\s+/g,
+	  rtrim = /^\s+|\s+$/g;
+
+function spaced(s) { return (" " + s + " ").replace(rspace, " "); }
+function trim(s)   { return s.replace(rtrim, ""); }
+
+function hasClass(elem, value) {
+  var className = spaced(elem.className || "");
+  return className.indexOf( " " + value + " " ) >= 0;
+}
+
+function addClass(elem, value) {
+  var className = spaced(elem.className || "");
+  if ( className.indexOf( " " + value + " " ) < 0 ) {
+    elem.className = trim(className + " " + value);
+  }
+}
+
+function removeClass(elem, value) {
+  var className = spaced(elem.className || "");
+  className = className.replace(" " + value + " ", " ");
+  elem.className = trim(className);
+}
+
+function toggleClass(elem, valueOn, valueOff, bool) {
+  if (bool == null) { bool = ! hasClass(elem, valueOn); }
+  if (bool) {
+    removeClass(elem, valueOff);
+    addClass(elem, valueOn);
+  }
+  else {
+    removeClass(elem, valueOn);
+    addClass(elem, valueOff);
+  }
+  return bool;
+}
+
+
+function makeClassToggle(valueOn, valueOff)
+{
+  return function(elem, bool) {
+    return toggleClass(elem, valueOn, valueOff, bool);
+  }
+}
+
+toggleShow = makeClassToggle("show", "hide");
+toggleCollapser = makeClassToggle("collapser", "expander");
+
+function toggleSection(id)
+{
+  var b = toggleShow(document.getElementById("section." + id));
+  toggleCollapser(document.getElementById("control." + id), b);
+  rememberCollapsed(id, b);
+  return b;
+}
+
+var collapsed = {};
+function rememberCollapsed(id, b)
+{
+  if(b)
+    delete collapsed[id]
+  else
+    collapsed[id] = null;
+
+  var sections = [];
+  for(var i in collapsed)
+  {
+    if(collapsed.hasOwnProperty(i))
+      sections.push(i);
+  }
+  // cookie specific to this page; don't use setCookie which sets path=/
+  document.cookie = "collapsed=" + escape(sections.join('+'));
+}
+
+function restoreCollapsed()
+{
+  var cookie = getCookie("collapsed");
+  if(!cookie)
+    return;
+
+  var ids = cookie.split('+');
+  for(var i in ids)
+  {
+    if(document.getElementById("section." + ids[i]))
+      toggleSection(ids[i]);
+  }
+}
+
+function setCookie(name, value) {
+  document.cookie = name + "=" + escape(value) + ";path=/;";
+}
+
+function clearCookie(name) {
+  document.cookie = name + "=;path=/;expires=Thu, 01-Jan-1970 00:00:01 GMT;";
+}
+
+function getCookie(name) {
+  var nameEQ = name + "=";
+  var ca = document.cookie.split(';');
+  for(var i=0;i < ca.length;i++) {
+    var c = ca[i];
+    while (c.charAt(0)==' ') c = c.substring(1,c.length);
+    if (c.indexOf(nameEQ) == 0) {
+      return unescape(c.substring(nameEQ.length,c.length));
+    }
+  }
+  return null;
+}
+
+
+
+var max_results = 75; // 50 is not enough to search for map in the base libraries
+var shown_range = null;
+var last_search = null;
+
+function quick_search()
+{
+    perform_search(false);
+}
+
+function full_search()
+{
+    perform_search(true);
+}
+
+
+function perform_search(full)
+{
+    var text = document.getElementById("searchbox").value.toLowerCase();
+    if (text == last_search && !full) return;
+    last_search = text;
+    
+    var table = document.getElementById("indexlist");
+    var status = document.getElementById("searchmsg");
+    var children = table.firstChild.childNodes;
+    
+    // first figure out the first node with the prefix
+    var first = bisect(-1);
+    var last = (first == -1 ? -1 : bisect(1));
+
+    if (first == -1)
+    {
+        table.className = "";
+        status.innerHTML = "No results found, displaying all";
+    }
+    else if (first == 0 && last == children.length - 1)
+    {
+        table.className = "";
+        status.innerHTML = "";
+    }
+    else if (last - first >= max_results && !full)
+    {
+        table.className = "";
+        status.innerHTML = "More than " + max_results + ", press Search to display";
+    }
+    else
+    {
+        // decide what you need to clear/show
+        if (shown_range)
+            setclass(shown_range[0], shown_range[1], "indexrow");
+        setclass(first, last, "indexshow");
+        shown_range = [first, last];
+        table.className = "indexsearch";
+        status.innerHTML = "";
+    }
+
+    
+    function setclass(first, last, status)
+    {
+        for (var i = first; i <= last; i++)
+        {
+            children[i].className = status;
+        }
+    }
+    
+    
+    // do a binary search, treating 0 as ...
+    // return either -1 (no 0's found) or location of most far match
+    function bisect(dir)
+    {
+        var first = 0, finish = children.length - 1;
+        var mid, success = false;
+
+        while (finish - first > 3)
+        {
+            mid = Math.floor((finish + first) / 2);
+
+            var i = checkitem(mid);
+            if (i == 0) i = dir;
+            if (i == -1)
+                finish = mid;
+            else
+                first = mid;
+        }
+        var a = (dir == 1 ? first : finish);
+        var b = (dir == 1 ? finish : first);
+        for (var i = b; i != a - dir; i -= dir)
+        {
+            if (checkitem(i) == 0) return i;
+        }
+        return -1;
+    }    
+    
+    
+    // from an index, decide what the result is
+    // 0 = match, -1 is lower, 1 is higher
+    function checkitem(i)
+    {
+        var s = getitem(i).toLowerCase().substr(0, text.length);
+        if (s == text) return 0;
+        else return (s > text ? -1 : 1);
+    }
+    
+    
+    // from an index, get its string
+    // this abstracts over alternates
+    function getitem(i)
+    {
+        for ( ; i >= 0; i--)
+        {
+            var s = children[i].firstChild.firstChild.data;
+            if (s.indexOf(' ') == -1)
+                return s;
+        }
+        return ""; // should never be reached
+    }
+}
+
+function setSynopsis(filename) {
+    if (parent.window.synopsis) {
+        if (parent.window.synopsis.location.replace) {
+            // In Firefox this avoids adding the change to the history.
+            parent.window.synopsis.location.replace(filename);
+        } else {
+            parent.window.synopsis.location = filename;
+        }
+    }
+}
+
+function addMenuItem(html) {
+  var menu = document.getElementById("page-menu");
+  if (menu) {
+    var btn = menu.firstChild.cloneNode(false);
+    btn.innerHTML = html;
+    menu.appendChild(btn);
+  }
+}
+
+function adjustForFrames() {
+  var bodyCls;
+  
+  if (parent.location.href == window.location.href) {
+    // not in frames, so add Frames button
+    addMenuItem("<a href='#' onclick='reframe();return true;'>Frames</a>");
+    bodyCls = "no-frame";
+  }
+  else {
+    bodyCls = "in-frame";
+  }
+  addClass(document.body, bodyCls);
+}
+
+function reframe() {
+  setCookie("haddock-reframe", document.URL);
+  window.location = "frames.html";
+}
+
+function postReframe() {
+  var s = getCookie("haddock-reframe");
+  if (s) {
+    parent.window.main.location = s;
+    clearCookie("haddock-reframe");
+  }
+}
+
+function styles() {
+  var i, a, es = document.getElementsByTagName("link"), rs = [];
+  for (i = 0; a = es[i]; i++) {
+    if(a.rel.indexOf("style") != -1 && a.title) {
+      rs.push(a);
+    }
+  }
+  return rs;
+}
+
+function addStyleMenu() {
+  var as = styles();
+  var i, a, btns = "";
+  for(i=0; a = as[i]; i++) {
+    btns += "<li><a href='#' onclick=\"setActiveStyleSheet('"
+      + a.title + "'); return false;\">"
+      + a.title + "</a></li>"
+  }
+  if (as.length > 1) {
+    var h = "<div id='style-menu-holder'>"
+      + "<a href='#' onclick='styleMenu(); return false;'>Style &#9662;</a>"
+      + "<ul id='style-menu' class='hide'>" + btns + "</ul>"
+      + "</div>";
+    addMenuItem(h);
+  }
+}
+
+function setActiveStyleSheet(title) {
+  var as = styles();
+  var i, a, found;
+  for(i=0; a = as[i]; i++) {
+    a.disabled = true;
+          // need to do this always, some browsers are edge triggered
+    if(a.title == title) {
+      found = a;
+    }
+  }
+  if (found) {
+    found.disabled = false;
+    setCookie("haddock-style", title);
+  }
+  else {
+    as[0].disabled = false;
+    clearCookie("haddock-style");
+  }
+  styleMenu(false);
+}
+
+function resetStyle() {
+  var s = getCookie("haddock-style");
+  if (s) setActiveStyleSheet(s);
+}
+
+
+function styleMenu(show) {
+  var m = document.getElementById('style-menu');
+  if (m) toggleShow(m, show);
+}
+
+
+function pageLoad() {
+  addStyleMenu();
+  adjustForFrames();
+  resetStyle();
+  restoreCollapsed();
+}
+
diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/hslogo-16.png b/docs/haddock/tensorflow-mnist-0.1.0.0/hslogo-16.png
new file mode 100644
index 0000000000000000000000000000000000000000..0ff8579fbd897417b0d6dad6e920f8882138a7c0
GIT binary patch
literal 1684
zcmV;F25b3=P)<h;3K|Lk000e1NJLTq000*N000mO1^@s6{V(Jg0008+X+uL$Nkc;*
zP;zf(X>4Tx0C)j~RL^S@K@|QrZmG~B2wH0nvUrdpNm;9CMbtL^5n^i$+aI<?kYqDS
z)^0ZI?k2URh~(nYD&9ryp$Gqf9(wR(FG_`4MJV2@C}N=(E%D86Otz)!z|Ma2-Z$@k
zZ+71R4RX>n^?(HA4aZWV5ov6ELTdbo0FI&wK{O>*+w4vx20?>!`FrQsdJlnHR>OPy
zcd~b_n$otK2Za4V;76L-DzNVtaSB-y0*E}{p()372;bw_^6ZZ}PI-92wGS&j#91PI
zKs7DSe@(bk%_Y-7gGe}(^>I=@oY#w#*Bu9GZf3^F5WP>3rn}7Ut74&?PWBFvy`A)a
zPP5)V!Xd&78LdA?xQ(9mjMYElVd13a#D+Z_7&Y|xU=_C-srWU*6kiZcC!$nw*)9$7
zn6CX+@=AhmkT}X@VSsa5NKe;HZuq)~1$`#h6R+ZTR#D-3j}vF!)ZOnz+5)dI4jl{{
z44Mr{P!L4~VVJN`K!!XTF*LGrKO?IK8<Tr7btG!LbYeuYL3=jbJ-1P$-8}v%B5{;M
zwFr{@LH;VQ$xr2Z`O93e*jD$Ht(%&<^58qg<(at}9@o>z<8w`3e3jI8lUGNUta*C8
zn(P`s>{pjD=7Kek#B;Fw@hxAK%$F&Q6vg9J^Xf~4by_hu-=A!MJ3Znq&n~srbFGPs
zH&&aMXZ>nO`|hf|ljc?VPhR!${AbO?W8x_>CU%PFA&Hm8F7cAsOREdwU~R_;ot1_u
z(ruCYB-LPGn!NQdT|ZlRy+(fw^-+`=%+gee_kY4FWHg<*4sZI8+sFJD<oAl_pC|$^
zY~aY5x@}W&?+~G7rEYVs0vEs0eekw!YomTR`~+A$s}`+NHJ>270UUORdLHO0nA4V)
z%{fwsET5CQ>B?eK%uw4yQc~9?*JVo<vzb+5>2}ze(;aRcp*ceL#HUJSllrgm5wQKR
zQu+C;QrUh^8rFfA`ftFz{YAidi-`aL010qNS#tmY4c7nw4c7reD4Tcy00T@(L_t(I
z5sj2vNEA^R$7gqDc6T=2^@fUA2(c`MltuL5<|<Z1YDGjBEr^N`3VJpQ$`C4&9(*Z6
z=|*B$y;-1^F6vXRAd7+(;)77Qps2euv)AeWmj>KW>RWz$&YbU@|M|{$E*8Tu-Ux!w
z1Y*Dr&Ubfr&v-nZaaB{3ilRumrjPmk{sZvQEWlW+{o~IH|8)=s6c#X9S5s5d%<Agu
zgM$153Ttk++ZPA~ZqgqN1|^pNV-|?9-zo48n9+Yiy$*su5czgp*VFj6!1h&DRZR>J
z4@)&QH5|xQY-)^L1n<EM!01(Q6kM8}ojndgRD~2KVIa@>0pTRu0Lx9`08YTjTwn^6
z0;b1+aQ@)n;Em$q;=7BBi)v0zj&o^g>0Whp^_^5IbxIUP8C@y9;R?*Ouu}rmfxbU=
zwtWVNke-m!=`7bYEhWpcI5#)9qp`8E0lr6IQ)ARL3Ui}Af@grj8aN1=r>Cb+prlzO
zNfJs*N_tUm2ZL<f%*@E?H`Ak_J`Q#Qri<;K{{H@Bg@uKWS5{WE+}vCx91geD*4DNm
zJ?D^~dGz7lWtbWH`o<0O#|{R6h9M*GMpjnV?Y_RgatOb~1le#ikPKTNIn%<;=>%5*
zPmL2??da$TR904gL(VDAQ<BE%k@5ZA>-Fv_Dk}Pdw*4T(%*f4MKLRg=4ekMjhe2mW
zMFsBwg%ftWT}0kxRaIk1k7qJ8*#cKB;Ft{i`zVIs-Nqge;!!Ld7#O&Qqu7e0sJmP)
z$MW*>L$vSB&dxp@iA3U9fo)-7!Czlr{|o7<LYh`Fd@X2-b0fIa*Wdy$({Zt@t7`&L
zhoaGF!l5{1P1Qm;<GaSj#=Oa7atf1JZ8n=F9*@uC?b(X48O9kPoH~h;!o2=WM?ZLl
z$Lj;4K~)fAdW@^N%nzT>Hv{1oyg3xsu%gn@(b1>$;SM-ZaQ`HV=V0s;lr%d8bd;xY
zGw<i+<$17ay36gdrqk(Lxc1|{y}dSeM5OIV(FjhBg5#OM<!Z*8(@<7crlVw+B-8R1
zl{3J6w9fYQ^vojauaRgZ=~SE&CK!R3pCe!v3>Nvm3=Iu=tyXIgtJnf@A(2S@M140N
ew{UA~tMxaJq;$xaSSi*30000<MNUMnLSTa0p)&^n

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/index-frames.html b/docs/haddock/tensorflow-mnist-0.1.0.0/index-frames.html
new file mode 100644
index 0000000..3463791
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-0.1.0.0/index-frames.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-mnist-0.1.0.0: TensorFlow demo application for learning MNIST model.</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-list"><p class="caption">Modules</p><ul><li class="module"><a href="TensorFlow-Examples-MNIST-Parse.html" target="main">TensorFlow.Examples.MNIST.Parse</a></li><li class="module"><a href="TensorFlow-Examples-MNIST-TrainedGraph.html" target="main">TensorFlow.Examples.MNIST.TrainedGraph</a></li></ul></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/index.html b/docs/haddock/tensorflow-mnist-0.1.0.0/index.html
new file mode 100644
index 0000000..3ccb460
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-0.1.0.0/index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-mnist-0.1.0.0: TensorFlow demo application for learning MNIST model.</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-mnist-0.1.0.0: TensorFlow demo application for learning MNIST model.</p></div><div id="content"><div id="description"><h1>tensorflow-mnist-0.1.0.0: TensorFlow demo application for learning MNIST model.</h1><div class="doc"><p>Please see README.md</p></div></div><div id="module-list"><p class="caption">Modules</p><ul><li><span id="control.n.1" class="module collapser" onclick="toggleSection('n.1')">TensorFlow</span><ul id="section.n.1" class="show"><li><span id="control.n.1.1" class="module collapser" onclick="toggleSection('n.1.1')">Examples</span><ul id="section.n.1.1" class="show"><li><span id="control.n.1.1.1" class="module collapser" onclick="toggleSection('n.1.1.1')">MNIST</span><ul id="section.n.1.1.1" class="show"><li><span class="module"><a href="TensorFlow-Examples-MNIST-Parse.html">TensorFlow.Examples.MNIST.Parse</a></span></li><li><span class="module"><a href="TensorFlow-Examples-MNIST-TrainedGraph.html">TensorFlow.Examples.MNIST.TrainedGraph</a></span></li></ul></li></ul></li></ul></li></ul></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/mini_TensorFlow-Examples-MNIST-Parse.html b/docs/haddock/tensorflow-mnist-0.1.0.0/mini_TensorFlow-Examples-MNIST-Parse.html
new file mode 100644
index 0000000..720a2bf
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-0.1.0.0/mini_TensorFlow-Examples-MNIST-Parse.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Examples.MNIST.Parse</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.Examples.MNIST.Parse</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">type</span> <a href="TensorFlow-Examples-MNIST-Parse.html#t:MNIST" target="main">MNIST</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Examples-MNIST-Parse.html#v:drawMNIST" target="main">drawMNIST</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Examples-MNIST-Parse.html#v:checkEndian" target="main">checkEndian</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Examples-MNIST-Parse.html#v:readMNISTSamples" target="main">readMNISTSamples</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Examples-MNIST-Parse.html#v:readMNISTLabels" target="main">readMNISTLabels</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Examples-MNIST-Parse.html#v:readMessageFromFileOrDie" target="main">readMessageFromFileOrDie</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/mini_TensorFlow-Examples-MNIST-TrainedGraph.html b/docs/haddock/tensorflow-mnist-0.1.0.0/mini_TensorFlow-Examples-MNIST-TrainedGraph.html
new file mode 100644
index 0000000..f4f119a
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-0.1.0.0/mini_TensorFlow-Examples-MNIST-TrainedGraph.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Examples.MNIST.TrainedGraph</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.Examples.MNIST.TrainedGraph</p></div><div id="interface"><div class="top"><p class="src"><a href="TensorFlow-Examples-MNIST-TrainedGraph.html#v:mnistPb" target="main">mnistPb</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Examples-MNIST-TrainedGraph.html#v:wtsCkpt" target="main">wtsCkpt</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Examples-MNIST-TrainedGraph.html#v:biasCkpt" target="main">biasCkpt</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/minus.gif b/docs/haddock/tensorflow-mnist-0.1.0.0/minus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..1deac2fe1a42e35b994f1b855488f392c50f6a89
GIT binary patch
literal 56
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS53HJ0a;6C;`j@u!^I(6E|
Jh4LZ{)&P#)4$%Mr

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/ocean.css b/docs/haddock/tensorflow-mnist-0.1.0.0/ocean.css
new file mode 100644
index 0000000..1110b40
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-0.1.0.0/ocean.css
@@ -0,0 +1,600 @@
+/* @group Fundamentals */
+
+* { margin: 0; padding: 0 }
+
+/* Is this portable? */
+html {
+  background-color: white;
+  width: 100%;
+  height: 100%;
+}
+
+body {
+  background: white;
+  color: black;
+  text-align: left;
+  min-height: 100%;
+  position: relative;
+}
+
+p {
+  margin: 0.8em 0;
+}
+
+ul, ol {
+  margin: 0.8em 0 0.8em 2em;
+}
+
+dl {
+  margin: 0.8em 0;
+}
+
+dt {
+  font-weight: bold;
+}
+dd {
+  margin-left: 2em;
+}
+
+a { text-decoration: none; }
+a[href]:link { color: rgb(196,69,29); }
+a[href]:visited { color: rgb(171,105,84); }
+a[href]:hover { text-decoration:underline; }
+
+/* @end */
+
+/* @group Fonts & Sizes */
+
+/* Basic technique & IE workarounds from YUI 3
+   For reasons, see:
+      http://yui.yahooapis.com/3.1.1/build/cssfonts/fonts.css
+ */
+
+body {
+	font:13px/1.4 sans-serif;
+	*font-size:small; /* for IE */
+	*font:x-small; /* for IE in quirks mode */
+}
+
+h1 { font-size: 146.5%; /* 19pt */ }
+h2 { font-size: 131%;   /* 17pt */ }
+h3 { font-size: 116%;   /* 15pt */ }
+h4 { font-size: 100%;   /* 13pt */ }
+h5 { font-size: 100%;   /* 13pt */ }
+
+select, input, button, textarea {
+	font:99% sans-serif;
+}
+
+table {
+	font-size:inherit;
+	font:100%;
+}
+
+pre, code, kbd, samp, tt, .src {
+	font-family:monospace;
+	*font-size:108%;
+	line-height: 124%;
+}
+
+.links, .link {
+  font-size: 85%; /* 11pt */
+}
+
+#module-header .caption {
+  font-size: 182%; /* 24pt */
+}
+
+.info  {
+  font-size: 85%; /* 11pt */
+}
+
+#table-of-contents, #synopsis  {
+  /* font-size: 85%; /* 11pt */
+}
+
+
+/* @end */
+
+/* @group Common */
+
+.caption, h1, h2, h3, h4, h5, h6 {
+  font-weight: bold;
+  color: rgb(78,98,114);
+  margin: 0.8em 0 0.4em;
+}
+
+* + h1, * + h2, * + h3, * + h4, * + h5, * + h6 {
+  margin-top: 2em;
+}
+
+h1 + h2, h2 + h3, h3 + h4, h4 + h5, h5 + h6 {
+  margin-top: inherit;
+}
+
+ul.links {
+  list-style: none;
+  text-align: left;
+  float: right;
+  display: inline-table;
+  margin: 0 0 0 1em;
+}
+
+ul.links li {
+  display: inline;
+  border-left: 1px solid #d5d5d5;
+  white-space: nowrap;
+  padding: 0;
+}
+
+ul.links li a {
+  padding: 0.2em 0.5em;
+}
+
+.hide { display: none; }
+.show { display: inherit; }
+.clear { clear: both; }
+
+.collapser {
+  background-image: url(minus.gif);
+  background-repeat: no-repeat;
+}
+.expander {
+  background-image: url(plus.gif);
+  background-repeat: no-repeat;
+}
+p.caption.collapser,
+p.caption.expander {
+  background-position: 0 0.4em;
+}
+.collapser, .expander {
+  padding-left: 14px;
+  margin-left: -14px;
+  cursor: pointer;
+}
+
+pre {
+  padding: 0.25em;
+  margin: 0.8em 0;
+  background: rgb(229,237,244);
+  overflow: auto;
+  border-bottom: 0.25em solid white;
+  /* white border adds some space below the box to compensate
+     for visual extra space that paragraphs have between baseline
+     and the bounding box */
+}
+
+.src {
+  background: #f0f0f0;
+  padding: 0.2em 0.5em;
+}
+
+.keyword { font-weight: normal; }
+.def { font-weight: bold; }
+
+
+/* @end */
+
+/* @group Page Structure */
+
+#content {
+  margin: 0 auto;
+  padding: 0 2em 6em;
+}
+
+#package-header {
+  background: rgb(41,56,69);
+  border-top: 5px solid rgb(78,98,114);
+  color: #ddd;
+  padding: 0.2em;
+  position: relative;
+  text-align: left;
+}
+
+#package-header .caption {
+  background: url(hslogo-16.png) no-repeat 0em;
+  color: white;
+  margin: 0 2em;
+  font-weight: normal;
+  font-style: normal;
+  padding-left: 2em;
+}
+
+#package-header a:link, #package-header a:visited { color: white; }
+#package-header a:hover { background: rgb(78,98,114); }
+
+#module-header .caption {
+  color: rgb(78,98,114);
+  font-weight: bold;
+  border-bottom: 1px solid #ddd;
+}
+
+table.info {
+  float: right;
+  padding: 0.5em 1em;
+  border: 1px solid #ddd;
+  color: rgb(78,98,114);
+  background-color: #fff;
+  max-width: 40%;
+  border-spacing: 0;
+  position: relative;
+  top: -0.5em;
+  margin: 0 0 0 2em;
+}
+
+.info th {
+	padding: 0 1em 0 0;
+}
+
+div#style-menu-holder {
+  position: relative;
+  z-index: 2;
+  display: inline;
+}
+
+#style-menu {
+  position: absolute;
+  z-index: 1;
+  overflow: visible;
+  background: #374c5e;
+  margin: 0;
+  text-align: center;
+  right: 0;
+  padding: 0;
+  top: 1.25em;
+}
+
+#style-menu li {
+	display: list-item;
+	border-style: none;
+	margin: 0;
+	padding: 0;
+	color: #000;
+	list-style-type: none;
+}
+
+#style-menu li + li {
+	border-top: 1px solid #919191;
+}
+
+#style-menu a {
+  width: 6em;
+  padding: 3px;
+  display: block;
+}
+
+#footer {
+  background: #ddd;
+  border-top: 1px solid #aaa;
+  padding: 0.5em 0;
+  color: #666;
+  text-align: center;
+  position: absolute;
+  bottom: 0;
+  width: 100%;
+  height: 3em;
+}
+
+/* @end */
+
+/* @group Front Matter */
+
+#table-of-contents {
+  float: right;
+  clear: right;
+  background: #faf9dc;
+  border: 1px solid #d8d7ad;
+  padding: 0.5em 1em;
+  max-width: 20em;
+  margin: 0.5em 0 1em 1em;
+}
+
+#table-of-contents .caption {
+  text-align: center;
+  margin: 0;
+}
+
+#table-of-contents ul {
+  list-style: none;
+  margin: 0;
+}
+
+#table-of-contents ul ul {
+  margin-left: 2em;
+}
+
+#description .caption {
+  display: none;
+}
+
+#synopsis {
+  display: none;
+}
+
+.no-frame #synopsis {
+  display: block;
+  position: fixed;
+  right: 0;
+  height: 80%;
+  top: 10%;
+  padding: 0;
+  max-width: 75%;
+}
+
+#synopsis .caption {
+  float: left;
+  width: 29px;
+  color: rgba(255,255,255,0);
+  height: 110px;
+  margin: 0;
+  font-size: 1px;
+  padding: 0;
+}
+
+#synopsis p.caption.collapser {
+  background: url(synopsis.png) no-repeat -64px -8px;
+}
+
+#synopsis p.caption.expander {
+  background: url(synopsis.png) no-repeat 0px -8px;
+}
+
+#synopsis ul {
+  height: 100%;
+  overflow: auto;
+  padding: 0.5em;
+  margin: 0;
+}
+
+#synopsis ul ul {
+  overflow: hidden;
+}
+
+#synopsis ul,
+#synopsis ul li.src {
+  background-color: #faf9dc;
+  white-space: nowrap;
+  list-style: none;
+  margin-left: 0;
+}
+
+/* @end */
+
+/* @group Main Content */
+
+#interface div.top { margin: 2em 0; }
+#interface h1 + div.top,
+#interface h2 + div.top,
+#interface h3 + div.top,
+#interface h4 + div.top,
+#interface h5 + div.top {
+ 	margin-top: 1em;
+}
+#interface p.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface td.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface span.fixity {
+  color: #919191;
+  border-left: 1px solid #919191;
+  padding: 0.2em 0.5em 0.2em 0.5em;
+  margin: 0 -1em 0 1em;
+}
+
+#interface span.rightedge {
+  border-left: 1px solid #919191;
+  padding: 0.2em 0 0.2em 0;
+  margin: 0 0 0 1em;
+}
+
+#interface table { border-spacing: 2px; }
+#interface td {
+  vertical-align: top;
+  padding-left: 0.5em;
+}
+#interface td.src {
+  white-space: nowrap;
+}
+#interface td.doc p {
+  margin: 0;
+}
+#interface td.doc p + p {
+  margin-top: 0.8em;
+}
+
+.clearfix:after {
+  clear: both;
+  content: " ";
+  display: block;
+  height: 0;
+  visibility: hidden;
+}
+
+.subs dl {
+  margin: 0;
+}
+
+.subs dt {
+  float: left;
+  clear: left;
+  display: block;
+  margin: 1px 0;
+}
+
+.subs dd {
+  float: right;
+  width: 90%;
+  display: block;
+  padding-left: 0.5em;
+  margin-bottom: 0.5em;
+}
+
+.subs dd.empty {
+  display: none;
+}
+
+.subs dd p {
+  margin: 0;
+}
+
+/* Render short-style data instances */
+.inst ul {
+  height: 100%;
+  padding: 0.5em;
+  margin: 0;
+}
+
+.inst, .inst li {
+  list-style: none;
+  margin-left: 1em;
+}
+
+/* Workaround for bug in Firefox (issue #384) */
+.inst-left {
+  float: left;
+}
+
+.top p.src {
+  border-top: 1px solid #ccc;
+}
+
+.subs, .doc {
+  /* use this selector for one level of indent */
+  padding-left: 2em;
+}
+
+.warning {
+  color: red;
+}
+
+.arguments {
+  margin-top: -0.4em;
+}
+.arguments .caption {
+  display: none;
+}
+
+.fields { padding-left: 1em; }
+
+.fields .caption { display: none; }
+
+.fields p { margin: 0 0; }
+
+/* this seems bulky to me
+.methods, .constructors {
+  background: #f8f8f8;
+  border: 1px solid #eee;
+}
+*/
+
+/* @end */
+
+/* @group Auxillary Pages */
+
+
+.extension-list {
+    list-style-type: none;
+    margin-left: 0;
+}
+
+#mini {
+  margin: 0 auto;
+  padding: 0 1em 1em;
+}
+
+#mini > * {
+  font-size: 93%; /* 12pt */
+}
+
+#mini #module-list .caption,
+#mini #module-header .caption {
+  font-size: 125%; /* 15pt */
+}
+
+#mini #interface h1,
+#mini #interface h2,
+#mini #interface h3,
+#mini #interface h4 {
+  font-size: 109%; /* 13pt */
+  margin: 1em 0 0;
+}
+
+#mini #interface .top,
+#mini #interface .src {
+  margin: 0;
+}
+
+#mini #module-list ul {
+  list-style: none;
+  margin: 0;
+}
+
+#alphabet ul {
+	list-style: none;
+	padding: 0;
+	margin: 0.5em 0 0;
+	text-align: center;
+}
+
+#alphabet li {
+	display: inline;
+	margin: 0 0.25em;
+}
+
+#alphabet a {
+	font-weight: bold;
+}
+
+#index .caption,
+#module-list .caption { font-size: 131%; /* 17pt */ }
+
+#index table {
+  margin-left: 2em;
+}
+
+#index .src {
+  font-weight: bold;
+}
+#index .alt {
+  font-size: 77%; /* 10pt */
+  font-style: italic;
+  padding-left: 2em;
+}
+
+#index td + td {
+  padding-left: 1em;
+}
+
+#module-list ul {
+  list-style: none;
+  margin: 0 0 0 2em;
+}
+
+#module-list li {
+  clear: right;
+}
+
+#module-list span.collapser,
+#module-list span.expander {
+  background-position: 0 0.3em;
+}
+
+#module-list .package {
+  float: right;
+}
+
+/* @end */
diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/plus.gif b/docs/haddock/tensorflow-mnist-0.1.0.0/plus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..2d15c14173d23f664b955cd24f51c82f5f09d91d
GIT binary patch
literal 59
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS5iS+a@;NG@rPSdH*>gbBX
M^XE!9f*2UA0nx1yDgXcg

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/src/Paths_tensorflow_mnist.html b/docs/haddock/tensorflow-mnist-0.1.0.0/src/Paths_tensorflow_mnist.html
new file mode 100644
index 0000000..96d3087
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-0.1.0.0/src/Paths_tensorflow_mnist.html
@@ -0,0 +1,46 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Paths_tensorflow_mnist.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Paths_tensorflow_mnist</span> <span class='hs-layout'>(</span>
+<a name="line-2"></a>    <span class='hs-varid'>version</span><span class='hs-layout'>,</span>
+<a name="line-3"></a>    <span class='hs-varid'>getBinDir</span><span class='hs-layout'>,</span> <span class='hs-varid'>getLibDir</span><span class='hs-layout'>,</span> <span class='hs-varid'>getDataDir</span><span class='hs-layout'>,</span> <span class='hs-varid'>getLibexecDir</span><span class='hs-layout'>,</span>
+<a name="line-4"></a>    <span class='hs-varid'>getDataFileName</span><span class='hs-layout'>,</span> <span class='hs-varid'>getSysconfDir</span>
+<a name="line-5"></a>  <span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-6"></a>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Exception</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Exception</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Version</span> <span class='hs-layout'>(</span><span class='hs-conid'>Version</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>System</span><span class='hs-varop'>.</span><span class='hs-conid'>Environment</span> <span class='hs-layout'>(</span><span class='hs-varid'>getEnv</span><span class='hs-layout'>)</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Prelude</span>
+<a name="line-11"></a>
+<a name="line-12"></a><a name="catchIO"></a><span class='hs-definition'>catchIO</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Exception</span><span class='hs-varop'>.</span><span class='hs-conid'>IOException</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span>
+<a name="line-13"></a><span class='hs-definition'>catchIO</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Exception</span><span class='hs-varop'>.</span><span class='hs-varid'>catch</span>
+<a name="line-14"></a>
+<a name="line-15"></a><a name="version"></a><span class='hs-definition'>version</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Version</span>
+<a name="line-16"></a><span class='hs-definition'>version</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Version</span> <span class='hs-keyglyph'>[</span><span class='hs-num'>0</span><span class='hs-layout'>,</span><span class='hs-num'>1</span><span class='hs-layout'>,</span><span class='hs-num'>0</span><span class='hs-layout'>,</span><span class='hs-num'>0</span><span class='hs-keyglyph'>]</span> <span class='hs-conid'>[]</span>
+<a name="line-17"></a><a name="bindir"></a><span class='hs-definition'>bindir</span><span class='hs-layout'>,</span> <span class='hs-varid'>libdir</span><span class='hs-layout'>,</span> <span class='hs-varid'>datadir</span><span class='hs-layout'>,</span> <span class='hs-varid'>libexecdir</span><span class='hs-layout'>,</span> <span class='hs-varid'>sysconfdir</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>FilePath</span>
+<a name="line-18"></a>
+<a name="line-19"></a><span class='hs-definition'>bindir</span>     <span class='hs-keyglyph'>=</span> <span class='hs-str'>"/home/gnezdo/tensorflow-haskell/.stack-work/install/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/lts-6.2/7.10.3/bin"</span>
+<a name="line-20"></a><a name="libdir"></a><span class='hs-definition'>libdir</span>     <span class='hs-keyglyph'>=</span> <span class='hs-str'>"/home/gnezdo/tensorflow-haskell/.stack-work/install/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/lts-6.2/7.10.3/lib/x86_64-linux-ghc-7.10.3/tensorflow-mnist-0.1.0.0-2agxd0imrn964MW1mWb4VF"</span>
+<a name="line-21"></a><a name="datadir"></a><span class='hs-definition'>datadir</span>    <span class='hs-keyglyph'>=</span> <span class='hs-str'>"/home/gnezdo/tensorflow-haskell/.stack-work/install/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/lts-6.2/7.10.3/share/x86_64-linux-ghc-7.10.3/tensorflow-mnist-0.1.0.0"</span>
+<a name="line-22"></a><a name="libexecdir"></a><span class='hs-definition'>libexecdir</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"/home/gnezdo/tensorflow-haskell/.stack-work/install/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/lts-6.2/7.10.3/libexec"</span>
+<a name="line-23"></a><a name="sysconfdir"></a><span class='hs-definition'>sysconfdir</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"/home/gnezdo/tensorflow-haskell/.stack-work/install/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/lts-6.2/7.10.3/etc"</span>
+<a name="line-24"></a>
+<a name="line-25"></a><a name="getBinDir"></a><span class='hs-definition'>getBinDir</span><span class='hs-layout'>,</span> <span class='hs-varid'>getLibDir</span><span class='hs-layout'>,</span> <span class='hs-varid'>getDataDir</span><span class='hs-layout'>,</span> <span class='hs-varid'>getLibexecDir</span><span class='hs-layout'>,</span> <span class='hs-varid'>getSysconfDir</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>FilePath</span>
+<a name="line-26"></a><span class='hs-definition'>getBinDir</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>catchIO</span> <span class='hs-layout'>(</span><span class='hs-varid'>getEnv</span> <span class='hs-str'>"tensorflow_mnist_bindir"</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-keyword'>_</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>return</span> <span class='hs-varid'>bindir</span><span class='hs-layout'>)</span>
+<a name="line-27"></a><a name="getLibDir"></a><span class='hs-definition'>getLibDir</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>catchIO</span> <span class='hs-layout'>(</span><span class='hs-varid'>getEnv</span> <span class='hs-str'>"tensorflow_mnist_libdir"</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-keyword'>_</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>return</span> <span class='hs-varid'>libdir</span><span class='hs-layout'>)</span>
+<a name="line-28"></a><a name="getDataDir"></a><span class='hs-definition'>getDataDir</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>catchIO</span> <span class='hs-layout'>(</span><span class='hs-varid'>getEnv</span> <span class='hs-str'>"tensorflow_mnist_datadir"</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-keyword'>_</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>return</span> <span class='hs-varid'>datadir</span><span class='hs-layout'>)</span>
+<a name="line-29"></a><a name="getLibexecDir"></a><span class='hs-definition'>getLibexecDir</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>catchIO</span> <span class='hs-layout'>(</span><span class='hs-varid'>getEnv</span> <span class='hs-str'>"tensorflow_mnist_libexecdir"</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-keyword'>_</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>return</span> <span class='hs-varid'>libexecdir</span><span class='hs-layout'>)</span>
+<a name="line-30"></a><a name="getSysconfDir"></a><span class='hs-definition'>getSysconfDir</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>catchIO</span> <span class='hs-layout'>(</span><span class='hs-varid'>getEnv</span> <span class='hs-str'>"tensorflow_mnist_sysconfdir"</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-keyword'>_</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>return</span> <span class='hs-varid'>sysconfdir</span><span class='hs-layout'>)</span>
+<a name="line-31"></a>
+<a name="line-32"></a><a name="getDataFileName"></a><span class='hs-definition'>getDataFileName</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>FilePath</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>FilePath</span>
+<a name="line-33"></a><span class='hs-definition'>getDataFileName</span> <span class='hs-varid'>name</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-34"></a>  <span class='hs-varid'>dir</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>getDataDir</span>
+<a name="line-35"></a>  <span class='hs-varid'>return</span> <span class='hs-layout'>(</span><span class='hs-varid'>dir</span> <span class='hs-varop'>++</span> <span class='hs-str'>"/"</span> <span class='hs-varop'>++</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/src/TensorFlow-Examples-MNIST-Parse.html b/docs/haddock/tensorflow-mnist-0.1.0.0/src/TensorFlow-Examples-MNIST-Parse.html
new file mode 100644
index 0000000..27b8468
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-0.1.0.0/src/TensorFlow-Examples-MNIST-Parse.html
@@ -0,0 +1,107 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/Examples/MNIST/Parse.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE OverloadedStrings #-}</span>
+<a name="line-16"></a><span class='hs-comment'>{-# LANGUAGE OverloadedLists #-}</span>
+<a name="line-17"></a><span class='hs-comment'>{-# LANGUAGE TypeSynonymInstances #-}</span>
+<a name="line-18"></a><span class='hs-comment'>{-# LANGUAGE FlexibleInstances #-}</span>
+<a name="line-19"></a><span class='hs-comment'>{-# LANGUAGE ViewPatterns #-}</span>
+<a name="line-20"></a>
+<a name="line-21"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Examples</span><span class='hs-varop'>.</span><span class='hs-conid'>MNIST</span><span class='hs-varop'>.</span><span class='hs-conid'>Parse</span> <span class='hs-keyword'>where</span>
+<a name="line-22"></a>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Monad</span> <span class='hs-layout'>(</span><span class='hs-varid'>when</span><span class='hs-layout'>,</span> <span class='hs-varid'>liftM</span><span class='hs-layout'>)</span>
+<a name="line-24"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Binary</span><span class='hs-varop'>.</span><span class='hs-conid'>Get</span> <span class='hs-layout'>(</span><span class='hs-conid'>Get</span><span class='hs-layout'>,</span> <span class='hs-varid'>runGet</span><span class='hs-layout'>,</span> <span class='hs-varid'>getWord32be</span><span class='hs-layout'>,</span> <span class='hs-varid'>getLazyByteString</span><span class='hs-layout'>)</span>
+<a name="line-25"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>Lazy</span> <span class='hs-layout'>(</span><span class='hs-varid'>toStrict</span><span class='hs-layout'>,</span> <span class='hs-varid'>readFile</span><span class='hs-layout'>)</span>
+<a name="line-26"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>List</span><span class='hs-varop'>.</span><span class='hs-conid'>Split</span> <span class='hs-layout'>(</span><span class='hs-varid'>chunksOf</span><span class='hs-layout'>)</span>
+<a name="line-27"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Monoid</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varop'>&lt;&gt;</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-28"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span> <span class='hs-layout'>(</span><span class='hs-conid'>Message</span><span class='hs-layout'>,</span> <span class='hs-varid'>decodeMessageOrDie</span><span class='hs-layout'>)</span>
+<a name="line-29"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-layout'>(</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-30"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span> <span class='hs-layout'>(</span><span class='hs-conid'>Word8</span><span class='hs-layout'>,</span> <span class='hs-conid'>Word32</span><span class='hs-layout'>)</span>
+<a name="line-31"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Prelude</span> <span class='hs-varid'>hiding</span> <span class='hs-layout'>(</span><span class='hs-varid'>readFile</span><span class='hs-layout'>)</span>
+<a name="line-32"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Codec</span><span class='hs-varop'>.</span><span class='hs-conid'>Compression</span><span class='hs-varop'>.</span><span class='hs-conid'>GZip</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>GZip</span>
+<a name="line-33"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>Lazy</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>L</span>
+<a name="line-34"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Text</span>
+<a name="line-35"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Vector</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>V</span>
+<a name="line-36"></a>
+<a name="line-37"></a><a name="MNIST"></a><span class='hs-comment'>-- | Utilities specific to MNIST.</span>
+<a name="line-38"></a><a name="MNIST"></a><span class='hs-keyword'>type</span> <span class='hs-conid'>MNIST</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>V</span><span class='hs-varop'>.</span><span class='hs-conid'>Vector</span> <span class='hs-conid'>Word8</span>
+<a name="line-39"></a>
+<a name="line-40"></a><a name="drawMNIST"></a><span class='hs-comment'>-- | Produces a unicode rendering of the MNIST digit sample.</span>
+<a name="line-41"></a><span class='hs-definition'>drawMNIST</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>MNIST</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Text</span>
+<a name="line-42"></a><span class='hs-definition'>drawMNIST</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>chunk</span> <span class='hs-varop'>.</span> <span class='hs-varid'>block</span>
+<a name="line-43"></a>  <span class='hs-keyword'>where</span>
+<a name="line-44"></a>    <span class='hs-varid'>block</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>V</span><span class='hs-varop'>.</span><span class='hs-conid'>Vector</span> <span class='hs-conid'>Word8</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Text</span>
+<a name="line-45"></a>    <span class='hs-varid'>block</span> <span class='hs-layout'>(</span><span class='hs-conid'>V</span><span class='hs-varop'>.</span><span class='hs-varid'>splitAt</span> <span class='hs-num'>1</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>[</span><span class='hs-num'>0</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span> <span class='hs-varid'>xs</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>" "</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-varid'>block</span> <span class='hs-varid'>xs</span>
+<a name="line-46"></a>    <span class='hs-varid'>block</span> <span class='hs-layout'>(</span><span class='hs-conid'>V</span><span class='hs-varop'>.</span><span class='hs-varid'>splitAt</span> <span class='hs-num'>1</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>[</span><span class='hs-varid'>n</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span> <span class='hs-varid'>xs</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>c</span> <span class='hs-varop'>`</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>cons</span><span class='hs-varop'>`</span> <span class='hs-varid'>block</span> <span class='hs-varid'>xs</span>
+<a name="line-47"></a>      <span class='hs-keyword'>where</span> <span class='hs-varid'>c</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"\9617\9618\9619\9608"</span> <span class='hs-varop'>!!</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-layout'>(</span><span class='hs-varid'>n</span> <span class='hs-varop'>`div`</span> <span class='hs-num'>64</span><span class='hs-layout'>)</span>
+<a name="line-48"></a>    <span class='hs-varid'>block</span> <span class='hs-layout'>(</span><span class='hs-conid'>V</span><span class='hs-varop'>.</span><span class='hs-varid'>splitAt</span> <span class='hs-num'>1</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyword'>_</span><span class='hs-layout'>)</span>   <span class='hs-keyglyph'>=</span> <span class='hs-str'>""</span>
+<a name="line-49"></a>    <span class='hs-varid'>chunk</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Text</span>
+<a name="line-50"></a>    <span class='hs-varid'>chunk</span> <span class='hs-str'>""</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"\n"</span>
+<a name="line-51"></a>    <span class='hs-varid'>chunk</span> <span class='hs-varid'>xs</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>take</span> <span class='hs-num'>28</span> <span class='hs-varid'>xs</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-str'>"\n"</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-varid'>chunk</span> <span class='hs-layout'>(</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>drop</span> <span class='hs-num'>28</span> <span class='hs-varid'>xs</span><span class='hs-layout'>)</span>
+<a name="line-52"></a>
+<a name="line-53"></a><a name="checkEndian"></a><span class='hs-comment'>-- | Check's the file's endianess, throwing an error if it's not as expected.</span>
+<a name="line-54"></a><span class='hs-definition'>checkEndian</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Get</span> <span class='hs-conid'>()</span>
+<a name="line-55"></a><span class='hs-definition'>checkEndian</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-56"></a>    <span class='hs-varid'>magic</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>getWord32be</span>
+<a name="line-57"></a>    <span class='hs-varid'>when</span> <span class='hs-layout'>(</span><span class='hs-varid'>magic</span> <span class='hs-varop'>`notElem`</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>[</span><span class='hs-num'>2049</span><span class='hs-layout'>,</span> <span class='hs-num'>2051</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Word32</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-varop'>$</span>
+<a name="line-58"></a>        <span class='hs-varid'>fail</span> <span class='hs-str'>"Expected big endian, but image file is little endian."</span>
+<a name="line-59"></a>
+<a name="line-60"></a><a name="readMNISTSamples"></a><span class='hs-comment'>-- | Reads an MNIST file and returns a list of samples.</span>
+<a name="line-61"></a><span class='hs-definition'>readMNISTSamples</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>FilePath</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>MNIST</span><span class='hs-keyglyph'>]</span>
+<a name="line-62"></a><span class='hs-definition'>readMNISTSamples</span> <span class='hs-varid'>path</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-63"></a>    <span class='hs-varid'>raw</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>GZip</span><span class='hs-varop'>.</span><span class='hs-varid'>decompress</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>readFile</span> <span class='hs-varid'>path</span>
+<a name="line-64"></a>    <span class='hs-varid'>return</span> <span class='hs-varop'>$</span> <span class='hs-varid'>runGet</span> <span class='hs-varid'>getMNIST</span> <span class='hs-varid'>raw</span>
+<a name="line-65"></a>  <span class='hs-keyword'>where</span>
+<a name="line-66"></a>    <span class='hs-varid'>getMNIST</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Get</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>MNIST</span><span class='hs-keyglyph'>]</span>
+<a name="line-67"></a>    <span class='hs-varid'>getMNIST</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-68"></a>        <span class='hs-varid'>checkEndian</span>
+<a name="line-69"></a>        <span class='hs-comment'>-- Parse header data.</span>
+<a name="line-70"></a>        <span class='hs-varid'>cnt</span>  <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>liftM</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-varid'>getWord32be</span>
+<a name="line-71"></a>        <span class='hs-varid'>rows</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>liftM</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-varid'>getWord32be</span>
+<a name="line-72"></a>        <span class='hs-varid'>cols</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>liftM</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-varid'>getWord32be</span>
+<a name="line-73"></a>        <span class='hs-comment'>-- Read all of the data, then split into samples.</span>
+<a name="line-74"></a>        <span class='hs-varid'>pixels</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>getLazyByteString</span> <span class='hs-varop'>$</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-varop'>$</span> <span class='hs-varid'>cnt</span> <span class='hs-varop'>*</span> <span class='hs-varid'>rows</span> <span class='hs-varop'>*</span> <span class='hs-varid'>cols</span>
+<a name="line-75"></a>        <span class='hs-varid'>return</span> <span class='hs-varop'>$</span> <span class='hs-conid'>V</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>chunksOf</span> <span class='hs-layout'>(</span><span class='hs-varid'>rows</span> <span class='hs-varop'>*</span> <span class='hs-varid'>cols</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-conid'>L</span><span class='hs-varop'>.</span><span class='hs-varid'>unpack</span> <span class='hs-varid'>pixels</span><span class='hs-layout'>)</span>
+<a name="line-76"></a>
+<a name="line-77"></a><a name="readMNISTLabels"></a><span class='hs-comment'>-- | Reads a list of MNIST labels from a file and returns them.</span>
+<a name="line-78"></a><span class='hs-definition'>readMNISTLabels</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>FilePath</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Word8</span><span class='hs-keyglyph'>]</span>
+<a name="line-79"></a><span class='hs-definition'>readMNISTLabels</span> <span class='hs-varid'>path</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-80"></a>    <span class='hs-varid'>raw</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>GZip</span><span class='hs-varop'>.</span><span class='hs-varid'>decompress</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>readFile</span> <span class='hs-varid'>path</span>
+<a name="line-81"></a>    <span class='hs-varid'>return</span> <span class='hs-varop'>$</span> <span class='hs-varid'>runGet</span> <span class='hs-varid'>getLabels</span> <span class='hs-varid'>raw</span>
+<a name="line-82"></a>  <span class='hs-keyword'>where</span> <span class='hs-varid'>getLabels</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Get</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Word8</span><span class='hs-keyglyph'>]</span>
+<a name="line-83"></a>        <span class='hs-varid'>getLabels</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-84"></a>            <span class='hs-varid'>checkEndian</span>
+<a name="line-85"></a>            <span class='hs-comment'>-- Parse header data.</span>
+<a name="line-86"></a>            <span class='hs-varid'>cnt</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>liftM</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-varid'>getWord32be</span>
+<a name="line-87"></a>            <span class='hs-comment'>-- Read all of the labels.</span>
+<a name="line-88"></a>            <span class='hs-conid'>L</span><span class='hs-varop'>.</span><span class='hs-varid'>unpack</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>getLazyByteString</span> <span class='hs-varid'>cnt</span>
+<a name="line-89"></a>
+<a name="line-90"></a><a name="readMessageFromFileOrDie"></a><span class='hs-definition'>readMessageFromFileOrDie</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Message</span> <span class='hs-varid'>m</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>FilePath</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>m</span>
+<a name="line-91"></a><span class='hs-definition'>readMessageFromFileOrDie</span> <span class='hs-varid'>path</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-92"></a>    <span class='hs-varid'>pb</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>readFile</span> <span class='hs-varid'>path</span>
+<a name="line-93"></a>    <span class='hs-varid'>return</span> <span class='hs-varop'>$</span> <span class='hs-varid'>decodeMessageOrDie</span> <span class='hs-varop'>$</span> <span class='hs-varid'>toStrict</span> <span class='hs-varid'>pb</span>
+<a name="line-94"></a>
+<a name="line-95"></a><span class='hs-comment'>-- TODO: Write a writeMessageFromFileOrDie and read/write non-lethal</span>
+<a name="line-96"></a><span class='hs-comment'>--             versions.</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/src/TensorFlow-Examples-MNIST-TrainedGraph.html b/docs/haddock/tensorflow-mnist-0.1.0.0/src/TensorFlow-Examples-MNIST-TrainedGraph.html
new file mode 100644
index 0000000..881e4d8
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-0.1.0.0/src/TensorFlow-Examples-MNIST-TrainedGraph.html
@@ -0,0 +1,41 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src-data/TensorFlow/Examples/MNIST/TrainedGraph.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE OverloadedStrings #-}</span>
+<a name="line-16"></a><span class='hs-comment'>-- | Paths to test helper files.</span>
+<a name="line-17"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Examples</span><span class='hs-varop'>.</span><span class='hs-conid'>MNIST</span><span class='hs-varop'>.</span><span class='hs-conid'>TrainedGraph</span> <span class='hs-keyword'>where</span>
+<a name="line-18"></a>
+<a name="line-19"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Paths_tensorflow_mnist</span> <span class='hs-layout'>(</span><span class='hs-varid'>getDataFileName</span><span class='hs-layout'>)</span>
+<a name="line-20"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-layout'>(</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>Char8</span> <span class='hs-layout'>(</span><span class='hs-varid'>pack</span><span class='hs-layout'>)</span>
+<a name="line-22"></a>
+<a name="line-23"></a><a name="mnistPb"></a><span class='hs-comment'>-- | File containing a Tensorflow serialized proto of MNIST.</span>
+<a name="line-24"></a><span class='hs-definition'>mnistPb</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>FilePath</span>
+<a name="line-25"></a><span class='hs-definition'>mnistPb</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>getDataFileName</span> <span class='hs-str'>"data/MNIST.pb"</span>
+<a name="line-26"></a>
+<a name="line-27"></a><a name="wtsCkpt"></a><span class='hs-comment'>-- | Files containing pre-trained weights for MNIST.</span>
+<a name="line-28"></a><span class='hs-definition'>wtsCkpt</span><span class='hs-layout'>,</span> <span class='hs-varid'>biasCkpt</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>ByteString</span>
+<a name="line-29"></a><span class='hs-definition'>wtsCkpt</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>pack</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>getDataFileName</span> <span class='hs-str'>"data/MNISTWts.ckpt"</span>
+<a name="line-30"></a><a name="biasCkpt"></a><span class='hs-definition'>biasCkpt</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>pack</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>getDataFileName</span> <span class='hs-str'>"data/MNISTBias.ckpt"</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/src/hscolour.css b/docs/haddock/tensorflow-mnist-0.1.0.0/src/hscolour.css
new file mode 100644
index 0000000..c15919e
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-0.1.0.0/src/hscolour.css
@@ -0,0 +1,5 @@
+.hs-keyglyph, .hs-layout {color: red;}
+.hs-keyword {color: blue;}
+.hs-comment, .hs-comment a {color: green;}
+.hs-str, .hs-chr {color: teal;}
+.hs-keyword, .hs-conid, .hs-varid, .hs-conop, .hs-varop, .hs-num, .hs-cpp, .hs-sel, .hs-definition {}
diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/synopsis.png b/docs/haddock/tensorflow-mnist-0.1.0.0/synopsis.png
new file mode 100644
index 0000000000000000000000000000000000000000..85fb86ec84907bcc86531dc82871948ff4d471fa
GIT binary patch
literal 11327
zcmV-FEWp!=P)<h;3K|Lk000e1NJLTq004jh003wR1^@s6IVt!g000+IX+uL$Nkc;*
zP;zf(X>4Tx0C)k_S!GyNTeqHT_l8Y(cXyX`gGi?cY`Qxn1VID|MJXwjPC)?)F$h6K
zMMOd+6hs7sqbPzXbr*U(-*=zy-hcPcUC*=TdiNM(jyd-lv&OpsU|J&v2m2!^0SE{T
z54F(O;<U4O;KWV=44?vZKnREv-NJ%3&CHI1|Ml9M1kj;!OzhCt|7!L>E2!K(!rTCW
z%wV;vdzf1QjBf#e&~gh74F>?Z4a=WLg$KhJ^$5nap>PLbJadS>e&h8+?D`9%QNL`g
zEVKbYGXj7k5Q(8)0Fd#*a?VIMFW3*64geVHKzE-&0BG!BtmfuTbO(T`0Jaeg2nagF
z{V*1E{Wm{e|AvV~*MEExiC+KU-~R=!2{)|c6Bg`GjQ;iG|FQ`1kAUCTuZtQk34#8{
z4r4<P-^=bq9b@>(3g7#|{=Z@d+d#}7f!3C=>=26vx*jwA8>@MS>RG@Tt_zt3hie^T
z_?<zX!Qp0Bf6L=zytR+Pm;lRjJ;HSV_$l=z8Xd-r9G16)Mp#(Gm>0%9VUd=)Fos7I
z^ghPh%Jy%YZ|)vCf6EaFPai$Q-!=$ppK!y&wrJs)bNdAuANB!m3n34Tfj{s75g-&U
z1A!Pg3bcXF-=!Gv1VmU93G2duANT;{0JugFTqg*|oPXPC|A$2HS3NJd-hcPV3EW`Y
zh=1Dr-5Mv{<{zIvz#Yba<V3iQL)~HT!km4Q{?+Cmi$j0EdL!8&xgxpyr%xz;lr%~X
zrG-*OsiBkrCyE}$48%~fC`FVeN*$#F>y&^Vcn^E_`qRfl{{bzYkp)4~$~NAx_VB;E
z{?P)PU)DbV{Qi#~0H0@T9czDj06@6MNq8OrpdAz(9qQxd9nPr<&s+~tPQySqaZyfb
zNh!%g_5YjeaLxMN*$s<dYY{|0@-La=c>v_p;d%b#U$Wpz0Geb0U>E+EOsEQ;I<Re%
zqFY30<Zpkd!v`9W019}7G68nL0|epsEDhv=3eW($zzCQED_{?t;PK-PkHJt71>!&=
zNC6q(BFF<p;2Nj^wV)BSfDUjM^n!=r33v`Bz%-Zxi(nP3gDtQFK@b)qgJ>Wohy&t-
zL?CHM5mJM6p`(xmWDmJOUQi$u0mVUQpbRJ*DuT+OI;a`C4fR4p&?xj8nuk`Puh35f
z5<x)FBG?f82nmEDLK9((utJ<bcp*X%F^FWu1w;X&9C01dj_5@^K};a#5UYqyBtViP
z>5*JWF{C0=8)=GkKzbrWk@3iMWInPS*@Wyu4kE{pbI3L14-^JPgW^Pq!Q<2bWsPz}
zg`nb5nW!REEvg;Wj~YYGqt;RTXfiY_S_G|(HbmQ@z0gtU6m&ki8r_B-Ku@3-(OVb{
zh8`n;QNS2r>@mKWSWG773g!l;2Q!LUz-(f%SSG9pRuyZCC1S&|DcC~nb!<2G1$Gg;
zjU&Zz;G}VSI0sxHE(w>9tH<5Py}&KucJP#VKD;vC6z`6Y#%JLx@m=^4{33pbgo;Ff
zM3uyf#Fr$Iq=2M}WPoIbWP_BHl$%tE)ST3Z^fYM!=}po{r1PXd2-E~&f;PdC5J9*=
zs3G(aUK2LR$jJD~G{_vt!pSa>)sa0QdqcKOPD3tEZb<G<o<Lqq-cJ64{1XM9f{#Lz
z!ige=BA=p_VuWIi5>LrbsZB|wjHfK7yiNI%a+8XNN{Y&qDu61Js-9|yYMB~K%}=dM
z?M|IcT|xbTdVvN>!$YG@<3@9arjllWW|0;{D?n>V>r0zK+erJ2cAbuzPL|Gw?j&6?
z-95TFdL%tRy&=6neHMKS{UrTQ1~vvw1`mcbh9-s=4Br`97&RC@7}FVVFitT3Wa4Df
zW%6UX#MHqw%Zy?cW;SPzV!p~ez`Vvn%c8>K#*)s`!ZO8*U=?PyV2x$1V13HE$;Qs6
z&lb#9$o7D3jh&udgWZ=sm;FBb3I`2`8ix-@E=M=VM@~9UO-_H#0?vNUbuLye1Fi_J
zGOlM_JKO@?*4#<lH@IhcNO%-^Jb7|?`gqoPIe1NY<9Hi*r}*%EihSODm-z<yzVi$4
z+w!OKcknL@FbWt7#0WG9%m|VRstbk)mJ5yxVT2Tg{DiIuJr_m@%L#i67YmPwAVlOv
zd_=B@jEbT~l|%zY%SB&`5yUjbBE=fS=Edp7kBTRWcZh$I;FWNYxFj(su`ekn=`UF+
zIW0vkWh9j()g|>+T3Fgmx>$N#hD=6JCPAiC=8LR|tcUDX*;jHjawc-Aa(!}p@(S{y
z@=fw93cLy~3MC3J6=@aC6f+ecDWR3LloFKgD*aHFR}NQhQU0tVrsAhkud;kZ;E2bO
z$|DP^+^R&?GSxXXPBj;`QnfjCE_I@Mx%xW|9u0SmYK<jLAx&S+>zbdmB(*}d+O)oF
zD{G(9?$JT&=<B5GJkzDtwbm`teWS;%=cU)6x2`XzAFJPEfHp8TxMc9skj>D|u+DJZ
zNWtioQNJ<4*wVPj_}x+AqoGH;Ob{kUCOIZE$M}u~9_ug#riP|Drn6=OW+7&G%rWL>
z=Ede8ETk;rECwxUES)XuEw`++tg@`8tp%+ktov*zY#eRsY`)v-*k;?#*-6-)vU_6B
zZ0}>=>40^xaj16KJg$2@@A#sloMVdP<OzWjCr=DHF+2G?-E*dJCOWsepj~WT>Ron;
zro?jMrmLZAiR-$Xw%cX5Rd)^dT=x|ZRgY|sB~Mk)Y|mvcRj(Yc6>oL#eD5_MZJ#2a
zFTMu8*L=VGnflfE9r)Y&-w413xCGn|qz?28><?lOIvF$?EE1d&JQt!Ik{7ZbY8+Y}
z28Epn>kOxb4~I`91S8Hy%txw47DsMJ*+jLTq&gXR@@ceibXxRMj9yGtEGpJ5wl9t=
zE-`NYl;)|jcqraAzAu3%Avt03wEpSZM3O|m#Ni~#r0k?`XKc@OC9@@;PF^^xf3_io
zJS8;cWvWW*wR5O*KIfjL$)pvg?Wen^KhBWM$j{i#bjy5vUg~_o`GX6d7oKIwXI;IB
zxfpnH@{;<c+H9)qnCyidlbrTk&fN6eZ+XsnL;14#*DjM>j<`HmaI~Pakhkz+;ck(4
z(L}LU@r@GJlC+ZVSKP0>xT<lr=^ER$jB7tjeM(=I>6f*a^OxsWU@9UjK2+LN4pu2v
z)m1ZBXH@Ui1lG*eTGaN}Db&@~v({%dAQ~bXR<1i<f8MCm*x4l3RCa^@M#fEWGvemQ
zW|!v47UPz_R>jt)TYR@l+GyI++oAU8_Vo_$j=4_z&e7XOxBI$Oy4voD->JFFb+`B)
z-My^)B=?i=A9TlbZ}tTDto3^JF7!F~O+T=EFy3$8|7^f`;L$_9hYtod2fH7sKDs-k
zJaqf9;^U4d@=w~I$~|oxmK$z+CjYE`L}8@!xzh8l(IcbxU#P$69n%?mIBq!pWa8Mw
z=%n@JtCx;1=U%zLT7K>S`pZ=0)Xwzj8T3s0Eahze8`d}FZ-w68n3JEoH?K4Q^qu9q
z=>@li)%RiVcNddCkbTHs;#jI%mR`QQqPOz=<MEG+tHG-~Ybk$G{Z;Zw@YAi&8lRu8
z+pRBt3H@^LHFJY$qvo6Jx4uoY%{N;CTRY#=x0$wUe<=PK{Av4hX(wtIyIZg)wAZz7
zygzdgaBvVz3?=?L3n1YV;pqv$*BUs7u!nPp=KzrO{LU$US-|NE1lz#a2I7HkB7Bib
zC_HKj-GeE>CgGy+9whdp4g`BLCvp!8U&;uov(!a2t+bEnRv6HXyi9t`-YglcEo`$K
zI8GTZXYLH1F5YE+b^&9-c%dfYc~N>X1MygiCdpZ8N*OKLV7W5+5rusvVP$KTgd_E;
zV`@J%*flk^Jhjj1)aX9cTQC5ItVZ(2W=FkE;*aH-)|+*kk6SET?pjmWaNEk+>D${o
z_#cmV%sNr-bj$gX%QW$m8{|&wA?SI;%go!uC))SCU%7vKz~jI-L0?1Ap^RZ7;i?hG
zB3+__P9{WW#uUa@#oavB8Q+`m==5;nXwvwZiR6j1<0+%5!{;8Q^`_s>XwIxTUvlAM
z)|rdpmprp=bM$iM@_6#8@((Vr7Q8HcP;{fXs3iGH;8nY8TBRaov}JqcixtC_ZB<vQ
z&(`?YTGlDob2eZb{<=Qi*ws{eBju)Fvqg(SE88vPt>w07?YBCLI#1vB=rX<|d6)j~
z?!9;SA9XkN<n~7QIX}?p7apJ-*nBuW*z>4rDD83J6N{$`!z{xG&lW}=KCd6md=WHe
zF)la3F!5t@`sLkMS6?Sg5vR3gcxTbGOK%>(y*_twKH{Cjg64anMViI^4{J-a%g<LH
ze(YYoyLRud-cLiHC*aZaV}tse#HQ(1*!QCCfgc+?yt~$WY5Tng2XOpFKtGg+ut(A%
z-=j*=VHi8C5{?;<#;=l0k&Y4u$@<72Q1np_Pz_U$(Jatz(CstOF!D30Fk7<(v!=0C
zv-fel=R$LH^XTz<^JVZi3XBSF3Nwow5p@<z60edNko+Kxkr9;Dk#m)gS143$Q5sTy
zqw-A^tHz`*q@kp#r)8miT*qBEKrc)`${^M-!6?x<@o1vS>0=3|@n*5+(H4=G;Z`Bm
z0XDw2UUnY#t`5ZG&W<NeoN#h<c62#GbaHieBf7hKczXJJ`FjWZMEFMgo$^l#ND0gg
zx)_`nQW$zAtSr1TqB^oJs{Uj{bVE#iY<*nqsjB$0ge#{D6LXTzpE;Wxe>ObDFO_)C
zCe0{aEki1k_dNXt+=U-mA1_W_8p^(%Q<IyQmz*DR*{Q&|P@#yYn4);6Wa-M-)$VKc
zrTJw^<pCA;mAX|@)oe8+HQ#IB*NxV9H#A-^X-sd5zTtDzzS*!vsa50_Ya2Pdf^Bpx
zbxz-Y-ZgON&fVsFb@$7<i+XZ;v-&a~r1qx_oPC%wnEEJVDC=?lldDf_hFhL>j|@Mb
z9sM+h7-yIepVWIvd=>Y)XzKR#)XeT1j<?!#-t!mU-CCGlggyu?nJkB_<bCW|eYv*z
ziR!cPy2cmFuWlPb-=a6;w@!ag*pB@X{?m8o#IEU{#=iIg)35lmfin;%)QHeRj3K>H
zI8-@&65hs?W6g0$Tn9b?K9MevmJ{6JljSOT6GbGYHWfM5G<6M41g#z&E8Qx6H$yI?
z50eHn6Z1ODBi1suSavH8F-{EUJXaTYHjh8AJ|73)7XPq7g<!OhsW6Z5mdKE3k(jTz
zvII_ILb6OMKw41-$PCI}kh79!lAl#5RdiS4Q~ID>t>OirQ5IDz)!g7S$y<#pnvPn`
zTCcP(>sag3>W=B<=vx}l7>pa{8`&AN7|$LpGx0noeC)GnyV)so9SefRgyl6WA8Q%w
zeVfO&`F8I1(hk7k+3~B6fhW|RD4pIpx4EPekGo2^q1>k2n?25Xx_BviQ+coYJoGK~
zi}SY&kPV~?{2VkK+z^r;>Jw%VE)ao<Sd8qCsyuln+BHTamLqmAZW>-y@)AN%A4?QY
z!X(X~xtpASHaNvFl_z!g+(cSqdP;^<ru=#G^Gg@HvkESTTr$e$&EC&>mD`$^mG5`i
zpn$&+Rk%>pUtCp^dd2Um*){o6wlZ|t=klqF!OHfk>gs};%-W>7nEHr@(CeX%5lwM7
zQg7xp*S7SwzHLLbOLn+*Uc0?`NAB*$d)wWCJsW)~{h|X4gV%@BpPU*_8L1qd8t0!(
zdySmVd!st{bK%K{=9Rj&=Ffv)KX1|hFxkC)82{hg(&3(fkq6-NB>?O?0kGBtAd?QJ
zm0$~|LIBLj0I*U5i1iA9XzK$|?dCuG2lOlFq=GX}9v}f{nuc(O=>uZH1yBw;!3bD_
zU{(i`gLA_m=mOLPjX+-zbO<GcGvX|w5itp8bfQQ*WD2qkUVrG}Of3%84DUniXj}9p
z^ivET&bv}EkFcb0hLww*!SUjPaCh){ydAy{ze{3HQcDVv9w+S}&=VpFugO%&D#;1t
zQRE9078DOC<tZDexT%V%>8W#QsA+O&>1m7Uxak_`<>>nu%o*kx!<ev4CCswS!z|vc
zXx2(L9kvhb=^T<AuQ(IA#JHxo(|J^QKJ!-bo#1ES9~a0Gv=U+xdMjKf5+SN1MlZG|
z-Y;=YGG59>T2DqomQ{`*59GHMHWa@qZ7S~^!Kl)z@vEz7SZjuAWovinywxMoS2FN7
zEH|1t%4A}H?2754xrD_j%Moi{n>gE7_6iP##}7_;J59Lg5Ifz(-D^B~y{dc!eQ)?H
z1`GsQ2d{)Cgfm98MOmHv9&;s5@6?xs(nO0hxa6LcxN|CLdl`M_GqP+i31t7w9nHU9
zkY40hVt!S*RG^%pl2DDR1@+)Ms)_U_Lks^c#r9*J-d)LeEAIFAEIl9{kQ}rbihXiz
zxOZfJbZ?wtQtXx5l+ld&8>=~scSi5kK8P(dtn9DO{nh=s_)Emb(M`^+uiKA)7VrA)
zEB#tO5ODlSVZM$P@WWh#2Fx+Iz|6u~m`%6|24UXdCqxG`1g0=2kOkd@#-Q&AR(P%P
zMdTpvAy(jBM;jT2tUyk{D~~EF3{{U>K(nFk;T(JdLx-`&6l3PF0@xsI7Y>87!d2q7
z@J9GD{0|aKlAELyq`{<P1SLWl8G$UCY?<7Fe1t-eqMK5MvWrTEsvFK%hiUBK+%$p?
zLsvvEML)#g&IlN<F{v}XV@_oeW0_<<%_hn=1;_jd$2w;fmm@bl_bAT=UK2hFz8Cxj
z0?vX$f}2A9!g(UzqH1CcVqe9dOEgR7NySNf$XLi~$;ru!DDWzBDsd|Fs0baAQPog0
zQ+L;h)y&gs)gIT`*5lMSHi$92ZZv<C*~A7u*Ss=kxA3yOWsS12wQaP=IJh3~J;CFY
z>in5#@A}YP&ZEYQ#XH-V)Gsvv6_^~14ao?j4lj=6k7|w9iW!UZJhhvUlPHq(FxfQ)
zq?V>>q`%8dxgeZ1aw#H*HTOZjUjc35y<*QR6jwV-iRB~}tyPXS=-S45n}+?ysv9OZ
zzqJ(K(rR1j$hs}xHG4PtzG(M&@2Lj@{VyISJQ5#z^W@U7{<DMUcV5Jft50HIPQ0$2
zPMURnYc#L4Ah{^9q_ASN>hV|l=i6Vte3RLV-yYuK+dKCw{z!laG%#N$3ABJM%p<0O
zYA^skKqQbP%m$r-WBwLFh0ujLomRwONMWQ8vL5*f<`CmhgJ?Rm2f718hVj63W7)9r
z*mpQXTq~XnpG|@xNg&xFjU_!Gq>|CVvs#J#1w}9=HDxE2J2egUAWZ`85!yYvKKcv>
zJ4PYKJ*G+KW|m8=VQlv7TJY|}%0<qN;Qq<;jdz`Io&T%Amf)T+R)kiRM@&}SK*Cw_
zq*ShSi_DnpraX&+wqk%%q4J>0wyKDli~41a=UN19Bb{{JVSQ=?d&3H&&qviwE*<+|
zre<Dfv2EpG-EAvqm*;>!9^?4cDF}{Txa*#Kx+jZQvyZXwvVVG@WYFu7)G)>HwaCho
zPBE;pGpDX4cqED@Z6)`nTsY^LE}F4-ek7|Lj+#LpTmF}Vfuf?4z^j_2v}GSEI;v7@
ztn0YySFg7=Mcq_r{?^*qM(m*I?Cd&z=li|$-7G!jeOwO;25=992S<m5AICiD8%90T
z7zuw~Iy(4bWgIcVFe&g-@|E0cxhctM!5QXRlG$%>X5MzsmCeV$vtN*Wk9q%cvGzm6
zlGZYQ`Nc~9M~79`)tR-DzwAEIeH!_EZe4SI`^$~5?i-97Prt=)N^Q<<UHq>3ePg@o
zht*Hi&(<B6UGm+oJ@<X`{k8*#!}y1J%19sX9snX}!Sv&sgM$rz0I=Br>|HuI*eO3a
z*sFk(4fq>KkN@xQ6^F(cm~$_2K14li9;XkV|9<@!M&f%8Nam8p00009a7bBm000XU
z000XU0RWnu7ytkil}SWFRCodHT?u#;Rkr@KbUNvfeG_5`YY-wNfPp{+o{ADgGcxep
z5O;8ydCWk3pWowCbe1RjK4lzy;4&jKqk}U-a1=+ud7z@;LLwlFC>S)v1jwFrI_XY2
zop;WyuIf%_F<t2n-Ic!IS6#QNZr{59Ip>~x?x|CCgE~7q5lBOq0>MKUdH^|7ARquk
zTn+*P5D<VNPJtl~KcE2u0x-lWFvQ`9_7{IpP!Ma2Ag8UrfAvN=8&%31Xc{{x^H{Co
z5Bf3s5%3VAbLh&IH$@c|uDl{TIw}*k&TyNplZ9bWQ*bafH>lHMG@8ELxbaVWHf?&T
znHpfF&E_pZ&^rD;1;7qozi0Q$(`V)7{8<+kI>wdbHk%E>!9AN2eO+^{$KB)hHtVU6
z4;0@%KYw`%{kM%aj|)L>`1``u*EM%B_Ep|f_<QCY-I>7iHT~t6&rZsneaT;XVt##n
z3*O&%0=#!k4Gq$@x_XoAC663)d$?Wm=U<vc73yj8s7C4ye?8%O?|sJH9hZdp8`Rda
zV)AbC6#gCsfWR{`**=IUEL^cLOs^a53WvcKmwX4axn1m?g`lRU8J`}Mu9cL0kEW(p
z*S~w@nA^2_^<7tUJ89F|+dbgAd&6H>XTrha?_sgD)BZa!4dhf)W5g$)o+5f!@!6p=
z7>#E6lGpa0z~7?)*juclePn!mT$U>W2F?VqT7?}(LqHHhL#3+DoNXk5_#Pb{(lwSP
zZ<=X|iSbjYeFoatR`H}3=!RdX3qeSTbc>FTPC<i4yKox}dW8l|7rAq|($3Sz1I6E)
z0PuMxfT6vuJ<7S!uFMY$3&G48V-OV;j??9JIDGgtTjt^~;o&;W${mB481A6nNfvWk
zG}X)k!0qh~kGnVT5kF&`L0DN3I5DFSlBQz~XE21}l1oOBfJjYXO~BgEYVC0M-fop2
zDE{6E#$wJW!`duuVRDQ5Vt#39HNO0^LN1EPNWEbGuwmiKId5c0K$-Du1N3`J+mCki
zh@bx#O+t{9<6v}qdxy{j&CM3X#uzYdYMN4g6Ac50s6((}=&dwR{Cx?4=mpk}4k7Lh
zja}8gbI(8i^%;C~;;e!(COSeeKO#br9p##*Gxqi|^SQmA8UnbUF8H0>&5WKoW3<xS
z+uAzF<~aqE$%3S$Xk2`8s#2WoA?J>vT<}n4p!j<e0B)WjD5-NnRG#WdwxQH=@&k}$
zOR8R0dom`Lc=?)q&RcuB7q?Ti)x-Dbg5R}n#wZVUe2wiO-Z*u-^!Q-=uplma_-CN_
z8Dw8c=~1luSn8d}KeqH68(WbvJ_VB|4Oi+hk~;kO6RAfiC;c;{;CF65PXUWd$`m2j
z>ve)Qtntp05&Y$`N~L&mauhNrjZlt#E%Rdnz*4RdA(~WsS0P~4Cker*^h9K3rID79
zAhx!)2_f*-6tD+E@|~5o_HbR*DQEm#fix64W;xPOIEsuwz3>ej`Mg}wlx+M?%^s;7
zt7<_1|D+24j|zb6{d*Duo)R*nQ%A&N`m}UK6}Gim#oV|jr-^I5{&3u6Y!z0&JjK=N
zf~iA{0UNr_&1RH*=FkdaRxmwXu@ih1pW6b!KwO1@&&hNB<m70YDveg=#B4Xn!v>f0
z=VYU~zns|bF>|Ig{pE8Oi&e4q8Sf>;d>$HnJ*g4^2E{@!BWJXj|MK2>t{)#4iCiKM
z_X3_Wd3!22SVWGECF_5t9Wx1ebd<xtk&rM9vu2GY%sRn*f5AMUxid3;-rS5H$2b;{
z-mVQ6!@O~ym}o<<xL6?O&7IIAJ-pEtPt>Ve1IRabo*K&M<b7Y)<TC!a?+aI7*L)G{
zfCa!8K?T~>e+mp(08G`jsI~A7O*rz=A?*I(Ym_y4*Z<vrc-}ABZsXaPWE_55X~M~}
zTIJaLcOUp224n-2zVikie7yn%YnEfm!5#SRe*5FmzkQB}p11+o{Fp|=>BHj<`2EIL
z@XCfeu<hfsczydWJ6bb}o?S)x0g-o5D5qWnbezFoiTRwunk%qdX1+L6i)|mA5jidY
z{0}jolhGWEzbnk=GvPS&`Ekzpd{Qp}jg2@(@Fp)^ip7aWa9kF~U5c$wF9JUgt8d{b
zE$t=?X?=dwwZ3ox^kn_>GtW8G6RGFlFM<@CjE-OtU#5a;0kB%yXw(N%<3n(~<Qsu8
zZ|+#=eP8Q&gdZK`A?MLQF(CmT9+sYjF8fKIN9rETN>sBeG(H{~)Y9EAyo%kT#Rg2j
zpdOnacnjrpoDswQL%S&=xD)LJZ^c?^7~tUKxVSW2U-+UJ`I8c2{Q|sd4FLUcTr-0M
zaqMa26wFKpz7U~s3AlNV<LRJWL`!?_{RP)vp+K58DGni_Hng{J{G%C}6UQspd^w|9
z18~65>^qhrHMbm9<`9gTLcVV_VCkY<jGCK;u}-q#%LEum+wQu00v693hqg8|E+F?e
zr#1I3!kny;lw%R`{p7X)SelGzum&R`J{*=hGlr!kVHo3VG@_P-B6~JTzO@CWjwnQm
zSR&sif;%k~Gk*Hv$KL#GDAqtA4UzOxf>cW$bp+1aV?*4j`n;5NQvl5P$NHC1)DVqF
ze?14Uta}S5dTDmrRR#Fn;tPAZ>c6M&cw`%zt17X5(`x+mXPZPMYENh$xHA{IIn#Q&
z<WuL|wgH?ccYQ_s*J*Sb=&abC^R8TNFm``&#I=ks9J{Fj!m;GPmSYjgeN40tkN)>^
zG}YF_5*3HIuofIEDMeLB1jc8M#;C+D(d52>)gx`#@~i9ZqkAV_+e~x*&R~QFvHtHw
zX=O8P?QIyJ9<S9QfZh~|rm@{l%9`SLom|1Yuu=8~?mQv6f;~cV->Ss9*B|&g;0hMp
z3Alm-uHb+xn7Ts16&!E{`__2XkJh+p1UhOAxPk+&;D9SQ;0g}7f`^<d*f9ujPKWd{
zNwA(ei(>~4p*<oBsnf?}Y_tY_z+HhWM{(ruZlLc|d{fqlC~^zWp3Kn@&XYx@s8428
z+W?HF@nkVpZ73FT0nX?mGjp-|=hGB_f%7#j<?C8Quwu<K-PpY#ZzNXSa*ba=b=wuJ
zprVklr=6G|NiR*v4kFXBPw?d4Dy67aGq*W0e|U1NJv0<}=fG*4{I=9DV5`;u*19?o
z5{B?pJoNC@$d2dNpxSZdpKs&0Z&je7$@!^)U(mv>Mp`Hum_uHM8Ep9TllPO>m-^Cs
zpVwg1bK6i`-w1z*2vDs7WXVaJJHyU=rk@Vk3#W^iKzdl}7D4^3u#E2B8*>%rGlt8u
z5=Bg)^vMF>N2OW-kTeo=C=#;#Uwg6hiz=At%UPznGuZL$9uX3jIcgXzEoL+}ne7De
zePX!NLIZ__1sfvp<d*$}QF`*n@c|qD`l$#Bn=o#aJJp@HvT{duzX6s;924PRO?YW@
z5nghs!8um&p{H&~w#Fia`1v1)e}&ROaMy!>aY5fTR<P*VpVM{qeXU#N9z3DGRUK^<
z`8ISD3xNMj_Il@&dJSMm!=vkNfHgFv_xE4u6KU4aaHMFg_cJ-U=WN?_Kka)L1Nw>(
zUH5HKQ7-^w@TCk-ATqS$+;^2Y-9Yg{p~En8>~LcE&~OCN2SO-y!qgT7qsff0kWR!$
z^D81!lBm$TfXL;}=Y9YJK+SF{!{d*<g%X#sAaChntiD3qVd2a){!IDu!avzV!TF4o
z9xGgkjB%soL!aKOw*koiw_`(h<}=y@MgPFg@A$O|ZKq#6{$@AkGvc*R?Si#UlQDr#
z3~i~w{&Resf;DUNF*}X?fEFV@D0-Vh!)meXqjD6C9^>=}ZDsk}pA}{0WdF3_)n|T5
zFNK7P(SF;zrP#jx9qieE2>F-K@p;gyHGt(@rI_!hEt)McpP}lbFn3v=a0JCAI=-Ld
z^HfmLKw}#PgVO)j-n&3BpR3@}{)WrPilHHGIK3w22T8R6=u<`rMwjnBh~jFy5zt}A
zN81hv!KkMXNNPDnh1mq7H@>uwma1@k3;2!wtQCOj+9tn%uigkWBw{AL|5)BofhX2&
zA+XZ302%fCsUzg9CimQPVv`f;C6O8|{n>ML#6sZcPqU_9<QKRdEq1%Pk+|_c^D$ZZ
zpTp?8<hlYZ8Sj`5DMd7vdH&JsA-;|ya5+f%=&CEBr`QO8Na{5Jql+JdN1k1QvXkYg
zs3w&k5rOovX-K7)MutMt)1dz*eGsClwG%b<Mnp!CZzRsYY@Y!MK;y}yczWw0oN9FR
ze!+NJMke*DMR;s+x~Aozt~D6zfBzb)s_VMoY^H}`58OE$vvRV#ocF~8^#ULQLiXc}
zUQkuE$PY|=DWHMVx2^eKu~LOA6rK$<8jY@}eezhf09Y%}*qLuY{&mxkmF%SY=ox@6
zF~uhVd#X`JdPSa>DPe!$!>g7c<pOq#ONp0~Ns93gcerW+2#u6Nz)abBxO;&=#J<Sv
z@ytFfgdg<)OXiP7*2GlLRC*q2ahR^|-aI}H3;8oRM~>oyleK6R!5=0O9Kit+4(r(6
ziv6QJ8-P(X4Sa3SakRGjFIv?a0G4_jZD3}d!^RD-cH>&cq5?d2jrKkeAp_;!Ur#;&
z9W7Y4e9epUX=T6m-g%gom8l&2YDT>Vpn#D2K2TLOYC9;D1)wkDRn>N#8T3J_^Lk0W
z2GEDo5^3Wxdgdfd9w7&WOIUcVywJ$#^9sz{H)rNATQUdN%*}+3f?}K#TL)6Cfb&`3
z%&Qjw3IaWJ_$1z;4dDsM&%YQ~=42pUgopbkSWmW!9lu+5e2Bl(Hp~!=)psw#l#5d7
z<59t4!9`Er%bRtn7l4p3WRMY9&31sf7Q0{HC$^-K>G(;07G_Pk5PmWfQbk{$>nD;C
z$aX+;iw(co_@<~Qn^p+B=a%_MiWA>XQ&sn1{z<(6(1#*dufHEF>#Fe8m!&8!F2%dw
zHlg}-8UFYJZG<8tdn)d^eHPNC3G-m$^7_440RBMV3*u1l6Q_-M<NrRRMWepLk*_Lo
z-`bbN{R`%2;?^Z|DbJ;SGxcT1^sB{@@yW=TI0_Rc#C0ow;>ckXuK!rmQ$k)#dR$sG
z@^U71!@qOSF|2)@pOpG;Qm+AE#NKTmpy<6aRJ-8I$ex7UR10>zRSMI&Dx4*+aC%oe
z$>ksZdHCl3@33X-u5M#~!F>8s>bP;(@Z1iZ5DQ57E<W`F&{Exn`H?9pD7b1W{*9JP
zOVBg7&1l$Lik;?q)RUhiU1Oo)Yy>(pe>^RmdH=2Rkv1Y;;r0f4a|kUQI?AO7tZbEf
zJ(*E203jiWBR5FKRnt*$=_L9l06hS)bRb+XpPQ(|6)W>G1u?i-W6WoCJgU<Q$tb+@
zD$Jffj@}LJ{)WH{zrK~CgV-P9o6Kh%ssy0+_)+}f)#J#erD6&w@S}9^J6Kv&%Joq)
zVCl;1aO2ciq)r?I|MikfXnOjsHU;7FXSU<j4Ywhj0zZnsHR6e<UbYvt;`zsJK|)+S
z62!-nogR|)RM~kx=eB4z9^a39Htt8sxmx;YG^zT!qj*qZKBH+t(F@!0k$*VC+$MjO
zDHQiTunA>lRkTWYJ9y;~2lKhQP~5|72z2_#^8q&npdI^OKWZnM4)jd~l<t#{4Y+Cf
zJmk?iep_h!z8~!6w6%M3%WgX$8wEf`;a??H_G2r4_wX%<Bi{h4{7*`b`2}z{1;B(K
z1YRBPTX!e&hj*a(pcMU@bm?3ynVpH1H&2p)w`(&X<ooZ-37-u`n^q8?b=dLG_XXa&
z|9vVMfYDf2c#}A8B~Ph8B=-cMr4A=W1b8$?C6oGZsl=!338XY{hCK?{ZdF8R@clrx
z<=e!HAY|AD>xFIKK%PKOm(9u+`!IG4P>PAtq9@Rh0<!HW6ucJiefk6H>JE!{0DuH!
zkK`y|6ZXDM&ju*fYcM2?dkd?0BQd?AvKl9=rI$l^%Bzo%82<d<-V^wnjLl;2>pwp_
z3!t@d`N^j}MPee&>2}gr!FRvB)4o^~UCPYDMfxiI>b@c+MsVI_ZG?n%#SdILF9)yD
z8iBv~&32h6$j=)^`5;_--)1F7aK==Pycf`JwRRcIa&EjD`NGhX@h9M+TM4YCmA;oJ
zrO3=nv3MeD1n(z%`&dZj&7(JU#eehVv~0XE^yJ%^arZ3+;^s6cinJi_LRv*8MlRsh
z{Xp^er2%-zvwii|iPQND<~cxwB;)S&_u$&{D%8_7aQMh%>8YP30yAe!z=De>;j*0J
zN>6b7(K|VAAJyy)=J$-BZpMp7n5{I{+sN@1<}jm{UYm<6az<l#QupT~-a!ux6n{`>
zC)2KLBDKeY!To$ha&qG2BZqfAotPNM^BbQ^H8u4$*;5z(vZ|_v=c1LgH4&aJ8cR)s
zhZ25=_;#ffO9d0sLd30K^&jiDoI6+3R|Htse-FYDw`bL=buUu;*yY6jR@v$9iMtOO
z{Jm)a77X@ba<Dr`c8-aSjUKyf*>%$<P8>f%7edh>l!!{woQDqvAyLn?wOiY*$B%zo
zv32X~pEWczvH$rLZ56cfy6vr`0a$epDA9d}4E`PkfT>4BU?%e$j!CrfB%e1P1~}M{
zuQ8DZRRHLI>|J6XE5CNbPoY`u^Tv~L_DESt0J@K9biv&;RPgs@1TwMtC4bqg&n_U&
z^RqpU@fmCZV8(Krcxd8Db|Y=v9v+%_sqO*ye5%7a4GH|cY5=AL^#T?U?(IAraOf}Z
znfd(s?_l?Sx}{(;kM%5!ES&ry9?r8?uz9NYQ(Ynr1^j&q08@d8z|&jaWMSaE-1`Sx
z2*lKk?$1KN8*2mJGw(g3`l+riN$dE3Q~;P7L<D9yC)d4ncJO{>Cd=wx?7hW&8J3pu
z_e%g|LIn2Oqk!C_wTCQ#s9zKa2tdEcq}@UR0njdQ`-LnZ0R1A9b_)drK)bx{7qWl=
z^ovZ|Eff#{?eex?$N~b;FEVMjP(T2*%iDe-`+v|7m{y$1dn*6{002ovPDHLkV1lnB
B5rhB$

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-mnist-0.1.0.0/tensorflow-mnist.txt b/docs/haddock/tensorflow-mnist-0.1.0.0/tensorflow-mnist.txt
new file mode 100644
index 0000000..5248c19
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-0.1.0.0/tensorflow-mnist.txt
@@ -0,0 +1,41 @@
+-- Hoogle documentation, generated by Haddock
+-- See Hoogle, http://www.haskell.org/hoogle/
+
+
+-- | TensorFlow demo application for learning MNIST model.
+--   
+--   Please see README.md
+@package tensorflow-mnist
+@version 0.1.0.0
+
+
+-- | Paths to test helper files.
+module TensorFlow.Examples.MNIST.TrainedGraph
+
+-- | File containing a Tensorflow serialized proto of MNIST.
+mnistPb :: IO FilePath
+
+-- | Files containing pre-trained weights for MNIST.
+wtsCkpt :: IO ByteString
+
+-- | Files containing pre-trained weights for MNIST.
+biasCkpt :: IO ByteString
+
+module TensorFlow.Examples.MNIST.Parse
+
+-- | Utilities specific to MNIST.
+type MNIST = Vector Word8
+
+-- | Produces a unicode rendering of the MNIST digit sample.
+drawMNIST :: MNIST -> Text
+
+-- | Check's the file's endianess, throwing an error if it's not as
+--   expected.
+checkEndian :: Get ()
+
+-- | Reads an MNIST file and returns a list of samples.
+readMNISTSamples :: FilePath -> IO [MNIST]
+
+-- | Reads a list of MNIST labels from a file and returns them.
+readMNISTLabels :: FilePath -> IO [Word8]
+readMessageFromFileOrDie :: Message m => FilePath -> IO m
diff --git a/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/TensorFlow-Examples-MNIST-InputData.html b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/TensorFlow-Examples-MNIST-InputData.html
new file mode 100644
index 0000000..22288df
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/TensorFlow-Examples-MNIST-InputData.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Examples.MNIST.InputData</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-Examples-MNIST-InputData.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-Examples-MNIST-InputData.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-mnist-input-data-0.1.0.0: Downloader of input data for training MNIST.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>Safe</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.Examples.MNIST.InputData</p></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><a href="#v:trainingImageData">trainingImageData</a> :: <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> <a href="../base-4.8.2.0/System-IO.html#t:FilePath">FilePath</a></li><li class="src short"><a href="#v:trainingLabelData">trainingLabelData</a> :: <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> <a href="../base-4.8.2.0/System-IO.html#t:FilePath">FilePath</a></li><li class="src short"><a href="#v:testImageData">testImageData</a> :: <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> <a href="../base-4.8.2.0/System-IO.html#t:FilePath">FilePath</a></li><li class="src short"><a href="#v:testLabelData">testLabelData</a> :: <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> <a href="../base-4.8.2.0/System-IO.html#t:FilePath">FilePath</a></li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><a name="v:trainingImageData" class="def">trainingImageData</a> :: <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> <a href="../base-4.8.2.0/System-IO.html#t:FilePath">FilePath</a> <a href="src/TensorFlow-Examples-MNIST-InputData.html#trainingImageData" class="link">Source</a></p><div class="doc"><p>Download the files containing the canonical MNIST samples and labels.</p></div></div><div class="top"><p class="src"><a name="v:trainingLabelData" class="def">trainingLabelData</a> :: <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> <a href="../base-4.8.2.0/System-IO.html#t:FilePath">FilePath</a> <a href="src/TensorFlow-Examples-MNIST-InputData.html#trainingLabelData" class="link">Source</a></p><div class="doc"><p>Download the files containing the canonical MNIST samples and labels.</p></div></div><div class="top"><p class="src"><a name="v:testImageData" class="def">testImageData</a> :: <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> <a href="../base-4.8.2.0/System-IO.html#t:FilePath">FilePath</a> <a href="src/TensorFlow-Examples-MNIST-InputData.html#testImageData" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:testLabelData" class="def">testLabelData</a> :: <a href="../base-4.8.2.0/System-IO.html#t:IO">IO</a> <a href="../base-4.8.2.0/System-IO.html#t:FilePath">FilePath</a> <a href="src/TensorFlow-Examples-MNIST-InputData.html#testLabelData" class="link">Source</a></p></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/doc-index.html b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/doc-index.html
new file mode 100644
index 0000000..0fe16fa
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/doc-index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-mnist-input-data-0.1.0.0: Downloader of input data for training MNIST. (Index)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-mnist-input-data-0.1.0.0: Downloader of input data for training MNIST.</p></div><div id="content"><div id="index"><p class="caption">Index</p><table><tr><td class="src">testImageData</td><td class="module"><a href="TensorFlow-Examples-MNIST-InputData.html#v:testImageData">TensorFlow.Examples.MNIST.InputData</a></td></tr><tr><td class="src">testLabelData</td><td class="module"><a href="TensorFlow-Examples-MNIST-InputData.html#v:testLabelData">TensorFlow.Examples.MNIST.InputData</a></td></tr><tr><td class="src">trainingImageData</td><td class="module"><a href="TensorFlow-Examples-MNIST-InputData.html#v:trainingImageData">TensorFlow.Examples.MNIST.InputData</a></td></tr><tr><td class="src">trainingLabelData</td><td class="module"><a href="TensorFlow-Examples-MNIST-InputData.html#v:trainingLabelData">TensorFlow.Examples.MNIST.InputData</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/frames.html b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/frames.html
new file mode 100644
index 0000000..1b4e38d
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/frames.html
@@ -0,0 +1,30 @@
+<!DOCTYPE html 
+     PUBLIC "-//W3C//DTD XHTML 1.0 Frameset//EN"
+     "http://www.w3.org/TR/xhtml1/DTD/xhtml1-frameset.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<title></title>
+<script src="haddock-util.js" type="text/javascript"></script>
+<script type="text/javascript"><!--
+/*
+
+  The synopsis frame needs to be updated using javascript, so we hide
+  it by default and only show it if javascript is enabled.
+
+  TODO: provide some means to disable it.
+*/
+function load() {
+  var d = document.getElementById("inner-fs");
+  d.rows = "50%,50%";
+  postReframe();
+}
+--></script>
+</head>
+<frameset id="outer-fs" cols="25%,75%" onload="load()">
+  <frameset id="inner-fs" rows="100%,0%">
+    <frame src="index-frames.html" name="modules" />
+    <frame src="" name="synopsis" />
+  </frameset>
+  <frame src="index.html" name="main" />
+</frameset>
+</html>
diff --git a/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/haddock-util.js b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/haddock-util.js
new file mode 100644
index 0000000..9a6fccf
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/haddock-util.js
@@ -0,0 +1,344 @@
+// Haddock JavaScript utilities
+
+var rspace = /\s\s+/g,
+	  rtrim = /^\s+|\s+$/g;
+
+function spaced(s) { return (" " + s + " ").replace(rspace, " "); }
+function trim(s)   { return s.replace(rtrim, ""); }
+
+function hasClass(elem, value) {
+  var className = spaced(elem.className || "");
+  return className.indexOf( " " + value + " " ) >= 0;
+}
+
+function addClass(elem, value) {
+  var className = spaced(elem.className || "");
+  if ( className.indexOf( " " + value + " " ) < 0 ) {
+    elem.className = trim(className + " " + value);
+  }
+}
+
+function removeClass(elem, value) {
+  var className = spaced(elem.className || "");
+  className = className.replace(" " + value + " ", " ");
+  elem.className = trim(className);
+}
+
+function toggleClass(elem, valueOn, valueOff, bool) {
+  if (bool == null) { bool = ! hasClass(elem, valueOn); }
+  if (bool) {
+    removeClass(elem, valueOff);
+    addClass(elem, valueOn);
+  }
+  else {
+    removeClass(elem, valueOn);
+    addClass(elem, valueOff);
+  }
+  return bool;
+}
+
+
+function makeClassToggle(valueOn, valueOff)
+{
+  return function(elem, bool) {
+    return toggleClass(elem, valueOn, valueOff, bool);
+  }
+}
+
+toggleShow = makeClassToggle("show", "hide");
+toggleCollapser = makeClassToggle("collapser", "expander");
+
+function toggleSection(id)
+{
+  var b = toggleShow(document.getElementById("section." + id));
+  toggleCollapser(document.getElementById("control." + id), b);
+  rememberCollapsed(id, b);
+  return b;
+}
+
+var collapsed = {};
+function rememberCollapsed(id, b)
+{
+  if(b)
+    delete collapsed[id]
+  else
+    collapsed[id] = null;
+
+  var sections = [];
+  for(var i in collapsed)
+  {
+    if(collapsed.hasOwnProperty(i))
+      sections.push(i);
+  }
+  // cookie specific to this page; don't use setCookie which sets path=/
+  document.cookie = "collapsed=" + escape(sections.join('+'));
+}
+
+function restoreCollapsed()
+{
+  var cookie = getCookie("collapsed");
+  if(!cookie)
+    return;
+
+  var ids = cookie.split('+');
+  for(var i in ids)
+  {
+    if(document.getElementById("section." + ids[i]))
+      toggleSection(ids[i]);
+  }
+}
+
+function setCookie(name, value) {
+  document.cookie = name + "=" + escape(value) + ";path=/;";
+}
+
+function clearCookie(name) {
+  document.cookie = name + "=;path=/;expires=Thu, 01-Jan-1970 00:00:01 GMT;";
+}
+
+function getCookie(name) {
+  var nameEQ = name + "=";
+  var ca = document.cookie.split(';');
+  for(var i=0;i < ca.length;i++) {
+    var c = ca[i];
+    while (c.charAt(0)==' ') c = c.substring(1,c.length);
+    if (c.indexOf(nameEQ) == 0) {
+      return unescape(c.substring(nameEQ.length,c.length));
+    }
+  }
+  return null;
+}
+
+
+
+var max_results = 75; // 50 is not enough to search for map in the base libraries
+var shown_range = null;
+var last_search = null;
+
+function quick_search()
+{
+    perform_search(false);
+}
+
+function full_search()
+{
+    perform_search(true);
+}
+
+
+function perform_search(full)
+{
+    var text = document.getElementById("searchbox").value.toLowerCase();
+    if (text == last_search && !full) return;
+    last_search = text;
+    
+    var table = document.getElementById("indexlist");
+    var status = document.getElementById("searchmsg");
+    var children = table.firstChild.childNodes;
+    
+    // first figure out the first node with the prefix
+    var first = bisect(-1);
+    var last = (first == -1 ? -1 : bisect(1));
+
+    if (first == -1)
+    {
+        table.className = "";
+        status.innerHTML = "No results found, displaying all";
+    }
+    else if (first == 0 && last == children.length - 1)
+    {
+        table.className = "";
+        status.innerHTML = "";
+    }
+    else if (last - first >= max_results && !full)
+    {
+        table.className = "";
+        status.innerHTML = "More than " + max_results + ", press Search to display";
+    }
+    else
+    {
+        // decide what you need to clear/show
+        if (shown_range)
+            setclass(shown_range[0], shown_range[1], "indexrow");
+        setclass(first, last, "indexshow");
+        shown_range = [first, last];
+        table.className = "indexsearch";
+        status.innerHTML = "";
+    }
+
+    
+    function setclass(first, last, status)
+    {
+        for (var i = first; i <= last; i++)
+        {
+            children[i].className = status;
+        }
+    }
+    
+    
+    // do a binary search, treating 0 as ...
+    // return either -1 (no 0's found) or location of most far match
+    function bisect(dir)
+    {
+        var first = 0, finish = children.length - 1;
+        var mid, success = false;
+
+        while (finish - first > 3)
+        {
+            mid = Math.floor((finish + first) / 2);
+
+            var i = checkitem(mid);
+            if (i == 0) i = dir;
+            if (i == -1)
+                finish = mid;
+            else
+                first = mid;
+        }
+        var a = (dir == 1 ? first : finish);
+        var b = (dir == 1 ? finish : first);
+        for (var i = b; i != a - dir; i -= dir)
+        {
+            if (checkitem(i) == 0) return i;
+        }
+        return -1;
+    }    
+    
+    
+    // from an index, decide what the result is
+    // 0 = match, -1 is lower, 1 is higher
+    function checkitem(i)
+    {
+        var s = getitem(i).toLowerCase().substr(0, text.length);
+        if (s == text) return 0;
+        else return (s > text ? -1 : 1);
+    }
+    
+    
+    // from an index, get its string
+    // this abstracts over alternates
+    function getitem(i)
+    {
+        for ( ; i >= 0; i--)
+        {
+            var s = children[i].firstChild.firstChild.data;
+            if (s.indexOf(' ') == -1)
+                return s;
+        }
+        return ""; // should never be reached
+    }
+}
+
+function setSynopsis(filename) {
+    if (parent.window.synopsis) {
+        if (parent.window.synopsis.location.replace) {
+            // In Firefox this avoids adding the change to the history.
+            parent.window.synopsis.location.replace(filename);
+        } else {
+            parent.window.synopsis.location = filename;
+        }
+    }
+}
+
+function addMenuItem(html) {
+  var menu = document.getElementById("page-menu");
+  if (menu) {
+    var btn = menu.firstChild.cloneNode(false);
+    btn.innerHTML = html;
+    menu.appendChild(btn);
+  }
+}
+
+function adjustForFrames() {
+  var bodyCls;
+  
+  if (parent.location.href == window.location.href) {
+    // not in frames, so add Frames button
+    addMenuItem("<a href='#' onclick='reframe();return true;'>Frames</a>");
+    bodyCls = "no-frame";
+  }
+  else {
+    bodyCls = "in-frame";
+  }
+  addClass(document.body, bodyCls);
+}
+
+function reframe() {
+  setCookie("haddock-reframe", document.URL);
+  window.location = "frames.html";
+}
+
+function postReframe() {
+  var s = getCookie("haddock-reframe");
+  if (s) {
+    parent.window.main.location = s;
+    clearCookie("haddock-reframe");
+  }
+}
+
+function styles() {
+  var i, a, es = document.getElementsByTagName("link"), rs = [];
+  for (i = 0; a = es[i]; i++) {
+    if(a.rel.indexOf("style") != -1 && a.title) {
+      rs.push(a);
+    }
+  }
+  return rs;
+}
+
+function addStyleMenu() {
+  var as = styles();
+  var i, a, btns = "";
+  for(i=0; a = as[i]; i++) {
+    btns += "<li><a href='#' onclick=\"setActiveStyleSheet('"
+      + a.title + "'); return false;\">"
+      + a.title + "</a></li>"
+  }
+  if (as.length > 1) {
+    var h = "<div id='style-menu-holder'>"
+      + "<a href='#' onclick='styleMenu(); return false;'>Style &#9662;</a>"
+      + "<ul id='style-menu' class='hide'>" + btns + "</ul>"
+      + "</div>";
+    addMenuItem(h);
+  }
+}
+
+function setActiveStyleSheet(title) {
+  var as = styles();
+  var i, a, found;
+  for(i=0; a = as[i]; i++) {
+    a.disabled = true;
+          // need to do this always, some browsers are edge triggered
+    if(a.title == title) {
+      found = a;
+    }
+  }
+  if (found) {
+    found.disabled = false;
+    setCookie("haddock-style", title);
+  }
+  else {
+    as[0].disabled = false;
+    clearCookie("haddock-style");
+  }
+  styleMenu(false);
+}
+
+function resetStyle() {
+  var s = getCookie("haddock-style");
+  if (s) setActiveStyleSheet(s);
+}
+
+
+function styleMenu(show) {
+  var m = document.getElementById('style-menu');
+  if (m) toggleShow(m, show);
+}
+
+
+function pageLoad() {
+  addStyleMenu();
+  adjustForFrames();
+  resetStyle();
+  restoreCollapsed();
+}
+
diff --git a/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/hslogo-16.png b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/hslogo-16.png
new file mode 100644
index 0000000000000000000000000000000000000000..0ff8579fbd897417b0d6dad6e920f8882138a7c0
GIT binary patch
literal 1684
zcmV;F25b3=P)<h;3K|Lk000e1NJLTq000*N000mO1^@s6{V(Jg0008+X+uL$Nkc;*
zP;zf(X>4Tx0C)j~RL^S@K@|QrZmG~B2wH0nvUrdpNm;9CMbtL^5n^i$+aI<?kYqDS
z)^0ZI?k2URh~(nYD&9ryp$Gqf9(wR(FG_`4MJV2@C}N=(E%D86Otz)!z|Ma2-Z$@k
zZ+71R4RX>n^?(HA4aZWV5ov6ELTdbo0FI&wK{O>*+w4vx20?>!`FrQsdJlnHR>OPy
zcd~b_n$otK2Za4V;76L-DzNVtaSB-y0*E}{p()372;bw_^6ZZ}PI-92wGS&j#91PI
zKs7DSe@(bk%_Y-7gGe}(^>I=@oY#w#*Bu9GZf3^F5WP>3rn}7Ut74&?PWBFvy`A)a
zPP5)V!Xd&78LdA?xQ(9mjMYElVd13a#D+Z_7&Y|xU=_C-srWU*6kiZcC!$nw*)9$7
zn6CX+@=AhmkT}X@VSsa5NKe;HZuq)~1$`#h6R+ZTR#D-3j}vF!)ZOnz+5)dI4jl{{
z44Mr{P!L4~VVJN`K!!XTF*LGrKO?IK8<Tr7btG!LbYeuYL3=jbJ-1P$-8}v%B5{;M
zwFr{@LH;VQ$xr2Z`O93e*jD$Ht(%&<^58qg<(at}9@o>z<8w`3e3jI8lUGNUta*C8
zn(P`s>{pjD=7Kek#B;Fw@hxAK%$F&Q6vg9J^Xf~4by_hu-=A!MJ3Znq&n~srbFGPs
zH&&aMXZ>nO`|hf|ljc?VPhR!${AbO?W8x_>CU%PFA&Hm8F7cAsOREdwU~R_;ot1_u
z(ruCYB-LPGn!NQdT|ZlRy+(fw^-+`=%+gee_kY4FWHg<*4sZI8+sFJD<oAl_pC|$^
zY~aY5x@}W&?+~G7rEYVs0vEs0eekw!YomTR`~+A$s}`+NHJ>270UUORdLHO0nA4V)
z%{fwsET5CQ>B?eK%uw4yQc~9?*JVo<vzb+5>2}ze(;aRcp*ceL#HUJSllrgm5wQKR
zQu+C;QrUh^8rFfA`ftFz{YAidi-`aL010qNS#tmY4c7nw4c7reD4Tcy00T@(L_t(I
z5sj2vNEA^R$7gqDc6T=2^@fUA2(c`MltuL5<|<Z1YDGjBEr^N`3VJpQ$`C4&9(*Z6
z=|*B$y;-1^F6vXRAd7+(;)77Qps2euv)AeWmj>KW>RWz$&YbU@|M|{$E*8Tu-Ux!w
z1Y*Dr&Ubfr&v-nZaaB{3ilRumrjPmk{sZvQEWlW+{o~IH|8)=s6c#X9S5s5d%<Agu
zgM$153Ttk++ZPA~ZqgqN1|^pNV-|?9-zo48n9+Yiy$*su5czgp*VFj6!1h&DRZR>J
z4@)&QH5|xQY-)^L1n<EM!01(Q6kM8}ojndgRD~2KVIa@>0pTRu0Lx9`08YTjTwn^6
z0;b1+aQ@)n;Em$q;=7BBi)v0zj&o^g>0Whp^_^5IbxIUP8C@y9;R?*Ouu}rmfxbU=
zwtWVNke-m!=`7bYEhWpcI5#)9qp`8E0lr6IQ)ARL3Ui}Af@grj8aN1=r>Cb+prlzO
zNfJs*N_tUm2ZL<f%*@E?H`Ak_J`Q#Qri<;K{{H@Bg@uKWS5{WE+}vCx91geD*4DNm
zJ?D^~dGz7lWtbWH`o<0O#|{R6h9M*GMpjnV?Y_RgatOb~1le#ikPKTNIn%<;=>%5*
zPmL2??da$TR904gL(VDAQ<BE%k@5ZA>-Fv_Dk}Pdw*4T(%*f4MKLRg=4ekMjhe2mW
zMFsBwg%ftWT}0kxRaIk1k7qJ8*#cKB;Ft{i`zVIs-Nqge;!!Ld7#O&Qqu7e0sJmP)
z$MW*>L$vSB&dxp@iA3U9fo)-7!Czlr{|o7<LYh`Fd@X2-b0fIa*Wdy$({Zt@t7`&L
zhoaGF!l5{1P1Qm;<GaSj#=Oa7atf1JZ8n=F9*@uC?b(X48O9kPoH~h;!o2=WM?ZLl
z$Lj;4K~)fAdW@^N%nzT>Hv{1oyg3xsu%gn@(b1>$;SM-ZaQ`HV=V0s;lr%d8bd;xY
zGw<i+<$17ay36gdrqk(Lxc1|{y}dSeM5OIV(FjhBg5#OM<!Z*8(@<7crlVw+B-8R1
zl{3J6w9fYQ^vojauaRgZ=~SE&CK!R3pCe!v3>Nvm3=Iu=tyXIgtJnf@A(2S@M140N
ew{UA~tMxaJq;$xaSSi*30000<MNUMnLSTa0p)&^n

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/index-frames.html b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/index-frames.html
new file mode 100644
index 0000000..7542511
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/index-frames.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-mnist-input-data-0.1.0.0: Downloader of input data for training MNIST.</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-list"><p class="caption">Modules</p><ul><li class="module"><a href="TensorFlow-Examples-MNIST-InputData.html" target="main">TensorFlow.Examples.MNIST.InputData</a></li></ul></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/index.html b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/index.html
new file mode 100644
index 0000000..c041e18
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-mnist-input-data-0.1.0.0: Downloader of input data for training MNIST.</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-mnist-input-data-0.1.0.0: Downloader of input data for training MNIST.</p></div><div id="content"><div id="description"><h1>tensorflow-mnist-input-data-0.1.0.0: Downloader of input data for training MNIST.</h1><div class="doc"><p>Please see README.md</p></div></div><div id="module-list"><p class="caption">Modules</p><ul><li><span id="control.n.1" class="module collapser" onclick="toggleSection('n.1')">TensorFlow</span><ul id="section.n.1" class="show"><li><span id="control.n.1.1" class="module collapser" onclick="toggleSection('n.1.1')">Examples</span><ul id="section.n.1.1" class="show"><li><span id="control.n.1.1.1" class="module collapser" onclick="toggleSection('n.1.1.1')">MNIST</span><ul id="section.n.1.1.1" class="show"><li><span class="module"><a href="TensorFlow-Examples-MNIST-InputData.html">TensorFlow.Examples.MNIST.InputData</a></span></li></ul></li></ul></li></ul></li></ul></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/mini_TensorFlow-Examples-MNIST-InputData.html b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/mini_TensorFlow-Examples-MNIST-InputData.html
new file mode 100644
index 0000000..d719ceb
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/mini_TensorFlow-Examples-MNIST-InputData.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Examples.MNIST.InputData</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.Examples.MNIST.InputData</p></div><div id="interface"><div class="top"><p class="src"><a href="TensorFlow-Examples-MNIST-InputData.html#v:trainingImageData" target="main">trainingImageData</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Examples-MNIST-InputData.html#v:trainingLabelData" target="main">trainingLabelData</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Examples-MNIST-InputData.html#v:testImageData" target="main">testImageData</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Examples-MNIST-InputData.html#v:testLabelData" target="main">testLabelData</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/minus.gif b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/minus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..1deac2fe1a42e35b994f1b855488f392c50f6a89
GIT binary patch
literal 56
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS53HJ0a;6C;`j@u!^I(6E|
Jh4LZ{)&P#)4$%Mr

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/ocean.css b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/ocean.css
new file mode 100644
index 0000000..1110b40
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/ocean.css
@@ -0,0 +1,600 @@
+/* @group Fundamentals */
+
+* { margin: 0; padding: 0 }
+
+/* Is this portable? */
+html {
+  background-color: white;
+  width: 100%;
+  height: 100%;
+}
+
+body {
+  background: white;
+  color: black;
+  text-align: left;
+  min-height: 100%;
+  position: relative;
+}
+
+p {
+  margin: 0.8em 0;
+}
+
+ul, ol {
+  margin: 0.8em 0 0.8em 2em;
+}
+
+dl {
+  margin: 0.8em 0;
+}
+
+dt {
+  font-weight: bold;
+}
+dd {
+  margin-left: 2em;
+}
+
+a { text-decoration: none; }
+a[href]:link { color: rgb(196,69,29); }
+a[href]:visited { color: rgb(171,105,84); }
+a[href]:hover { text-decoration:underline; }
+
+/* @end */
+
+/* @group Fonts & Sizes */
+
+/* Basic technique & IE workarounds from YUI 3
+   For reasons, see:
+      http://yui.yahooapis.com/3.1.1/build/cssfonts/fonts.css
+ */
+
+body {
+	font:13px/1.4 sans-serif;
+	*font-size:small; /* for IE */
+	*font:x-small; /* for IE in quirks mode */
+}
+
+h1 { font-size: 146.5%; /* 19pt */ }
+h2 { font-size: 131%;   /* 17pt */ }
+h3 { font-size: 116%;   /* 15pt */ }
+h4 { font-size: 100%;   /* 13pt */ }
+h5 { font-size: 100%;   /* 13pt */ }
+
+select, input, button, textarea {
+	font:99% sans-serif;
+}
+
+table {
+	font-size:inherit;
+	font:100%;
+}
+
+pre, code, kbd, samp, tt, .src {
+	font-family:monospace;
+	*font-size:108%;
+	line-height: 124%;
+}
+
+.links, .link {
+  font-size: 85%; /* 11pt */
+}
+
+#module-header .caption {
+  font-size: 182%; /* 24pt */
+}
+
+.info  {
+  font-size: 85%; /* 11pt */
+}
+
+#table-of-contents, #synopsis  {
+  /* font-size: 85%; /* 11pt */
+}
+
+
+/* @end */
+
+/* @group Common */
+
+.caption, h1, h2, h3, h4, h5, h6 {
+  font-weight: bold;
+  color: rgb(78,98,114);
+  margin: 0.8em 0 0.4em;
+}
+
+* + h1, * + h2, * + h3, * + h4, * + h5, * + h6 {
+  margin-top: 2em;
+}
+
+h1 + h2, h2 + h3, h3 + h4, h4 + h5, h5 + h6 {
+  margin-top: inherit;
+}
+
+ul.links {
+  list-style: none;
+  text-align: left;
+  float: right;
+  display: inline-table;
+  margin: 0 0 0 1em;
+}
+
+ul.links li {
+  display: inline;
+  border-left: 1px solid #d5d5d5;
+  white-space: nowrap;
+  padding: 0;
+}
+
+ul.links li a {
+  padding: 0.2em 0.5em;
+}
+
+.hide { display: none; }
+.show { display: inherit; }
+.clear { clear: both; }
+
+.collapser {
+  background-image: url(minus.gif);
+  background-repeat: no-repeat;
+}
+.expander {
+  background-image: url(plus.gif);
+  background-repeat: no-repeat;
+}
+p.caption.collapser,
+p.caption.expander {
+  background-position: 0 0.4em;
+}
+.collapser, .expander {
+  padding-left: 14px;
+  margin-left: -14px;
+  cursor: pointer;
+}
+
+pre {
+  padding: 0.25em;
+  margin: 0.8em 0;
+  background: rgb(229,237,244);
+  overflow: auto;
+  border-bottom: 0.25em solid white;
+  /* white border adds some space below the box to compensate
+     for visual extra space that paragraphs have between baseline
+     and the bounding box */
+}
+
+.src {
+  background: #f0f0f0;
+  padding: 0.2em 0.5em;
+}
+
+.keyword { font-weight: normal; }
+.def { font-weight: bold; }
+
+
+/* @end */
+
+/* @group Page Structure */
+
+#content {
+  margin: 0 auto;
+  padding: 0 2em 6em;
+}
+
+#package-header {
+  background: rgb(41,56,69);
+  border-top: 5px solid rgb(78,98,114);
+  color: #ddd;
+  padding: 0.2em;
+  position: relative;
+  text-align: left;
+}
+
+#package-header .caption {
+  background: url(hslogo-16.png) no-repeat 0em;
+  color: white;
+  margin: 0 2em;
+  font-weight: normal;
+  font-style: normal;
+  padding-left: 2em;
+}
+
+#package-header a:link, #package-header a:visited { color: white; }
+#package-header a:hover { background: rgb(78,98,114); }
+
+#module-header .caption {
+  color: rgb(78,98,114);
+  font-weight: bold;
+  border-bottom: 1px solid #ddd;
+}
+
+table.info {
+  float: right;
+  padding: 0.5em 1em;
+  border: 1px solid #ddd;
+  color: rgb(78,98,114);
+  background-color: #fff;
+  max-width: 40%;
+  border-spacing: 0;
+  position: relative;
+  top: -0.5em;
+  margin: 0 0 0 2em;
+}
+
+.info th {
+	padding: 0 1em 0 0;
+}
+
+div#style-menu-holder {
+  position: relative;
+  z-index: 2;
+  display: inline;
+}
+
+#style-menu {
+  position: absolute;
+  z-index: 1;
+  overflow: visible;
+  background: #374c5e;
+  margin: 0;
+  text-align: center;
+  right: 0;
+  padding: 0;
+  top: 1.25em;
+}
+
+#style-menu li {
+	display: list-item;
+	border-style: none;
+	margin: 0;
+	padding: 0;
+	color: #000;
+	list-style-type: none;
+}
+
+#style-menu li + li {
+	border-top: 1px solid #919191;
+}
+
+#style-menu a {
+  width: 6em;
+  padding: 3px;
+  display: block;
+}
+
+#footer {
+  background: #ddd;
+  border-top: 1px solid #aaa;
+  padding: 0.5em 0;
+  color: #666;
+  text-align: center;
+  position: absolute;
+  bottom: 0;
+  width: 100%;
+  height: 3em;
+}
+
+/* @end */
+
+/* @group Front Matter */
+
+#table-of-contents {
+  float: right;
+  clear: right;
+  background: #faf9dc;
+  border: 1px solid #d8d7ad;
+  padding: 0.5em 1em;
+  max-width: 20em;
+  margin: 0.5em 0 1em 1em;
+}
+
+#table-of-contents .caption {
+  text-align: center;
+  margin: 0;
+}
+
+#table-of-contents ul {
+  list-style: none;
+  margin: 0;
+}
+
+#table-of-contents ul ul {
+  margin-left: 2em;
+}
+
+#description .caption {
+  display: none;
+}
+
+#synopsis {
+  display: none;
+}
+
+.no-frame #synopsis {
+  display: block;
+  position: fixed;
+  right: 0;
+  height: 80%;
+  top: 10%;
+  padding: 0;
+  max-width: 75%;
+}
+
+#synopsis .caption {
+  float: left;
+  width: 29px;
+  color: rgba(255,255,255,0);
+  height: 110px;
+  margin: 0;
+  font-size: 1px;
+  padding: 0;
+}
+
+#synopsis p.caption.collapser {
+  background: url(synopsis.png) no-repeat -64px -8px;
+}
+
+#synopsis p.caption.expander {
+  background: url(synopsis.png) no-repeat 0px -8px;
+}
+
+#synopsis ul {
+  height: 100%;
+  overflow: auto;
+  padding: 0.5em;
+  margin: 0;
+}
+
+#synopsis ul ul {
+  overflow: hidden;
+}
+
+#synopsis ul,
+#synopsis ul li.src {
+  background-color: #faf9dc;
+  white-space: nowrap;
+  list-style: none;
+  margin-left: 0;
+}
+
+/* @end */
+
+/* @group Main Content */
+
+#interface div.top { margin: 2em 0; }
+#interface h1 + div.top,
+#interface h2 + div.top,
+#interface h3 + div.top,
+#interface h4 + div.top,
+#interface h5 + div.top {
+ 	margin-top: 1em;
+}
+#interface p.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface td.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface span.fixity {
+  color: #919191;
+  border-left: 1px solid #919191;
+  padding: 0.2em 0.5em 0.2em 0.5em;
+  margin: 0 -1em 0 1em;
+}
+
+#interface span.rightedge {
+  border-left: 1px solid #919191;
+  padding: 0.2em 0 0.2em 0;
+  margin: 0 0 0 1em;
+}
+
+#interface table { border-spacing: 2px; }
+#interface td {
+  vertical-align: top;
+  padding-left: 0.5em;
+}
+#interface td.src {
+  white-space: nowrap;
+}
+#interface td.doc p {
+  margin: 0;
+}
+#interface td.doc p + p {
+  margin-top: 0.8em;
+}
+
+.clearfix:after {
+  clear: both;
+  content: " ";
+  display: block;
+  height: 0;
+  visibility: hidden;
+}
+
+.subs dl {
+  margin: 0;
+}
+
+.subs dt {
+  float: left;
+  clear: left;
+  display: block;
+  margin: 1px 0;
+}
+
+.subs dd {
+  float: right;
+  width: 90%;
+  display: block;
+  padding-left: 0.5em;
+  margin-bottom: 0.5em;
+}
+
+.subs dd.empty {
+  display: none;
+}
+
+.subs dd p {
+  margin: 0;
+}
+
+/* Render short-style data instances */
+.inst ul {
+  height: 100%;
+  padding: 0.5em;
+  margin: 0;
+}
+
+.inst, .inst li {
+  list-style: none;
+  margin-left: 1em;
+}
+
+/* Workaround for bug in Firefox (issue #384) */
+.inst-left {
+  float: left;
+}
+
+.top p.src {
+  border-top: 1px solid #ccc;
+}
+
+.subs, .doc {
+  /* use this selector for one level of indent */
+  padding-left: 2em;
+}
+
+.warning {
+  color: red;
+}
+
+.arguments {
+  margin-top: -0.4em;
+}
+.arguments .caption {
+  display: none;
+}
+
+.fields { padding-left: 1em; }
+
+.fields .caption { display: none; }
+
+.fields p { margin: 0 0; }
+
+/* this seems bulky to me
+.methods, .constructors {
+  background: #f8f8f8;
+  border: 1px solid #eee;
+}
+*/
+
+/* @end */
+
+/* @group Auxillary Pages */
+
+
+.extension-list {
+    list-style-type: none;
+    margin-left: 0;
+}
+
+#mini {
+  margin: 0 auto;
+  padding: 0 1em 1em;
+}
+
+#mini > * {
+  font-size: 93%; /* 12pt */
+}
+
+#mini #module-list .caption,
+#mini #module-header .caption {
+  font-size: 125%; /* 15pt */
+}
+
+#mini #interface h1,
+#mini #interface h2,
+#mini #interface h3,
+#mini #interface h4 {
+  font-size: 109%; /* 13pt */
+  margin: 1em 0 0;
+}
+
+#mini #interface .top,
+#mini #interface .src {
+  margin: 0;
+}
+
+#mini #module-list ul {
+  list-style: none;
+  margin: 0;
+}
+
+#alphabet ul {
+	list-style: none;
+	padding: 0;
+	margin: 0.5em 0 0;
+	text-align: center;
+}
+
+#alphabet li {
+	display: inline;
+	margin: 0 0.25em;
+}
+
+#alphabet a {
+	font-weight: bold;
+}
+
+#index .caption,
+#module-list .caption { font-size: 131%; /* 17pt */ }
+
+#index table {
+  margin-left: 2em;
+}
+
+#index .src {
+  font-weight: bold;
+}
+#index .alt {
+  font-size: 77%; /* 10pt */
+  font-style: italic;
+  padding-left: 2em;
+}
+
+#index td + td {
+  padding-left: 1em;
+}
+
+#module-list ul {
+  list-style: none;
+  margin: 0 0 0 2em;
+}
+
+#module-list li {
+  clear: right;
+}
+
+#module-list span.collapser,
+#module-list span.expander {
+  background-position: 0 0.3em;
+}
+
+#module-list .package {
+  float: right;
+}
+
+/* @end */
diff --git a/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/plus.gif b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/plus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..2d15c14173d23f664b955cd24f51c82f5f09d91d
GIT binary patch
literal 59
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS5iS+a@;NG@rPSdH*>gbBX
M^XE!9f*2UA0nx1yDgXcg

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/src/Paths_tensorflow_mnist_input_data.html b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/src/Paths_tensorflow_mnist_input_data.html
new file mode 100644
index 0000000..9fead8c
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/src/Paths_tensorflow_mnist_input_data.html
@@ -0,0 +1,46 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Paths_tensorflow_mnist_input_data.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Paths_tensorflow_mnist_input_data</span> <span class='hs-layout'>(</span>
+<a name="line-2"></a>    <span class='hs-varid'>version</span><span class='hs-layout'>,</span>
+<a name="line-3"></a>    <span class='hs-varid'>getBinDir</span><span class='hs-layout'>,</span> <span class='hs-varid'>getLibDir</span><span class='hs-layout'>,</span> <span class='hs-varid'>getDataDir</span><span class='hs-layout'>,</span> <span class='hs-varid'>getLibexecDir</span><span class='hs-layout'>,</span>
+<a name="line-4"></a>    <span class='hs-varid'>getDataFileName</span><span class='hs-layout'>,</span> <span class='hs-varid'>getSysconfDir</span>
+<a name="line-5"></a>  <span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-6"></a>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Exception</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Exception</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Version</span> <span class='hs-layout'>(</span><span class='hs-conid'>Version</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>System</span><span class='hs-varop'>.</span><span class='hs-conid'>Environment</span> <span class='hs-layout'>(</span><span class='hs-varid'>getEnv</span><span class='hs-layout'>)</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Prelude</span>
+<a name="line-11"></a>
+<a name="line-12"></a><a name="catchIO"></a><span class='hs-definition'>catchIO</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Exception</span><span class='hs-varop'>.</span><span class='hs-conid'>IOException</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-varid'>a</span>
+<a name="line-13"></a><span class='hs-definition'>catchIO</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Exception</span><span class='hs-varop'>.</span><span class='hs-varid'>catch</span>
+<a name="line-14"></a>
+<a name="line-15"></a><a name="version"></a><span class='hs-definition'>version</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Version</span>
+<a name="line-16"></a><span class='hs-definition'>version</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Version</span> <span class='hs-keyglyph'>[</span><span class='hs-num'>0</span><span class='hs-layout'>,</span><span class='hs-num'>1</span><span class='hs-layout'>,</span><span class='hs-num'>0</span><span class='hs-layout'>,</span><span class='hs-num'>0</span><span class='hs-keyglyph'>]</span> <span class='hs-conid'>[]</span>
+<a name="line-17"></a><a name="bindir"></a><span class='hs-definition'>bindir</span><span class='hs-layout'>,</span> <span class='hs-varid'>libdir</span><span class='hs-layout'>,</span> <span class='hs-varid'>datadir</span><span class='hs-layout'>,</span> <span class='hs-varid'>libexecdir</span><span class='hs-layout'>,</span> <span class='hs-varid'>sysconfdir</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>FilePath</span>
+<a name="line-18"></a>
+<a name="line-19"></a><span class='hs-definition'>bindir</span>     <span class='hs-keyglyph'>=</span> <span class='hs-str'>"/home/gnezdo/tensorflow-haskell/.stack-work/install/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/lts-6.2/7.10.3/bin"</span>
+<a name="line-20"></a><a name="libdir"></a><span class='hs-definition'>libdir</span>     <span class='hs-keyglyph'>=</span> <span class='hs-str'>"/home/gnezdo/tensorflow-haskell/.stack-work/install/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/lts-6.2/7.10.3/lib/x86_64-linux-ghc-7.10.3/tensorflow-mnist-input-data-0.1.0.0-JIQTTyi85Nv6pdBnglu33Q"</span>
+<a name="line-21"></a><a name="datadir"></a><span class='hs-definition'>datadir</span>    <span class='hs-keyglyph'>=</span> <span class='hs-str'>"/home/gnezdo/tensorflow-haskell/.stack-work/install/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/lts-6.2/7.10.3/share/x86_64-linux-ghc-7.10.3/tensorflow-mnist-input-data-0.1.0.0"</span>
+<a name="line-22"></a><a name="libexecdir"></a><span class='hs-definition'>libexecdir</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"/home/gnezdo/tensorflow-haskell/.stack-work/install/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/lts-6.2/7.10.3/libexec"</span>
+<a name="line-23"></a><a name="sysconfdir"></a><span class='hs-definition'>sysconfdir</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"/home/gnezdo/tensorflow-haskell/.stack-work/install/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/lts-6.2/7.10.3/etc"</span>
+<a name="line-24"></a>
+<a name="line-25"></a><a name="getBinDir"></a><span class='hs-definition'>getBinDir</span><span class='hs-layout'>,</span> <span class='hs-varid'>getLibDir</span><span class='hs-layout'>,</span> <span class='hs-varid'>getDataDir</span><span class='hs-layout'>,</span> <span class='hs-varid'>getLibexecDir</span><span class='hs-layout'>,</span> <span class='hs-varid'>getSysconfDir</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>FilePath</span>
+<a name="line-26"></a><span class='hs-definition'>getBinDir</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>catchIO</span> <span class='hs-layout'>(</span><span class='hs-varid'>getEnv</span> <span class='hs-str'>"tensorflow_mnist_input_data_bindir"</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-keyword'>_</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>return</span> <span class='hs-varid'>bindir</span><span class='hs-layout'>)</span>
+<a name="line-27"></a><a name="getLibDir"></a><span class='hs-definition'>getLibDir</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>catchIO</span> <span class='hs-layout'>(</span><span class='hs-varid'>getEnv</span> <span class='hs-str'>"tensorflow_mnist_input_data_libdir"</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-keyword'>_</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>return</span> <span class='hs-varid'>libdir</span><span class='hs-layout'>)</span>
+<a name="line-28"></a><a name="getDataDir"></a><span class='hs-definition'>getDataDir</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>catchIO</span> <span class='hs-layout'>(</span><span class='hs-varid'>getEnv</span> <span class='hs-str'>"tensorflow_mnist_input_data_datadir"</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-keyword'>_</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>return</span> <span class='hs-varid'>datadir</span><span class='hs-layout'>)</span>
+<a name="line-29"></a><a name="getLibexecDir"></a><span class='hs-definition'>getLibexecDir</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>catchIO</span> <span class='hs-layout'>(</span><span class='hs-varid'>getEnv</span> <span class='hs-str'>"tensorflow_mnist_input_data_libexecdir"</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-keyword'>_</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>return</span> <span class='hs-varid'>libexecdir</span><span class='hs-layout'>)</span>
+<a name="line-30"></a><a name="getSysconfDir"></a><span class='hs-definition'>getSysconfDir</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>catchIO</span> <span class='hs-layout'>(</span><span class='hs-varid'>getEnv</span> <span class='hs-str'>"tensorflow_mnist_input_data_sysconfdir"</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-keyword'>_</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>return</span> <span class='hs-varid'>sysconfdir</span><span class='hs-layout'>)</span>
+<a name="line-31"></a>
+<a name="line-32"></a><a name="getDataFileName"></a><span class='hs-definition'>getDataFileName</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>FilePath</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>FilePath</span>
+<a name="line-33"></a><span class='hs-definition'>getDataFileName</span> <span class='hs-varid'>name</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-34"></a>  <span class='hs-varid'>dir</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>getDataDir</span>
+<a name="line-35"></a>  <span class='hs-varid'>return</span> <span class='hs-layout'>(</span><span class='hs-varid'>dir</span> <span class='hs-varop'>++</span> <span class='hs-str'>"/"</span> <span class='hs-varop'>++</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/src/TensorFlow-Examples-MNIST-InputData.html b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/src/TensorFlow-Examples-MNIST-InputData.html
new file mode 100644
index 0000000..2dd0400
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/src/TensorFlow-Examples-MNIST-InputData.html
@@ -0,0 +1,42 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/Examples/MNIST/InputData.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Examples</span><span class='hs-varop'>.</span><span class='hs-conid'>MNIST</span><span class='hs-varop'>.</span><span class='hs-conid'>InputData</span>
+<a name="line-16"></a>  <span class='hs-layout'>(</span> <span class='hs-varid'>trainingImageData</span>
+<a name="line-17"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>trainingLabelData</span>
+<a name="line-18"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>testImageData</span>
+<a name="line-19"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>testLabelData</span>
+<a name="line-20"></a>  <span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-21"></a>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Paths_tensorflow_mnist_input_data</span> <span class='hs-layout'>(</span><span class='hs-varid'>getDataFileName</span><span class='hs-layout'>)</span>
+<a name="line-23"></a>
+<a name="line-24"></a><a name="trainingImageData"></a><span class='hs-comment'>-- | Download the files containing the canonical MNIST samples and labels.</span>
+<a name="line-25"></a><span class='hs-definition'>trainingImageData</span><span class='hs-layout'>,</span> <span class='hs-varid'>trainingLabelData</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>FilePath</span>
+<a name="line-26"></a><span class='hs-definition'>trainingImageData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>getDataFileName</span> <span class='hs-str'>"train-images-idx3-ubyte.gz"</span>
+<a name="line-27"></a><a name="trainingLabelData"></a><span class='hs-definition'>trainingLabelData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>getDataFileName</span> <span class='hs-str'>"train-labels-idx1-ubyte.gz"</span>
+<a name="line-28"></a>
+<a name="line-29"></a><a name="testImageData"></a><span class='hs-definition'>testImageData</span><span class='hs-layout'>,</span> <span class='hs-varid'>testLabelData</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>IO</span> <span class='hs-conid'>FilePath</span>
+<a name="line-30"></a><span class='hs-definition'>testImageData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>getDataFileName</span> <span class='hs-str'>"t10k-images-idx3-ubyte.gz"</span>
+<a name="line-31"></a><a name="testLabelData"></a><span class='hs-definition'>testLabelData</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>getDataFileName</span> <span class='hs-str'>"t10k-labels-idx1-ubyte.gz"</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/src/hscolour.css b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/src/hscolour.css
new file mode 100644
index 0000000..c15919e
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/src/hscolour.css
@@ -0,0 +1,5 @@
+.hs-keyglyph, .hs-layout {color: red;}
+.hs-keyword {color: blue;}
+.hs-comment, .hs-comment a {color: green;}
+.hs-str, .hs-chr {color: teal;}
+.hs-keyword, .hs-conid, .hs-varid, .hs-conop, .hs-varop, .hs-num, .hs-cpp, .hs-sel, .hs-definition {}
diff --git a/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/synopsis.png b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/synopsis.png
new file mode 100644
index 0000000000000000000000000000000000000000..85fb86ec84907bcc86531dc82871948ff4d471fa
GIT binary patch
literal 11327
zcmV-FEWp!=P)<h;3K|Lk000e1NJLTq004jh003wR1^@s6IVt!g000+IX+uL$Nkc;*
zP;zf(X>4Tx0C)k_S!GyNTeqHT_l8Y(cXyX`gGi?cY`Qxn1VID|MJXwjPC)?)F$h6K
zMMOd+6hs7sqbPzXbr*U(-*=zy-hcPcUC*=TdiNM(jyd-lv&OpsU|J&v2m2!^0SE{T
z54F(O;<U4O;KWV=44?vZKnREv-NJ%3&CHI1|Ml9M1kj;!OzhCt|7!L>E2!K(!rTCW
z%wV;vdzf1QjBf#e&~gh74F>?Z4a=WLg$KhJ^$5nap>PLbJadS>e&h8+?D`9%QNL`g
zEVKbYGXj7k5Q(8)0Fd#*a?VIMFW3*64geVHKzE-&0BG!BtmfuTbO(T`0Jaeg2nagF
z{V*1E{Wm{e|AvV~*MEExiC+KU-~R=!2{)|c6Bg`GjQ;iG|FQ`1kAUCTuZtQk34#8{
z4r4<P-^=bq9b@>(3g7#|{=Z@d+d#}7f!3C=>=26vx*jwA8>@MS>RG@Tt_zt3hie^T
z_?<zX!Qp0Bf6L=zytR+Pm;lRjJ;HSV_$l=z8Xd-r9G16)Mp#(Gm>0%9VUd=)Fos7I
z^ghPh%Jy%YZ|)vCf6EaFPai$Q-!=$ppK!y&wrJs)bNdAuANB!m3n34Tfj{s75g-&U
z1A!Pg3bcXF-=!Gv1VmU93G2duANT;{0JugFTqg*|oPXPC|A$2HS3NJd-hcPV3EW`Y
zh=1Dr-5Mv{<{zIvz#Yba<V3iQL)~HT!km4Q{?+Cmi$j0EdL!8&xgxpyr%xz;lr%~X
zrG-*OsiBkrCyE}$48%~fC`FVeN*$#F>y&^Vcn^E_`qRfl{{bzYkp)4~$~NAx_VB;E
z{?P)PU)DbV{Qi#~0H0@T9czDj06@6MNq8OrpdAz(9qQxd9nPr<&s+~tPQySqaZyfb
zNh!%g_5YjeaLxMN*$s<dYY{|0@-La=c>v_p;d%b#U$Wpz0Geb0U>E+EOsEQ;I<Re%
zqFY30<Zpkd!v`9W019}7G68nL0|epsEDhv=3eW($zzCQED_{?t;PK-PkHJt71>!&=
zNC6q(BFF<p;2Nj^wV)BSfDUjM^n!=r33v`Bz%-Zxi(nP3gDtQFK@b)qgJ>Wohy&t-
zL?CHM5mJM6p`(xmWDmJOUQi$u0mVUQpbRJ*DuT+OI;a`C4fR4p&?xj8nuk`Puh35f
z5<x)FBG?f82nmEDLK9((utJ<bcp*X%F^FWu1w;X&9C01dj_5@^K};a#5UYqyBtViP
z>5*JWF{C0=8)=GkKzbrWk@3iMWInPS*@Wyu4kE{pbI3L14-^JPgW^Pq!Q<2bWsPz}
zg`nb5nW!REEvg;Wj~YYGqt;RTXfiY_S_G|(HbmQ@z0gtU6m&ki8r_B-Ku@3-(OVb{
zh8`n;QNS2r>@mKWSWG773g!l;2Q!LUz-(f%SSG9pRuyZCC1S&|DcC~nb!<2G1$Gg;
zjU&Zz;G}VSI0sxHE(w>9tH<5Py}&KucJP#VKD;vC6z`6Y#%JLx@m=^4{33pbgo;Ff
zM3uyf#Fr$Iq=2M}WPoIbWP_BHl$%tE)ST3Z^fYM!=}po{r1PXd2-E~&f;PdC5J9*=
zs3G(aUK2LR$jJD~G{_vt!pSa>)sa0QdqcKOPD3tEZb<G<o<Lqq-cJ64{1XM9f{#Lz
z!ige=BA=p_VuWIi5>LrbsZB|wjHfK7yiNI%a+8XNN{Y&qDu61Js-9|yYMB~K%}=dM
z?M|IcT|xbTdVvN>!$YG@<3@9arjllWW|0;{D?n>V>r0zK+erJ2cAbuzPL|Gw?j&6?
z-95TFdL%tRy&=6neHMKS{UrTQ1~vvw1`mcbh9-s=4Br`97&RC@7}FVVFitT3Wa4Df
zW%6UX#MHqw%Zy?cW;SPzV!p~ez`Vvn%c8>K#*)s`!ZO8*U=?PyV2x$1V13HE$;Qs6
z&lb#9$o7D3jh&udgWZ=sm;FBb3I`2`8ix-@E=M=VM@~9UO-_H#0?vNUbuLye1Fi_J
zGOlM_JKO@?*4#<lH@IhcNO%-^Jb7|?`gqoPIe1NY<9Hi*r}*%EihSODm-z<yzVi$4
z+w!OKcknL@FbWt7#0WG9%m|VRstbk)mJ5yxVT2Tg{DiIuJr_m@%L#i67YmPwAVlOv
zd_=B@jEbT~l|%zY%SB&`5yUjbBE=fS=Edp7kBTRWcZh$I;FWNYxFj(su`ekn=`UF+
zIW0vkWh9j()g|>+T3Fgmx>$N#hD=6JCPAiC=8LR|tcUDX*;jHjawc-Aa(!}p@(S{y
z@=fw93cLy~3MC3J6=@aC6f+ecDWR3LloFKgD*aHFR}NQhQU0tVrsAhkud;kZ;E2bO
z$|DP^+^R&?GSxXXPBj;`QnfjCE_I@Mx%xW|9u0SmYK<jLAx&S+>zbdmB(*}d+O)oF
zD{G(9?$JT&=<B5GJkzDtwbm`teWS;%=cU)6x2`XzAFJPEfHp8TxMc9skj>D|u+DJZ
zNWtioQNJ<4*wVPj_}x+AqoGH;Ob{kUCOIZE$M}u~9_ug#riP|Drn6=OW+7&G%rWL>
z=Ede8ETk;rECwxUES)XuEw`++tg@`8tp%+ktov*zY#eRsY`)v-*k;?#*-6-)vU_6B
zZ0}>=>40^xaj16KJg$2@@A#sloMVdP<OzWjCr=DHF+2G?-E*dJCOWsepj~WT>Ron;
zro?jMrmLZAiR-$Xw%cX5Rd)^dT=x|ZRgY|sB~Mk)Y|mvcRj(Yc6>oL#eD5_MZJ#2a
zFTMu8*L=VGnflfE9r)Y&-w413xCGn|qz?28><?lOIvF$?EE1d&JQt!Ik{7ZbY8+Y}
z28Epn>kOxb4~I`91S8Hy%txw47DsMJ*+jLTq&gXR@@ceibXxRMj9yGtEGpJ5wl9t=
zE-`NYl;)|jcqraAzAu3%Avt03wEpSZM3O|m#Ni~#r0k?`XKc@OC9@@;PF^^xf3_io
zJS8;cWvWW*wR5O*KIfjL$)pvg?Wen^KhBWM$j{i#bjy5vUg~_o`GX6d7oKIwXI;IB
zxfpnH@{;<c+H9)qnCyidlbrTk&fN6eZ+XsnL;14#*DjM>j<`HmaI~Pakhkz+;ck(4
z(L}LU@r@GJlC+ZVSKP0>xT<lr=^ER$jB7tjeM(=I>6f*a^OxsWU@9UjK2+LN4pu2v
z)m1ZBXH@Ui1lG*eTGaN}Db&@~v({%dAQ~bXR<1i<f8MCm*x4l3RCa^@M#fEWGvemQ
zW|!v47UPz_R>jt)TYR@l+GyI++oAU8_Vo_$j=4_z&e7XOxBI$Oy4voD->JFFb+`B)
z-My^)B=?i=A9TlbZ}tTDto3^JF7!F~O+T=EFy3$8|7^f`;L$_9hYtod2fH7sKDs-k
zJaqf9;^U4d@=w~I$~|oxmK$z+CjYE`L}8@!xzh8l(IcbxU#P$69n%?mIBq!pWa8Mw
z=%n@JtCx;1=U%zLT7K>S`pZ=0)Xwzj8T3s0Eahze8`d}FZ-w68n3JEoH?K4Q^qu9q
z=>@li)%RiVcNddCkbTHs;#jI%mR`QQqPOz=<MEG+tHG-~Ybk$G{Z;Zw@YAi&8lRu8
z+pRBt3H@^LHFJY$qvo6Jx4uoY%{N;CTRY#=x0$wUe<=PK{Av4hX(wtIyIZg)wAZz7
zygzdgaBvVz3?=?L3n1YV;pqv$*BUs7u!nPp=KzrO{LU$US-|NE1lz#a2I7HkB7Bib
zC_HKj-GeE>CgGy+9whdp4g`BLCvp!8U&;uov(!a2t+bEnRv6HXyi9t`-YglcEo`$K
zI8GTZXYLH1F5YE+b^&9-c%dfYc~N>X1MygiCdpZ8N*OKLV7W5+5rusvVP$KTgd_E;
zV`@J%*flk^Jhjj1)aX9cTQC5ItVZ(2W=FkE;*aH-)|+*kk6SET?pjmWaNEk+>D${o
z_#cmV%sNr-bj$gX%QW$m8{|&wA?SI;%go!uC))SCU%7vKz~jI-L0?1Ap^RZ7;i?hG
zB3+__P9{WW#uUa@#oavB8Q+`m==5;nXwvwZiR6j1<0+%5!{;8Q^`_s>XwIxTUvlAM
z)|rdpmprp=bM$iM@_6#8@((Vr7Q8HcP;{fXs3iGH;8nY8TBRaov}JqcixtC_ZB<vQ
z&(`?YTGlDob2eZb{<=Qi*ws{eBju)Fvqg(SE88vPt>w07?YBCLI#1vB=rX<|d6)j~
z?!9;SA9XkN<n~7QIX}?p7apJ-*nBuW*z>4rDD83J6N{$`!z{xG&lW}=KCd6md=WHe
zF)la3F!5t@`sLkMS6?Sg5vR3gcxTbGOK%>(y*_twKH{Cjg64anMViI^4{J-a%g<LH
ze(YYoyLRud-cLiHC*aZaV}tse#HQ(1*!QCCfgc+?yt~$WY5Tng2XOpFKtGg+ut(A%
z-=j*=VHi8C5{?;<#;=l0k&Y4u$@<72Q1np_Pz_U$(Jatz(CstOF!D30Fk7<(v!=0C
zv-fel=R$LH^XTz<^JVZi3XBSF3Nwow5p@<z60edNko+Kxkr9;Dk#m)gS143$Q5sTy
zqw-A^tHz`*q@kp#r)8miT*qBEKrc)`${^M-!6?x<@o1vS>0=3|@n*5+(H4=G;Z`Bm
z0XDw2UUnY#t`5ZG&W<NeoN#h<c62#GbaHieBf7hKczXJJ`FjWZMEFMgo$^l#ND0gg
zx)_`nQW$zAtSr1TqB^oJs{Uj{bVE#iY<*nqsjB$0ge#{D6LXTzpE;Wxe>ObDFO_)C
zCe0{aEki1k_dNXt+=U-mA1_W_8p^(%Q<IyQmz*DR*{Q&|P@#yYn4);6Wa-M-)$VKc
zrTJw^<pCA;mAX|@)oe8+HQ#IB*NxV9H#A-^X-sd5zTtDzzS*!vsa50_Ya2Pdf^Bpx
zbxz-Y-ZgON&fVsFb@$7<i+XZ;v-&a~r1qx_oPC%wnEEJVDC=?lldDf_hFhL>j|@Mb
z9sM+h7-yIepVWIvd=>Y)XzKR#)XeT1j<?!#-t!mU-CCGlggyu?nJkB_<bCW|eYv*z
ziR!cPy2cmFuWlPb-=a6;w@!ag*pB@X{?m8o#IEU{#=iIg)35lmfin;%)QHeRj3K>H
zI8-@&65hs?W6g0$Tn9b?K9MevmJ{6JljSOT6GbGYHWfM5G<6M41g#z&E8Qx6H$yI?
z50eHn6Z1ODBi1suSavH8F-{EUJXaTYHjh8AJ|73)7XPq7g<!OhsW6Z5mdKE3k(jTz
zvII_ILb6OMKw41-$PCI}kh79!lAl#5RdiS4Q~ID>t>OirQ5IDz)!g7S$y<#pnvPn`
zTCcP(>sag3>W=B<=vx}l7>pa{8`&AN7|$LpGx0noeC)GnyV)so9SefRgyl6WA8Q%w
zeVfO&`F8I1(hk7k+3~B6fhW|RD4pIpx4EPekGo2^q1>k2n?25Xx_BviQ+coYJoGK~
zi}SY&kPV~?{2VkK+z^r;>Jw%VE)ao<Sd8qCsyuln+BHTamLqmAZW>-y@)AN%A4?QY
z!X(X~xtpASHaNvFl_z!g+(cSqdP;^<ru=#G^Gg@HvkESTTr$e$&EC&>mD`$^mG5`i
zpn$&+Rk%>pUtCp^dd2Um*){o6wlZ|t=klqF!OHfk>gs};%-W>7nEHr@(CeX%5lwM7
zQg7xp*S7SwzHLLbOLn+*Uc0?`NAB*$d)wWCJsW)~{h|X4gV%@BpPU*_8L1qd8t0!(
zdySmVd!st{bK%K{=9Rj&=Ffv)KX1|hFxkC)82{hg(&3(fkq6-NB>?O?0kGBtAd?QJ
zm0$~|LIBLj0I*U5i1iA9XzK$|?dCuG2lOlFq=GX}9v}f{nuc(O=>uZH1yBw;!3bD_
zU{(i`gLA_m=mOLPjX+-zbO<GcGvX|w5itp8bfQQ*WD2qkUVrG}Of3%84DUniXj}9p
z^ivET&bv}EkFcb0hLww*!SUjPaCh){ydAy{ze{3HQcDVv9w+S}&=VpFugO%&D#;1t
zQRE9078DOC<tZDexT%V%>8W#QsA+O&>1m7Uxak_`<>>nu%o*kx!<ev4CCswS!z|vc
zXx2(L9kvhb=^T<AuQ(IA#JHxo(|J^QKJ!-bo#1ES9~a0Gv=U+xdMjKf5+SN1MlZG|
z-Y;=YGG59>T2DqomQ{`*59GHMHWa@qZ7S~^!Kl)z@vEz7SZjuAWovinywxMoS2FN7
zEH|1t%4A}H?2754xrD_j%Moi{n>gE7_6iP##}7_;J59Lg5Ifz(-D^B~y{dc!eQ)?H
z1`GsQ2d{)Cgfm98MOmHv9&;s5@6?xs(nO0hxa6LcxN|CLdl`M_GqP+i31t7w9nHU9
zkY40hVt!S*RG^%pl2DDR1@+)Ms)_U_Lks^c#r9*J-d)LeEAIFAEIl9{kQ}rbihXiz
zxOZfJbZ?wtQtXx5l+ld&8>=~scSi5kK8P(dtn9DO{nh=s_)Emb(M`^+uiKA)7VrA)
zEB#tO5ODlSVZM$P@WWh#2Fx+Iz|6u~m`%6|24UXdCqxG`1g0=2kOkd@#-Q&AR(P%P
zMdTpvAy(jBM;jT2tUyk{D~~EF3{{U>K(nFk;T(JdLx-`&6l3PF0@xsI7Y>87!d2q7
z@J9GD{0|aKlAELyq`{<P1SLWl8G$UCY?<7Fe1t-eqMK5MvWrTEsvFK%hiUBK+%$p?
zLsvvEML)#g&IlN<F{v}XV@_oeW0_<<%_hn=1;_jd$2w;fmm@bl_bAT=UK2hFz8Cxj
z0?vX$f}2A9!g(UzqH1CcVqe9dOEgR7NySNf$XLi~$;ru!DDWzBDsd|Fs0baAQPog0
zQ+L;h)y&gs)gIT`*5lMSHi$92ZZv<C*~A7u*Ss=kxA3yOWsS12wQaP=IJh3~J;CFY
z>in5#@A}YP&ZEYQ#XH-V)Gsvv6_^~14ao?j4lj=6k7|w9iW!UZJhhvUlPHq(FxfQ)
zq?V>>q`%8dxgeZ1aw#H*HTOZjUjc35y<*QR6jwV-iRB~}tyPXS=-S45n}+?ysv9OZ
zzqJ(K(rR1j$hs}xHG4PtzG(M&@2Lj@{VyISJQ5#z^W@U7{<DMUcV5Jft50HIPQ0$2
zPMURnYc#L4Ah{^9q_ASN>hV|l=i6Vte3RLV-yYuK+dKCw{z!laG%#N$3ABJM%p<0O
zYA^skKqQbP%m$r-WBwLFh0ujLomRwONMWQ8vL5*f<`CmhgJ?Rm2f718hVj63W7)9r
z*mpQXTq~XnpG|@xNg&xFjU_!Gq>|CVvs#J#1w}9=HDxE2J2egUAWZ`85!yYvKKcv>
zJ4PYKJ*G+KW|m8=VQlv7TJY|}%0<qN;Qq<;jdz`Io&T%Amf)T+R)kiRM@&}SK*Cw_
zq*ShSi_DnpraX&+wqk%%q4J>0wyKDli~41a=UN19Bb{{JVSQ=?d&3H&&qviwE*<+|
zre<Dfv2EpG-EAvqm*;>!9^?4cDF}{Txa*#Kx+jZQvyZXwvVVG@WYFu7)G)>HwaCho
zPBE;pGpDX4cqED@Z6)`nTsY^LE}F4-ek7|Lj+#LpTmF}Vfuf?4z^j_2v}GSEI;v7@
ztn0YySFg7=Mcq_r{?^*qM(m*I?Cd&z=li|$-7G!jeOwO;25=992S<m5AICiD8%90T
z7zuw~Iy(4bWgIcVFe&g-@|E0cxhctM!5QXRlG$%>X5MzsmCeV$vtN*Wk9q%cvGzm6
zlGZYQ`Nc~9M~79`)tR-DzwAEIeH!_EZe4SI`^$~5?i-97Prt=)N^Q<<UHq>3ePg@o
zht*Hi&(<B6UGm+oJ@<X`{k8*#!}y1J%19sX9snX}!Sv&sgM$rz0I=Br>|HuI*eO3a
z*sFk(4fq>KkN@xQ6^F(cm~$_2K14li9;XkV|9<@!M&f%8Nam8p00009a7bBm000XU
z000XU0RWnu7ytkil}SWFRCodHT?u#;Rkr@KbUNvfeG_5`YY-wNfPp{+o{ADgGcxep
z5O;8ydCWk3pWowCbe1RjK4lzy;4&jKqk}U-a1=+ud7z@;LLwlFC>S)v1jwFrI_XY2
zop;WyuIf%_F<t2n-Ic!IS6#QNZr{59Ip>~x?x|CCgE~7q5lBOq0>MKUdH^|7ARquk
zTn+*P5D<VNPJtl~KcE2u0x-lWFvQ`9_7{IpP!Ma2Ag8UrfAvN=8&%31Xc{{x^H{Co
z5Bf3s5%3VAbLh&IH$@c|uDl{TIw}*k&TyNplZ9bWQ*bafH>lHMG@8ELxbaVWHf?&T
znHpfF&E_pZ&^rD;1;7qozi0Q$(`V)7{8<+kI>wdbHk%E>!9AN2eO+^{$KB)hHtVU6
z4;0@%KYw`%{kM%aj|)L>`1``u*EM%B_Ep|f_<QCY-I>7iHT~t6&rZsneaT;XVt##n
z3*O&%0=#!k4Gq$@x_XoAC663)d$?Wm=U<vc73yj8s7C4ye?8%O?|sJH9hZdp8`Rda
zV)AbC6#gCsfWR{`**=IUEL^cLOs^a53WvcKmwX4axn1m?g`lRU8J`}Mu9cL0kEW(p
z*S~w@nA^2_^<7tUJ89F|+dbgAd&6H>XTrha?_sgD)BZa!4dhf)W5g$)o+5f!@!6p=
z7>#E6lGpa0z~7?)*juclePn!mT$U>W2F?VqT7?}(LqHHhL#3+DoNXk5_#Pb{(lwSP
zZ<=X|iSbjYeFoatR`H}3=!RdX3qeSTbc>FTPC<i4yKox}dW8l|7rAq|($3Sz1I6E)
z0PuMxfT6vuJ<7S!uFMY$3&G48V-OV;j??9JIDGgtTjt^~;o&;W${mB481A6nNfvWk
zG}X)k!0qh~kGnVT5kF&`L0DN3I5DFSlBQz~XE21}l1oOBfJjYXO~BgEYVC0M-fop2
zDE{6E#$wJW!`duuVRDQ5Vt#39HNO0^LN1EPNWEbGuwmiKId5c0K$-Du1N3`J+mCki
zh@bx#O+t{9<6v}qdxy{j&CM3X#uzYdYMN4g6Ac50s6((}=&dwR{Cx?4=mpk}4k7Lh
zja}8gbI(8i^%;C~;;e!(COSeeKO#br9p##*Gxqi|^SQmA8UnbUF8H0>&5WKoW3<xS
z+uAzF<~aqE$%3S$Xk2`8s#2WoA?J>vT<}n4p!j<e0B)WjD5-NnRG#WdwxQH=@&k}$
zOR8R0dom`Lc=?)q&RcuB7q?Ti)x-Dbg5R}n#wZVUe2wiO-Z*u-^!Q-=uplma_-CN_
z8Dw8c=~1luSn8d}KeqH68(WbvJ_VB|4Oi+hk~;kO6RAfiC;c;{;CF65PXUWd$`m2j
z>ve)Qtntp05&Y$`N~L&mauhNrjZlt#E%Rdnz*4RdA(~WsS0P~4Cker*^h9K3rID79
zAhx!)2_f*-6tD+E@|~5o_HbR*DQEm#fix64W;xPOIEsuwz3>ej`Mg}wlx+M?%^s;7
zt7<_1|D+24j|zb6{d*Duo)R*nQ%A&N`m}UK6}Gim#oV|jr-^I5{&3u6Y!z0&JjK=N
zf~iA{0UNr_&1RH*=FkdaRxmwXu@ih1pW6b!KwO1@&&hNB<m70YDveg=#B4Xn!v>f0
z=VYU~zns|bF>|Ig{pE8Oi&e4q8Sf>;d>$HnJ*g4^2E{@!BWJXj|MK2>t{)#4iCiKM
z_X3_Wd3!22SVWGECF_5t9Wx1ebd<xtk&rM9vu2GY%sRn*f5AMUxid3;-rS5H$2b;{
z-mVQ6!@O~ym}o<<xL6?O&7IIAJ-pEtPt>Ve1IRabo*K&M<b7Y)<TC!a?+aI7*L)G{
zfCa!8K?T~>e+mp(08G`jsI~A7O*rz=A?*I(Ym_y4*Z<vrc-}ABZsXaPWE_55X~M~}
zTIJaLcOUp224n-2zVikie7yn%YnEfm!5#SRe*5FmzkQB}p11+o{Fp|=>BHj<`2EIL
z@XCfeu<hfsczydWJ6bb}o?S)x0g-o5D5qWnbezFoiTRwunk%qdX1+L6i)|mA5jidY
z{0}jolhGWEzbnk=GvPS&`Ekzpd{Qp}jg2@(@Fp)^ip7aWa9kF~U5c$wF9JUgt8d{b
zE$t=?X?=dwwZ3ox^kn_>GtW8G6RGFlFM<@CjE-OtU#5a;0kB%yXw(N%<3n(~<Qsu8
zZ|+#=eP8Q&gdZK`A?MLQF(CmT9+sYjF8fKIN9rETN>sBeG(H{~)Y9EAyo%kT#Rg2j
zpdOnacnjrpoDswQL%S&=xD)LJZ^c?^7~tUKxVSW2U-+UJ`I8c2{Q|sd4FLUcTr-0M
zaqMa26wFKpz7U~s3AlNV<LRJWL`!?_{RP)vp+K58DGni_Hng{J{G%C}6UQspd^w|9
z18~65>^qhrHMbm9<`9gTLcVV_VCkY<jGCK;u}-q#%LEum+wQu00v693hqg8|E+F?e
zr#1I3!kny;lw%R`{p7X)SelGzum&R`J{*=hGlr!kVHo3VG@_P-B6~JTzO@CWjwnQm
zSR&sif;%k~Gk*Hv$KL#GDAqtA4UzOxf>cW$bp+1aV?*4j`n;5NQvl5P$NHC1)DVqF
ze?14Uta}S5dTDmrRR#Fn;tPAZ>c6M&cw`%zt17X5(`x+mXPZPMYENh$xHA{IIn#Q&
z<WuL|wgH?ccYQ_s*J*Sb=&abC^R8TNFm``&#I=ks9J{Fj!m;GPmSYjgeN40tkN)>^
zG}YF_5*3HIuofIEDMeLB1jc8M#;C+D(d52>)gx`#@~i9ZqkAV_+e~x*&R~QFvHtHw
zX=O8P?QIyJ9<S9QfZh~|rm@{l%9`SLom|1Yuu=8~?mQv6f;~cV->Ss9*B|&g;0hMp
z3Alm-uHb+xn7Ts16&!E{`__2XkJh+p1UhOAxPk+&;D9SQ;0g}7f`^<d*f9ujPKWd{
zNwA(ei(>~4p*<oBsnf?}Y_tY_z+HhWM{(ruZlLc|d{fqlC~^zWp3Kn@&XYx@s8428
z+W?HF@nkVpZ73FT0nX?mGjp-|=hGB_f%7#j<?C8Quwu<K-PpY#ZzNXSa*ba=b=wuJ
zprVklr=6G|NiR*v4kFXBPw?d4Dy67aGq*W0e|U1NJv0<}=fG*4{I=9DV5`;u*19?o
z5{B?pJoNC@$d2dNpxSZdpKs&0Z&je7$@!^)U(mv>Mp`Hum_uHM8Ep9TllPO>m-^Cs
zpVwg1bK6i`-w1z*2vDs7WXVaJJHyU=rk@Vk3#W^iKzdl}7D4^3u#E2B8*>%rGlt8u
z5=Bg)^vMF>N2OW-kTeo=C=#;#Uwg6hiz=At%UPznGuZL$9uX3jIcgXzEoL+}ne7De
zePX!NLIZ__1sfvp<d*$}QF`*n@c|qD`l$#Bn=o#aJJp@HvT{duzX6s;924PRO?YW@
z5nghs!8um&p{H&~w#Fia`1v1)e}&ROaMy!>aY5fTR<P*VpVM{qeXU#N9z3DGRUK^<
z`8ISD3xNMj_Il@&dJSMm!=vkNfHgFv_xE4u6KU4aaHMFg_cJ-U=WN?_Kka)L1Nw>(
zUH5HKQ7-^w@TCk-ATqS$+;^2Y-9Yg{p~En8>~LcE&~OCN2SO-y!qgT7qsff0kWR!$
z^D81!lBm$TfXL;}=Y9YJK+SF{!{d*<g%X#sAaChntiD3qVd2a){!IDu!avzV!TF4o
z9xGgkjB%soL!aKOw*koiw_`(h<}=y@MgPFg@A$O|ZKq#6{$@AkGvc*R?Si#UlQDr#
z3~i~w{&Resf;DUNF*}X?fEFV@D0-Vh!)meXqjD6C9^>=}ZDsk}pA}{0WdF3_)n|T5
zFNK7P(SF;zrP#jx9qieE2>F-K@p;gyHGt(@rI_!hEt)McpP}lbFn3v=a0JCAI=-Ld
z^HfmLKw}#PgVO)j-n&3BpR3@}{)WrPilHHGIK3w22T8R6=u<`rMwjnBh~jFy5zt}A
zN81hv!KkMXNNPDnh1mq7H@>uwma1@k3;2!wtQCOj+9tn%uigkWBw{AL|5)BofhX2&
zA+XZ302%fCsUzg9CimQPVv`f;C6O8|{n>ML#6sZcPqU_9<QKRdEq1%Pk+|_c^D$ZZ
zpTp?8<hlYZ8Sj`5DMd7vdH&JsA-;|ya5+f%=&CEBr`QO8Na{5Jql+JdN1k1QvXkYg
zs3w&k5rOovX-K7)MutMt)1dz*eGsClwG%b<Mnp!CZzRsYY@Y!MK;y}yczWw0oN9FR
ze!+NJMke*DMR;s+x~Aozt~D6zfBzb)s_VMoY^H}`58OE$vvRV#ocF~8^#ULQLiXc}
zUQkuE$PY|=DWHMVx2^eKu~LOA6rK$<8jY@}eezhf09Y%}*qLuY{&mxkmF%SY=ox@6
zF~uhVd#X`JdPSa>DPe!$!>g7c<pOq#ONp0~Ns93gcerW+2#u6Nz)abBxO;&=#J<Sv
z@ytFfgdg<)OXiP7*2GlLRC*q2ahR^|-aI}H3;8oRM~>oyleK6R!5=0O9Kit+4(r(6
ziv6QJ8-P(X4Sa3SakRGjFIv?a0G4_jZD3}d!^RD-cH>&cq5?d2jrKkeAp_;!Ur#;&
z9W7Y4e9epUX=T6m-g%gom8l&2YDT>Vpn#D2K2TLOYC9;D1)wkDRn>N#8T3J_^Lk0W
z2GEDo5^3Wxdgdfd9w7&WOIUcVywJ$#^9sz{H)rNATQUdN%*}+3f?}K#TL)6Cfb&`3
z%&Qjw3IaWJ_$1z;4dDsM&%YQ~=42pUgopbkSWmW!9lu+5e2Bl(Hp~!=)psw#l#5d7
z<59t4!9`Er%bRtn7l4p3WRMY9&31sf7Q0{HC$^-K>G(;07G_Pk5PmWfQbk{$>nD;C
z$aX+;iw(co_@<~Qn^p+B=a%_MiWA>XQ&sn1{z<(6(1#*dufHEF>#Fe8m!&8!F2%dw
zHlg}-8UFYJZG<8tdn)d^eHPNC3G-m$^7_440RBMV3*u1l6Q_-M<NrRRMWepLk*_Lo
z-`bbN{R`%2;?^Z|DbJ;SGxcT1^sB{@@yW=TI0_Rc#C0ow;>ckXuK!rmQ$k)#dR$sG
z@^U71!@qOSF|2)@pOpG;Qm+AE#NKTmpy<6aRJ-8I$ex7UR10>zRSMI&Dx4*+aC%oe
z$>ksZdHCl3@33X-u5M#~!F>8s>bP;(@Z1iZ5DQ57E<W`F&{Exn`H?9pD7b1W{*9JP
zOVBg7&1l$Lik;?q)RUhiU1Oo)Yy>(pe>^RmdH=2Rkv1Y;;r0f4a|kUQI?AO7tZbEf
zJ(*E203jiWBR5FKRnt*$=_L9l06hS)bRb+XpPQ(|6)W>G1u?i-W6WoCJgU<Q$tb+@
zD$Jffj@}LJ{)WH{zrK~CgV-P9o6Kh%ssy0+_)+}f)#J#erD6&w@S}9^J6Kv&%Joq)
zVCl;1aO2ciq)r?I|MikfXnOjsHU;7FXSU<j4Ywhj0zZnsHR6e<UbYvt;`zsJK|)+S
z62!-nogR|)RM~kx=eB4z9^a39Htt8sxmx;YG^zT!qj*qZKBH+t(F@!0k$*VC+$MjO
zDHQiTunA>lRkTWYJ9y;~2lKhQP~5|72z2_#^8q&npdI^OKWZnM4)jd~l<t#{4Y+Cf
zJmk?iep_h!z8~!6w6%M3%WgX$8wEf`;a??H_G2r4_wX%<Bi{h4{7*`b`2}z{1;B(K
z1YRBPTX!e&hj*a(pcMU@bm?3ynVpH1H&2p)w`(&X<ooZ-37-u`n^q8?b=dLG_XXa&
z|9vVMfYDf2c#}A8B~Ph8B=-cMr4A=W1b8$?C6oGZsl=!338XY{hCK?{ZdF8R@clrx
z<=e!HAY|AD>xFIKK%PKOm(9u+`!IG4P>PAtq9@Rh0<!HW6ucJiefk6H>JE!{0DuH!
zkK`y|6ZXDM&ju*fYcM2?dkd?0BQd?AvKl9=rI$l^%Bzo%82<d<-V^wnjLl;2>pwp_
z3!t@d`N^j}MPee&>2}gr!FRvB)4o^~UCPYDMfxiI>b@c+MsVI_ZG?n%#SdILF9)yD
z8iBv~&32h6$j=)^`5;_--)1F7aK==Pycf`JwRRcIa&EjD`NGhX@h9M+TM4YCmA;oJ
zrO3=nv3MeD1n(z%`&dZj&7(JU#eehVv~0XE^yJ%^arZ3+;^s6cinJi_LRv*8MlRsh
z{Xp^er2%-zvwii|iPQND<~cxwB;)S&_u$&{D%8_7aQMh%>8YP30yAe!z=De>;j*0J
zN>6b7(K|VAAJyy)=J$-BZpMp7n5{I{+sN@1<}jm{UYm<6az<l#QupT~-a!ux6n{`>
zC)2KLBDKeY!To$ha&qG2BZqfAotPNM^BbQ^H8u4$*;5z(vZ|_v=c1LgH4&aJ8cR)s
zhZ25=_;#ffO9d0sLd30K^&jiDoI6+3R|Htse-FYDw`bL=buUu;*yY6jR@v$9iMtOO
z{Jm)a77X@ba<Dr`c8-aSjUKyf*>%$<P8>f%7edh>l!!{woQDqvAyLn?wOiY*$B%zo
zv32X~pEWczvH$rLZ56cfy6vr`0a$epDA9d}4E`PkfT>4BU?%e$j!CrfB%e1P1~}M{
zuQ8DZRRHLI>|J6XE5CNbPoY`u^Tv~L_DESt0J@K9biv&;RPgs@1TwMtC4bqg&n_U&
z^RqpU@fmCZV8(Krcxd8Db|Y=v9v+%_sqO*ye5%7a4GH|cY5=AL^#T?U?(IAraOf}Z
znfd(s?_l?Sx}{(;kM%5!ES&ry9?r8?uz9NYQ(Ynr1^j&q08@d8z|&jaWMSaE-1`Sx
z2*lKk?$1KN8*2mJGw(g3`l+riN$dE3Q~;P7L<D9yC)d4ncJO{>Cd=wx?7hW&8J3pu
z_e%g|LIn2Oqk!C_wTCQ#s9zKa2tdEcq}@UR0njdQ`-LnZ0R1A9b_)drK)bx{7qWl=
z^ovZ|Eff#{?eex?$N~b;FEVMjP(T2*%iDe-`+v|7m{y$1dn*6{002ovPDHLkV1lnB
B5rhB$

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/tensorflow-mnist-input-data.txt b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/tensorflow-mnist-input-data.txt
new file mode 100644
index 0000000..afb87ad
--- /dev/null
+++ b/docs/haddock/tensorflow-mnist-input-data-0.1.0.0/tensorflow-mnist-input-data.txt
@@ -0,0 +1,19 @@
+-- Hoogle documentation, generated by Haddock
+-- See Hoogle, http://www.haskell.org/hoogle/
+
+
+-- | Downloader of input data for training MNIST.
+--   
+--   Please see README.md
+@package tensorflow-mnist-input-data
+@version 0.1.0.0
+
+module TensorFlow.Examples.MNIST.InputData
+
+-- | Download the files containing the canonical MNIST samples and labels.
+trainingImageData :: IO FilePath
+
+-- | Download the files containing the canonical MNIST samples and labels.
+trainingLabelData :: IO FilePath
+testImageData :: IO FilePath
+testLabelData :: IO FilePath
diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html b/docs/haddock/tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html
new file mode 100644
index 0000000..783aa85
--- /dev/null
+++ b/docs/haddock/tensorflow-opgen-0.1.0.0/TensorFlow-OpGen-AttrVal.html
@@ -0,0 +1,6 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.OpGen.AttrVal</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-OpGen-AttrVal.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-OpGen-AttrVal.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-opgen-0.1.0.0: Code generation for TensorFlow operations.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.OpGen.AttrVal</p></div><div id="description"><p class="caption">Description</p><div class="doc"><p>Wrapping of TensorFlow attributes into Haskell entities.</p></div></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><span class="keyword">data</span> <a href="#t:AttrDef">AttrDef</a></li><li class="src short"><span class="keyword">data</span> <a href="#t:AttrCase">AttrCase</a> f<ul class="subs"><li>= <a href="#v:AttrBytes">AttrBytes</a> (f <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>)</li><li>| <a href="#v:AttrInt64">AttrInt64</a> (f <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</li><li>| <a href="#v:AttrFloat">AttrFloat</a> (f <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</li><li>| <a href="#v:AttrBool">AttrBool</a> (f <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a>)</li><li>| <a href="#v:AttrType">AttrType</a> (f <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a>)</li><li>| <a href="#v:AttrShape">AttrShape</a> (f <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a>)</li></ul></li><li class="src short"><span class="keyword">data</span> <a href="#t:AttrTemplate">AttrTemplate</a><ul class="subs"><li>= <a href="#v:AttrSingle">AttrSingle</a> (<a href="TensorFlow-OpGen-AttrVal.html#t:AttrCase">AttrCase</a> <a href="TensorFlow-OpGen-AttrVal.html#t:Template">Template</a>)</li><li>| <a href="#v:AttrList">AttrList</a> (<a href="TensorFlow-OpGen-AttrVal.html#t:AttrCase">AttrCase</a> [])</li><li>| <a href="#v:AttrTensor">AttrTensor</a> UnusedTensor</li></ul></li><li class="src short"><span class="keyword">data</span> <a href="#t:Template">Template</a> a</li><li class="src short"><a href="#v:attrDef">attrDef</a> :: <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> -&gt; <a href="TensorFlow-OpGen-AttrVal.html#t:AttrDef">AttrDef</a></li><li class="src short"><a href="#v:attrOriginal">attrOriginal</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-OpGen-AttrVal.html#t:AttrDef">AttrDef</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a></li><li class="src short"><a href="#v:attrTemplate">attrTemplate</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-OpGen-AttrVal.html#t:AttrDef">AttrDef</a> <a href="TensorFlow-OpGen-AttrVal.html#t:AttrTemplate">AttrTemplate</a></li><li class="src short"><a href="#v:templateDefault">templateDefault</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> (<a href="TensorFlow-OpGen-AttrVal.html#t:Template">Template</a> a) (<a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> a)</li><li class="src short"><a href="#v:templateRestrictions">templateRestrictions</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> (<a href="TensorFlow-OpGen-AttrVal.html#t:Template">Template</a> a) [a]</li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:AttrDef" class="def">AttrDef</a> <a href="src/TensorFlow-OpGen-AttrVal.html#AttrDef" class="link">Source</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:AttrCase" class="def">AttrCase</a> f <a href="src/TensorFlow-OpGen-AttrVal.html#AttrCase" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:AttrBytes" class="def">AttrBytes</a> (f <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:AttrInt64" class="def">AttrInt64</a> (f <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:AttrFloat" class="def">AttrFloat</a> (f <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:AttrBool" class="def">AttrBool</a> (f <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a>)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:AttrType" class="def">AttrType</a> (f <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a>)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:AttrShape" class="def">AttrShape</a> (f <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a>)</td><td class="doc empty">&nbsp;</td></tr></table></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:AttrTemplate" class="def">AttrTemplate</a> <a href="src/TensorFlow-OpGen-AttrVal.html#AttrTemplate" class="link">Source</a></p><div class="doc"><p>Type-reified representation of TensorFlow AttrDef.
+ Initially limited to just the types in Op descriptors.</p></div><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:AttrSingle" class="def">AttrSingle</a> (<a href="TensorFlow-OpGen-AttrVal.html#t:AttrCase">AttrCase</a> <a href="TensorFlow-OpGen-AttrVal.html#t:Template">Template</a>)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:AttrList" class="def">AttrList</a> (<a href="TensorFlow-OpGen-AttrVal.html#t:AttrCase">AttrCase</a> [])</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:AttrTensor" class="def">AttrTensor</a> UnusedTensor</td><td class="doc empty">&nbsp;</td></tr></table></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:Template" class="def">Template</a> a <a href="src/TensorFlow-OpGen-AttrVal.html#Template" class="link">Source</a></p><div class="doc"><p>Specifies the optional default value and a set of allowed values
+ for the given type.</p></div></div><div class="top"><p class="src"><a name="v:attrDef" class="def">attrDef</a> :: <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> -&gt; <a href="TensorFlow-OpGen-AttrVal.html#t:AttrDef">AttrDef</a> <a href="src/TensorFlow-OpGen-AttrVal.html#attrDef" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:attrOriginal" class="def">attrOriginal</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-OpGen-AttrVal.html#t:AttrDef">AttrDef</a> <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> <a href="src/TensorFlow-OpGen-AttrVal.html#attrOriginal" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:attrTemplate" class="def">attrTemplate</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> <a href="TensorFlow-OpGen-AttrVal.html#t:AttrDef">AttrDef</a> <a href="TensorFlow-OpGen-AttrVal.html#t:AttrTemplate">AttrTemplate</a> <a href="src/TensorFlow-OpGen-AttrVal.html#attrTemplate" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:templateDefault" class="def">templateDefault</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> (<a href="TensorFlow-OpGen-AttrVal.html#t:Template">Template</a> a) (<a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> a) <a href="src/TensorFlow-OpGen-AttrVal.html#templateDefault" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:templateRestrictions" class="def">templateRestrictions</a> :: <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens-39-">Lens'</a> (<a href="TensorFlow-OpGen-AttrVal.html#t:Template">Template</a> a) [a] <a href="src/TensorFlow-OpGen-AttrVal.html#templateRestrictions" class="link">Source</a></p></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html b/docs/haddock/tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html
new file mode 100644
index 0000000..2e411d7
--- /dev/null
+++ b/docs/haddock/tensorflow-opgen-0.1.0.0/TensorFlow-OpGen.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.OpGen</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-OpGen.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-OpGen.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-opgen-0.1.0.0: Code generation for TensorFlow operations.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.OpGen</p></div><div id="description"><p class="caption">Description</p><div class="doc"><p>Rendering of TensorFlow operations as Haskell functions.</p></div></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:OpGenFlags" class="def">OpGenFlags</a> <a href="src/TensorFlow-OpGen.html#OpGenFlags" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:OpGenFlags" class="def">OpGenFlags</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:outputFile" class="def">outputFile</a> :: <a href="../base-4.8.2.0/Data-String.html#t:String">String</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:prefix" class="def">prefix</a> :: <a href="../base-4.8.2.0/Data-String.html#t:String">String</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:excludeList" class="def">excludeList</a> :: <a href="../base-4.8.2.0/Data-String.html#t:String">String</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div></div><div class="top"><p class="src"><a name="v:docOpList" class="def">docOpList</a> :: <a href="TensorFlow-OpGen.html#t:OpGenFlags">OpGenFlags</a> -&gt; <a href="../tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html#t:OpList">OpList</a> -&gt; <a href="../mainland-pretty-0.4.1.4/Text-PrettyPrint-Mainland.html#t:Doc">Doc</a> <a href="src/TensorFlow-OpGen.html#docOpList" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:flagParser" class="def">flagParser</a> :: <a href="../optparse-applicative-0.12.1.0/Options-Applicative-Types.html#t:Parser">Parser</a> <a href="TensorFlow-OpGen.html#t:OpGenFlags">OpGenFlags</a> <a href="src/TensorFlow-OpGen.html#flagParser" class="link">Source</a></p></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/doc-index.html b/docs/haddock/tensorflow-opgen-0.1.0.0/doc-index.html
new file mode 100644
index 0000000..e02dd2e
--- /dev/null
+++ b/docs/haddock/tensorflow-opgen-0.1.0.0/doc-index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-opgen-0.1.0.0: Code generation for TensorFlow operations. (Index)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-opgen-0.1.0.0: Code generation for TensorFlow operations.</p></div><div id="content"><div id="index"><p class="caption">Index</p><table><tr><td class="src">AttrBool</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#v:AttrBool">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrBytes</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#v:AttrBytes">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrCase</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#t:AttrCase">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrDef</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#t:AttrDef">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">attrDef</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#v:attrDef">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrFloat</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#v:AttrFloat">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrInt64</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#v:AttrInt64">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrList</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#v:AttrList">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">attrOriginal</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#v:attrOriginal">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrShape</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#v:AttrShape">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrSingle</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#v:AttrSingle">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrTemplate</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#t:AttrTemplate">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">attrTemplate</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#v:attrTemplate">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrTensor</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#v:AttrTensor">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">AttrType</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#v:AttrType">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">docOpList</td><td class="module"><a href="TensorFlow-OpGen.html#v:docOpList">TensorFlow.OpGen</a></td></tr><tr><td class="src">excludeList</td><td class="module"><a href="TensorFlow-OpGen.html#v:excludeList">TensorFlow.OpGen</a></td></tr><tr><td class="src">flagParser</td><td class="module"><a href="TensorFlow-OpGen.html#v:flagParser">TensorFlow.OpGen</a></td></tr><tr><td class="src">OpGenFlags</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="TensorFlow-OpGen.html#t:OpGenFlags">TensorFlow.OpGen</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="TensorFlow-OpGen.html#v:OpGenFlags">TensorFlow.OpGen</a></td></tr><tr><td class="src">outputFile</td><td class="module"><a href="TensorFlow-OpGen.html#v:outputFile">TensorFlow.OpGen</a></td></tr><tr><td class="src">prefix</td><td class="module"><a href="TensorFlow-OpGen.html#v:prefix">TensorFlow.OpGen</a></td></tr><tr><td class="src">Template</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#t:Template">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">templateDefault</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#v:templateDefault">TensorFlow.OpGen.AttrVal</a></td></tr><tr><td class="src">templateRestrictions</td><td class="module"><a href="TensorFlow-OpGen-AttrVal.html#v:templateRestrictions">TensorFlow.OpGen.AttrVal</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/frames.html b/docs/haddock/tensorflow-opgen-0.1.0.0/frames.html
new file mode 100644
index 0000000..1b4e38d
--- /dev/null
+++ b/docs/haddock/tensorflow-opgen-0.1.0.0/frames.html
@@ -0,0 +1,30 @@
+<!DOCTYPE html 
+     PUBLIC "-//W3C//DTD XHTML 1.0 Frameset//EN"
+     "http://www.w3.org/TR/xhtml1/DTD/xhtml1-frameset.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<title></title>
+<script src="haddock-util.js" type="text/javascript"></script>
+<script type="text/javascript"><!--
+/*
+
+  The synopsis frame needs to be updated using javascript, so we hide
+  it by default and only show it if javascript is enabled.
+
+  TODO: provide some means to disable it.
+*/
+function load() {
+  var d = document.getElementById("inner-fs");
+  d.rows = "50%,50%";
+  postReframe();
+}
+--></script>
+</head>
+<frameset id="outer-fs" cols="25%,75%" onload="load()">
+  <frameset id="inner-fs" rows="100%,0%">
+    <frame src="index-frames.html" name="modules" />
+    <frame src="" name="synopsis" />
+  </frameset>
+  <frame src="index.html" name="main" />
+</frameset>
+</html>
diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/haddock-util.js b/docs/haddock/tensorflow-opgen-0.1.0.0/haddock-util.js
new file mode 100644
index 0000000..9a6fccf
--- /dev/null
+++ b/docs/haddock/tensorflow-opgen-0.1.0.0/haddock-util.js
@@ -0,0 +1,344 @@
+// Haddock JavaScript utilities
+
+var rspace = /\s\s+/g,
+	  rtrim = /^\s+|\s+$/g;
+
+function spaced(s) { return (" " + s + " ").replace(rspace, " "); }
+function trim(s)   { return s.replace(rtrim, ""); }
+
+function hasClass(elem, value) {
+  var className = spaced(elem.className || "");
+  return className.indexOf( " " + value + " " ) >= 0;
+}
+
+function addClass(elem, value) {
+  var className = spaced(elem.className || "");
+  if ( className.indexOf( " " + value + " " ) < 0 ) {
+    elem.className = trim(className + " " + value);
+  }
+}
+
+function removeClass(elem, value) {
+  var className = spaced(elem.className || "");
+  className = className.replace(" " + value + " ", " ");
+  elem.className = trim(className);
+}
+
+function toggleClass(elem, valueOn, valueOff, bool) {
+  if (bool == null) { bool = ! hasClass(elem, valueOn); }
+  if (bool) {
+    removeClass(elem, valueOff);
+    addClass(elem, valueOn);
+  }
+  else {
+    removeClass(elem, valueOn);
+    addClass(elem, valueOff);
+  }
+  return bool;
+}
+
+
+function makeClassToggle(valueOn, valueOff)
+{
+  return function(elem, bool) {
+    return toggleClass(elem, valueOn, valueOff, bool);
+  }
+}
+
+toggleShow = makeClassToggle("show", "hide");
+toggleCollapser = makeClassToggle("collapser", "expander");
+
+function toggleSection(id)
+{
+  var b = toggleShow(document.getElementById("section." + id));
+  toggleCollapser(document.getElementById("control." + id), b);
+  rememberCollapsed(id, b);
+  return b;
+}
+
+var collapsed = {};
+function rememberCollapsed(id, b)
+{
+  if(b)
+    delete collapsed[id]
+  else
+    collapsed[id] = null;
+
+  var sections = [];
+  for(var i in collapsed)
+  {
+    if(collapsed.hasOwnProperty(i))
+      sections.push(i);
+  }
+  // cookie specific to this page; don't use setCookie which sets path=/
+  document.cookie = "collapsed=" + escape(sections.join('+'));
+}
+
+function restoreCollapsed()
+{
+  var cookie = getCookie("collapsed");
+  if(!cookie)
+    return;
+
+  var ids = cookie.split('+');
+  for(var i in ids)
+  {
+    if(document.getElementById("section." + ids[i]))
+      toggleSection(ids[i]);
+  }
+}
+
+function setCookie(name, value) {
+  document.cookie = name + "=" + escape(value) + ";path=/;";
+}
+
+function clearCookie(name) {
+  document.cookie = name + "=;path=/;expires=Thu, 01-Jan-1970 00:00:01 GMT;";
+}
+
+function getCookie(name) {
+  var nameEQ = name + "=";
+  var ca = document.cookie.split(';');
+  for(var i=0;i < ca.length;i++) {
+    var c = ca[i];
+    while (c.charAt(0)==' ') c = c.substring(1,c.length);
+    if (c.indexOf(nameEQ) == 0) {
+      return unescape(c.substring(nameEQ.length,c.length));
+    }
+  }
+  return null;
+}
+
+
+
+var max_results = 75; // 50 is not enough to search for map in the base libraries
+var shown_range = null;
+var last_search = null;
+
+function quick_search()
+{
+    perform_search(false);
+}
+
+function full_search()
+{
+    perform_search(true);
+}
+
+
+function perform_search(full)
+{
+    var text = document.getElementById("searchbox").value.toLowerCase();
+    if (text == last_search && !full) return;
+    last_search = text;
+    
+    var table = document.getElementById("indexlist");
+    var status = document.getElementById("searchmsg");
+    var children = table.firstChild.childNodes;
+    
+    // first figure out the first node with the prefix
+    var first = bisect(-1);
+    var last = (first == -1 ? -1 : bisect(1));
+
+    if (first == -1)
+    {
+        table.className = "";
+        status.innerHTML = "No results found, displaying all";
+    }
+    else if (first == 0 && last == children.length - 1)
+    {
+        table.className = "";
+        status.innerHTML = "";
+    }
+    else if (last - first >= max_results && !full)
+    {
+        table.className = "";
+        status.innerHTML = "More than " + max_results + ", press Search to display";
+    }
+    else
+    {
+        // decide what you need to clear/show
+        if (shown_range)
+            setclass(shown_range[0], shown_range[1], "indexrow");
+        setclass(first, last, "indexshow");
+        shown_range = [first, last];
+        table.className = "indexsearch";
+        status.innerHTML = "";
+    }
+
+    
+    function setclass(first, last, status)
+    {
+        for (var i = first; i <= last; i++)
+        {
+            children[i].className = status;
+        }
+    }
+    
+    
+    // do a binary search, treating 0 as ...
+    // return either -1 (no 0's found) or location of most far match
+    function bisect(dir)
+    {
+        var first = 0, finish = children.length - 1;
+        var mid, success = false;
+
+        while (finish - first > 3)
+        {
+            mid = Math.floor((finish + first) / 2);
+
+            var i = checkitem(mid);
+            if (i == 0) i = dir;
+            if (i == -1)
+                finish = mid;
+            else
+                first = mid;
+        }
+        var a = (dir == 1 ? first : finish);
+        var b = (dir == 1 ? finish : first);
+        for (var i = b; i != a - dir; i -= dir)
+        {
+            if (checkitem(i) == 0) return i;
+        }
+        return -1;
+    }    
+    
+    
+    // from an index, decide what the result is
+    // 0 = match, -1 is lower, 1 is higher
+    function checkitem(i)
+    {
+        var s = getitem(i).toLowerCase().substr(0, text.length);
+        if (s == text) return 0;
+        else return (s > text ? -1 : 1);
+    }
+    
+    
+    // from an index, get its string
+    // this abstracts over alternates
+    function getitem(i)
+    {
+        for ( ; i >= 0; i--)
+        {
+            var s = children[i].firstChild.firstChild.data;
+            if (s.indexOf(' ') == -1)
+                return s;
+        }
+        return ""; // should never be reached
+    }
+}
+
+function setSynopsis(filename) {
+    if (parent.window.synopsis) {
+        if (parent.window.synopsis.location.replace) {
+            // In Firefox this avoids adding the change to the history.
+            parent.window.synopsis.location.replace(filename);
+        } else {
+            parent.window.synopsis.location = filename;
+        }
+    }
+}
+
+function addMenuItem(html) {
+  var menu = document.getElementById("page-menu");
+  if (menu) {
+    var btn = menu.firstChild.cloneNode(false);
+    btn.innerHTML = html;
+    menu.appendChild(btn);
+  }
+}
+
+function adjustForFrames() {
+  var bodyCls;
+  
+  if (parent.location.href == window.location.href) {
+    // not in frames, so add Frames button
+    addMenuItem("<a href='#' onclick='reframe();return true;'>Frames</a>");
+    bodyCls = "no-frame";
+  }
+  else {
+    bodyCls = "in-frame";
+  }
+  addClass(document.body, bodyCls);
+}
+
+function reframe() {
+  setCookie("haddock-reframe", document.URL);
+  window.location = "frames.html";
+}
+
+function postReframe() {
+  var s = getCookie("haddock-reframe");
+  if (s) {
+    parent.window.main.location = s;
+    clearCookie("haddock-reframe");
+  }
+}
+
+function styles() {
+  var i, a, es = document.getElementsByTagName("link"), rs = [];
+  for (i = 0; a = es[i]; i++) {
+    if(a.rel.indexOf("style") != -1 && a.title) {
+      rs.push(a);
+    }
+  }
+  return rs;
+}
+
+function addStyleMenu() {
+  var as = styles();
+  var i, a, btns = "";
+  for(i=0; a = as[i]; i++) {
+    btns += "<li><a href='#' onclick=\"setActiveStyleSheet('"
+      + a.title + "'); return false;\">"
+      + a.title + "</a></li>"
+  }
+  if (as.length > 1) {
+    var h = "<div id='style-menu-holder'>"
+      + "<a href='#' onclick='styleMenu(); return false;'>Style &#9662;</a>"
+      + "<ul id='style-menu' class='hide'>" + btns + "</ul>"
+      + "</div>";
+    addMenuItem(h);
+  }
+}
+
+function setActiveStyleSheet(title) {
+  var as = styles();
+  var i, a, found;
+  for(i=0; a = as[i]; i++) {
+    a.disabled = true;
+          // need to do this always, some browsers are edge triggered
+    if(a.title == title) {
+      found = a;
+    }
+  }
+  if (found) {
+    found.disabled = false;
+    setCookie("haddock-style", title);
+  }
+  else {
+    as[0].disabled = false;
+    clearCookie("haddock-style");
+  }
+  styleMenu(false);
+}
+
+function resetStyle() {
+  var s = getCookie("haddock-style");
+  if (s) setActiveStyleSheet(s);
+}
+
+
+function styleMenu(show) {
+  var m = document.getElementById('style-menu');
+  if (m) toggleShow(m, show);
+}
+
+
+function pageLoad() {
+  addStyleMenu();
+  adjustForFrames();
+  resetStyle();
+  restoreCollapsed();
+}
+
diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/hslogo-16.png b/docs/haddock/tensorflow-opgen-0.1.0.0/hslogo-16.png
new file mode 100644
index 0000000000000000000000000000000000000000..0ff8579fbd897417b0d6dad6e920f8882138a7c0
GIT binary patch
literal 1684
zcmV;F25b3=P)<h;3K|Lk000e1NJLTq000*N000mO1^@s6{V(Jg0008+X+uL$Nkc;*
zP;zf(X>4Tx0C)j~RL^S@K@|QrZmG~B2wH0nvUrdpNm;9CMbtL^5n^i$+aI<?kYqDS
z)^0ZI?k2URh~(nYD&9ryp$Gqf9(wR(FG_`4MJV2@C}N=(E%D86Otz)!z|Ma2-Z$@k
zZ+71R4RX>n^?(HA4aZWV5ov6ELTdbo0FI&wK{O>*+w4vx20?>!`FrQsdJlnHR>OPy
zcd~b_n$otK2Za4V;76L-DzNVtaSB-y0*E}{p()372;bw_^6ZZ}PI-92wGS&j#91PI
zKs7DSe@(bk%_Y-7gGe}(^>I=@oY#w#*Bu9GZf3^F5WP>3rn}7Ut74&?PWBFvy`A)a
zPP5)V!Xd&78LdA?xQ(9mjMYElVd13a#D+Z_7&Y|xU=_C-srWU*6kiZcC!$nw*)9$7
zn6CX+@=AhmkT}X@VSsa5NKe;HZuq)~1$`#h6R+ZTR#D-3j}vF!)ZOnz+5)dI4jl{{
z44Mr{P!L4~VVJN`K!!XTF*LGrKO?IK8<Tr7btG!LbYeuYL3=jbJ-1P$-8}v%B5{;M
zwFr{@LH;VQ$xr2Z`O93e*jD$Ht(%&<^58qg<(at}9@o>z<8w`3e3jI8lUGNUta*C8
zn(P`s>{pjD=7Kek#B;Fw@hxAK%$F&Q6vg9J^Xf~4by_hu-=A!MJ3Znq&n~srbFGPs
zH&&aMXZ>nO`|hf|ljc?VPhR!${AbO?W8x_>CU%PFA&Hm8F7cAsOREdwU~R_;ot1_u
z(ruCYB-LPGn!NQdT|ZlRy+(fw^-+`=%+gee_kY4FWHg<*4sZI8+sFJD<oAl_pC|$^
zY~aY5x@}W&?+~G7rEYVs0vEs0eekw!YomTR`~+A$s}`+NHJ>270UUORdLHO0nA4V)
z%{fwsET5CQ>B?eK%uw4yQc~9?*JVo<vzb+5>2}ze(;aRcp*ceL#HUJSllrgm5wQKR
zQu+C;QrUh^8rFfA`ftFz{YAidi-`aL010qNS#tmY4c7nw4c7reD4Tcy00T@(L_t(I
z5sj2vNEA^R$7gqDc6T=2^@fUA2(c`MltuL5<|<Z1YDGjBEr^N`3VJpQ$`C4&9(*Z6
z=|*B$y;-1^F6vXRAd7+(;)77Qps2euv)AeWmj>KW>RWz$&YbU@|M|{$E*8Tu-Ux!w
z1Y*Dr&Ubfr&v-nZaaB{3ilRumrjPmk{sZvQEWlW+{o~IH|8)=s6c#X9S5s5d%<Agu
zgM$153Ttk++ZPA~ZqgqN1|^pNV-|?9-zo48n9+Yiy$*su5czgp*VFj6!1h&DRZR>J
z4@)&QH5|xQY-)^L1n<EM!01(Q6kM8}ojndgRD~2KVIa@>0pTRu0Lx9`08YTjTwn^6
z0;b1+aQ@)n;Em$q;=7BBi)v0zj&o^g>0Whp^_^5IbxIUP8C@y9;R?*Ouu}rmfxbU=
zwtWVNke-m!=`7bYEhWpcI5#)9qp`8E0lr6IQ)ARL3Ui}Af@grj8aN1=r>Cb+prlzO
zNfJs*N_tUm2ZL<f%*@E?H`Ak_J`Q#Qri<;K{{H@Bg@uKWS5{WE+}vCx91geD*4DNm
zJ?D^~dGz7lWtbWH`o<0O#|{R6h9M*GMpjnV?Y_RgatOb~1le#ikPKTNIn%<;=>%5*
zPmL2??da$TR904gL(VDAQ<BE%k@5ZA>-Fv_Dk}Pdw*4T(%*f4MKLRg=4ekMjhe2mW
zMFsBwg%ftWT}0kxRaIk1k7qJ8*#cKB;Ft{i`zVIs-Nqge;!!Ld7#O&Qqu7e0sJmP)
z$MW*>L$vSB&dxp@iA3U9fo)-7!Czlr{|o7<LYh`Fd@X2-b0fIa*Wdy$({Zt@t7`&L
zhoaGF!l5{1P1Qm;<GaSj#=Oa7atf1JZ8n=F9*@uC?b(X48O9kPoH~h;!o2=WM?ZLl
z$Lj;4K~)fAdW@^N%nzT>Hv{1oyg3xsu%gn@(b1>$;SM-ZaQ`HV=V0s;lr%d8bd;xY
zGw<i+<$17ay36gdrqk(Lxc1|{y}dSeM5OIV(FjhBg5#OM<!Z*8(@<7crlVw+B-8R1
zl{3J6w9fYQ^vojauaRgZ=~SE&CK!R3pCe!v3>Nvm3=Iu=tyXIgtJnf@A(2S@M140N
ew{UA~tMxaJq;$xaSSi*30000<MNUMnLSTa0p)&^n

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/index-frames.html b/docs/haddock/tensorflow-opgen-0.1.0.0/index-frames.html
new file mode 100644
index 0000000..a8ff440
--- /dev/null
+++ b/docs/haddock/tensorflow-opgen-0.1.0.0/index-frames.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-opgen-0.1.0.0: Code generation for TensorFlow operations.</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-list"><p class="caption">Modules</p><ul><li class="module"><a href="TensorFlow-OpGen.html" target="main">TensorFlow.OpGen</a></li><li class="module"><a href="TensorFlow-OpGen-AttrVal.html" target="main">TensorFlow.OpGen.AttrVal</a></li></ul></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/index.html b/docs/haddock/tensorflow-opgen-0.1.0.0/index.html
new file mode 100644
index 0000000..de25a07
--- /dev/null
+++ b/docs/haddock/tensorflow-opgen-0.1.0.0/index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-opgen-0.1.0.0: Code generation for TensorFlow operations.</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-opgen-0.1.0.0: Code generation for TensorFlow operations.</p></div><div id="content"><div id="description"><h1>tensorflow-opgen-0.1.0.0: Code generation for TensorFlow operations.</h1><div class="doc"><p>Please see README.md</p></div></div><div id="module-list"><p class="caption">Modules</p><ul><li><span id="control.n.1" class="module collapser" onclick="toggleSection('n.1')">TensorFlow</span><ul id="section.n.1" class="show"><li><span class="module"><span id="control.n.1.1" class="collapser" onclick="toggleSection('n.1.1')">&nbsp;</span><a href="TensorFlow-OpGen.html">TensorFlow.OpGen</a></span><ul id="section.n.1.1" class="show"><li><span class="module"><a href="TensorFlow-OpGen-AttrVal.html">TensorFlow.OpGen.AttrVal</a></span></li></ul></li></ul></li></ul></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/mini_TensorFlow-OpGen-AttrVal.html b/docs/haddock/tensorflow-opgen-0.1.0.0/mini_TensorFlow-OpGen-AttrVal.html
new file mode 100644
index 0000000..c945e9a
--- /dev/null
+++ b/docs/haddock/tensorflow-opgen-0.1.0.0/mini_TensorFlow-OpGen-AttrVal.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.OpGen.AttrVal</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.OpGen.AttrVal</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-OpGen-AttrVal.html#t:AttrDef" target="main">AttrDef</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-OpGen-AttrVal.html#t:AttrCase" target="main">AttrCase</a> f</p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-OpGen-AttrVal.html#t:AttrTemplate" target="main">AttrTemplate</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-OpGen-AttrVal.html#t:Template" target="main">Template</a> a</p></div><div class="top"><p class="src"><a href="TensorFlow-OpGen-AttrVal.html#v:attrDef" target="main">attrDef</a></p></div><div class="top"><p class="src"><a href="TensorFlow-OpGen-AttrVal.html#v:attrOriginal" target="main">attrOriginal</a></p></div><div class="top"><p class="src"><a href="TensorFlow-OpGen-AttrVal.html#v:attrTemplate" target="main">attrTemplate</a></p></div><div class="top"><p class="src"><a href="TensorFlow-OpGen-AttrVal.html#v:templateDefault" target="main">templateDefault</a></p></div><div class="top"><p class="src"><a href="TensorFlow-OpGen-AttrVal.html#v:templateRestrictions" target="main">templateRestrictions</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/mini_TensorFlow-OpGen.html b/docs/haddock/tensorflow-opgen-0.1.0.0/mini_TensorFlow-OpGen.html
new file mode 100644
index 0000000..694d0c0
--- /dev/null
+++ b/docs/haddock/tensorflow-opgen-0.1.0.0/mini_TensorFlow-OpGen.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.OpGen</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.OpGen</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-OpGen.html#t:OpGenFlags" target="main">OpGenFlags</a></p></div><div class="top"><p class="src"><a href="TensorFlow-OpGen.html#v:docOpList" target="main">docOpList</a></p></div><div class="top"><p class="src"><a href="TensorFlow-OpGen.html#v:flagParser" target="main">flagParser</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/minus.gif b/docs/haddock/tensorflow-opgen-0.1.0.0/minus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..1deac2fe1a42e35b994f1b855488f392c50f6a89
GIT binary patch
literal 56
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS53HJ0a;6C;`j@u!^I(6E|
Jh4LZ{)&P#)4$%Mr

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/ocean.css b/docs/haddock/tensorflow-opgen-0.1.0.0/ocean.css
new file mode 100644
index 0000000..1110b40
--- /dev/null
+++ b/docs/haddock/tensorflow-opgen-0.1.0.0/ocean.css
@@ -0,0 +1,600 @@
+/* @group Fundamentals */
+
+* { margin: 0; padding: 0 }
+
+/* Is this portable? */
+html {
+  background-color: white;
+  width: 100%;
+  height: 100%;
+}
+
+body {
+  background: white;
+  color: black;
+  text-align: left;
+  min-height: 100%;
+  position: relative;
+}
+
+p {
+  margin: 0.8em 0;
+}
+
+ul, ol {
+  margin: 0.8em 0 0.8em 2em;
+}
+
+dl {
+  margin: 0.8em 0;
+}
+
+dt {
+  font-weight: bold;
+}
+dd {
+  margin-left: 2em;
+}
+
+a { text-decoration: none; }
+a[href]:link { color: rgb(196,69,29); }
+a[href]:visited { color: rgb(171,105,84); }
+a[href]:hover { text-decoration:underline; }
+
+/* @end */
+
+/* @group Fonts & Sizes */
+
+/* Basic technique & IE workarounds from YUI 3
+   For reasons, see:
+      http://yui.yahooapis.com/3.1.1/build/cssfonts/fonts.css
+ */
+
+body {
+	font:13px/1.4 sans-serif;
+	*font-size:small; /* for IE */
+	*font:x-small; /* for IE in quirks mode */
+}
+
+h1 { font-size: 146.5%; /* 19pt */ }
+h2 { font-size: 131%;   /* 17pt */ }
+h3 { font-size: 116%;   /* 15pt */ }
+h4 { font-size: 100%;   /* 13pt */ }
+h5 { font-size: 100%;   /* 13pt */ }
+
+select, input, button, textarea {
+	font:99% sans-serif;
+}
+
+table {
+	font-size:inherit;
+	font:100%;
+}
+
+pre, code, kbd, samp, tt, .src {
+	font-family:monospace;
+	*font-size:108%;
+	line-height: 124%;
+}
+
+.links, .link {
+  font-size: 85%; /* 11pt */
+}
+
+#module-header .caption {
+  font-size: 182%; /* 24pt */
+}
+
+.info  {
+  font-size: 85%; /* 11pt */
+}
+
+#table-of-contents, #synopsis  {
+  /* font-size: 85%; /* 11pt */
+}
+
+
+/* @end */
+
+/* @group Common */
+
+.caption, h1, h2, h3, h4, h5, h6 {
+  font-weight: bold;
+  color: rgb(78,98,114);
+  margin: 0.8em 0 0.4em;
+}
+
+* + h1, * + h2, * + h3, * + h4, * + h5, * + h6 {
+  margin-top: 2em;
+}
+
+h1 + h2, h2 + h3, h3 + h4, h4 + h5, h5 + h6 {
+  margin-top: inherit;
+}
+
+ul.links {
+  list-style: none;
+  text-align: left;
+  float: right;
+  display: inline-table;
+  margin: 0 0 0 1em;
+}
+
+ul.links li {
+  display: inline;
+  border-left: 1px solid #d5d5d5;
+  white-space: nowrap;
+  padding: 0;
+}
+
+ul.links li a {
+  padding: 0.2em 0.5em;
+}
+
+.hide { display: none; }
+.show { display: inherit; }
+.clear { clear: both; }
+
+.collapser {
+  background-image: url(minus.gif);
+  background-repeat: no-repeat;
+}
+.expander {
+  background-image: url(plus.gif);
+  background-repeat: no-repeat;
+}
+p.caption.collapser,
+p.caption.expander {
+  background-position: 0 0.4em;
+}
+.collapser, .expander {
+  padding-left: 14px;
+  margin-left: -14px;
+  cursor: pointer;
+}
+
+pre {
+  padding: 0.25em;
+  margin: 0.8em 0;
+  background: rgb(229,237,244);
+  overflow: auto;
+  border-bottom: 0.25em solid white;
+  /* white border adds some space below the box to compensate
+     for visual extra space that paragraphs have between baseline
+     and the bounding box */
+}
+
+.src {
+  background: #f0f0f0;
+  padding: 0.2em 0.5em;
+}
+
+.keyword { font-weight: normal; }
+.def { font-weight: bold; }
+
+
+/* @end */
+
+/* @group Page Structure */
+
+#content {
+  margin: 0 auto;
+  padding: 0 2em 6em;
+}
+
+#package-header {
+  background: rgb(41,56,69);
+  border-top: 5px solid rgb(78,98,114);
+  color: #ddd;
+  padding: 0.2em;
+  position: relative;
+  text-align: left;
+}
+
+#package-header .caption {
+  background: url(hslogo-16.png) no-repeat 0em;
+  color: white;
+  margin: 0 2em;
+  font-weight: normal;
+  font-style: normal;
+  padding-left: 2em;
+}
+
+#package-header a:link, #package-header a:visited { color: white; }
+#package-header a:hover { background: rgb(78,98,114); }
+
+#module-header .caption {
+  color: rgb(78,98,114);
+  font-weight: bold;
+  border-bottom: 1px solid #ddd;
+}
+
+table.info {
+  float: right;
+  padding: 0.5em 1em;
+  border: 1px solid #ddd;
+  color: rgb(78,98,114);
+  background-color: #fff;
+  max-width: 40%;
+  border-spacing: 0;
+  position: relative;
+  top: -0.5em;
+  margin: 0 0 0 2em;
+}
+
+.info th {
+	padding: 0 1em 0 0;
+}
+
+div#style-menu-holder {
+  position: relative;
+  z-index: 2;
+  display: inline;
+}
+
+#style-menu {
+  position: absolute;
+  z-index: 1;
+  overflow: visible;
+  background: #374c5e;
+  margin: 0;
+  text-align: center;
+  right: 0;
+  padding: 0;
+  top: 1.25em;
+}
+
+#style-menu li {
+	display: list-item;
+	border-style: none;
+	margin: 0;
+	padding: 0;
+	color: #000;
+	list-style-type: none;
+}
+
+#style-menu li + li {
+	border-top: 1px solid #919191;
+}
+
+#style-menu a {
+  width: 6em;
+  padding: 3px;
+  display: block;
+}
+
+#footer {
+  background: #ddd;
+  border-top: 1px solid #aaa;
+  padding: 0.5em 0;
+  color: #666;
+  text-align: center;
+  position: absolute;
+  bottom: 0;
+  width: 100%;
+  height: 3em;
+}
+
+/* @end */
+
+/* @group Front Matter */
+
+#table-of-contents {
+  float: right;
+  clear: right;
+  background: #faf9dc;
+  border: 1px solid #d8d7ad;
+  padding: 0.5em 1em;
+  max-width: 20em;
+  margin: 0.5em 0 1em 1em;
+}
+
+#table-of-contents .caption {
+  text-align: center;
+  margin: 0;
+}
+
+#table-of-contents ul {
+  list-style: none;
+  margin: 0;
+}
+
+#table-of-contents ul ul {
+  margin-left: 2em;
+}
+
+#description .caption {
+  display: none;
+}
+
+#synopsis {
+  display: none;
+}
+
+.no-frame #synopsis {
+  display: block;
+  position: fixed;
+  right: 0;
+  height: 80%;
+  top: 10%;
+  padding: 0;
+  max-width: 75%;
+}
+
+#synopsis .caption {
+  float: left;
+  width: 29px;
+  color: rgba(255,255,255,0);
+  height: 110px;
+  margin: 0;
+  font-size: 1px;
+  padding: 0;
+}
+
+#synopsis p.caption.collapser {
+  background: url(synopsis.png) no-repeat -64px -8px;
+}
+
+#synopsis p.caption.expander {
+  background: url(synopsis.png) no-repeat 0px -8px;
+}
+
+#synopsis ul {
+  height: 100%;
+  overflow: auto;
+  padding: 0.5em;
+  margin: 0;
+}
+
+#synopsis ul ul {
+  overflow: hidden;
+}
+
+#synopsis ul,
+#synopsis ul li.src {
+  background-color: #faf9dc;
+  white-space: nowrap;
+  list-style: none;
+  margin-left: 0;
+}
+
+/* @end */
+
+/* @group Main Content */
+
+#interface div.top { margin: 2em 0; }
+#interface h1 + div.top,
+#interface h2 + div.top,
+#interface h3 + div.top,
+#interface h4 + div.top,
+#interface h5 + div.top {
+ 	margin-top: 1em;
+}
+#interface p.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface td.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface span.fixity {
+  color: #919191;
+  border-left: 1px solid #919191;
+  padding: 0.2em 0.5em 0.2em 0.5em;
+  margin: 0 -1em 0 1em;
+}
+
+#interface span.rightedge {
+  border-left: 1px solid #919191;
+  padding: 0.2em 0 0.2em 0;
+  margin: 0 0 0 1em;
+}
+
+#interface table { border-spacing: 2px; }
+#interface td {
+  vertical-align: top;
+  padding-left: 0.5em;
+}
+#interface td.src {
+  white-space: nowrap;
+}
+#interface td.doc p {
+  margin: 0;
+}
+#interface td.doc p + p {
+  margin-top: 0.8em;
+}
+
+.clearfix:after {
+  clear: both;
+  content: " ";
+  display: block;
+  height: 0;
+  visibility: hidden;
+}
+
+.subs dl {
+  margin: 0;
+}
+
+.subs dt {
+  float: left;
+  clear: left;
+  display: block;
+  margin: 1px 0;
+}
+
+.subs dd {
+  float: right;
+  width: 90%;
+  display: block;
+  padding-left: 0.5em;
+  margin-bottom: 0.5em;
+}
+
+.subs dd.empty {
+  display: none;
+}
+
+.subs dd p {
+  margin: 0;
+}
+
+/* Render short-style data instances */
+.inst ul {
+  height: 100%;
+  padding: 0.5em;
+  margin: 0;
+}
+
+.inst, .inst li {
+  list-style: none;
+  margin-left: 1em;
+}
+
+/* Workaround for bug in Firefox (issue #384) */
+.inst-left {
+  float: left;
+}
+
+.top p.src {
+  border-top: 1px solid #ccc;
+}
+
+.subs, .doc {
+  /* use this selector for one level of indent */
+  padding-left: 2em;
+}
+
+.warning {
+  color: red;
+}
+
+.arguments {
+  margin-top: -0.4em;
+}
+.arguments .caption {
+  display: none;
+}
+
+.fields { padding-left: 1em; }
+
+.fields .caption { display: none; }
+
+.fields p { margin: 0 0; }
+
+/* this seems bulky to me
+.methods, .constructors {
+  background: #f8f8f8;
+  border: 1px solid #eee;
+}
+*/
+
+/* @end */
+
+/* @group Auxillary Pages */
+
+
+.extension-list {
+    list-style-type: none;
+    margin-left: 0;
+}
+
+#mini {
+  margin: 0 auto;
+  padding: 0 1em 1em;
+}
+
+#mini > * {
+  font-size: 93%; /* 12pt */
+}
+
+#mini #module-list .caption,
+#mini #module-header .caption {
+  font-size: 125%; /* 15pt */
+}
+
+#mini #interface h1,
+#mini #interface h2,
+#mini #interface h3,
+#mini #interface h4 {
+  font-size: 109%; /* 13pt */
+  margin: 1em 0 0;
+}
+
+#mini #interface .top,
+#mini #interface .src {
+  margin: 0;
+}
+
+#mini #module-list ul {
+  list-style: none;
+  margin: 0;
+}
+
+#alphabet ul {
+	list-style: none;
+	padding: 0;
+	margin: 0.5em 0 0;
+	text-align: center;
+}
+
+#alphabet li {
+	display: inline;
+	margin: 0 0.25em;
+}
+
+#alphabet a {
+	font-weight: bold;
+}
+
+#index .caption,
+#module-list .caption { font-size: 131%; /* 17pt */ }
+
+#index table {
+  margin-left: 2em;
+}
+
+#index .src {
+  font-weight: bold;
+}
+#index .alt {
+  font-size: 77%; /* 10pt */
+  font-style: italic;
+  padding-left: 2em;
+}
+
+#index td + td {
+  padding-left: 1em;
+}
+
+#module-list ul {
+  list-style: none;
+  margin: 0 0 0 2em;
+}
+
+#module-list li {
+  clear: right;
+}
+
+#module-list span.collapser,
+#module-list span.expander {
+  background-position: 0 0.3em;
+}
+
+#module-list .package {
+  float: right;
+}
+
+/* @end */
diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/plus.gif b/docs/haddock/tensorflow-opgen-0.1.0.0/plus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..2d15c14173d23f664b955cd24f51c82f5f09d91d
GIT binary patch
literal 59
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS5iS+a@;NG@rPSdH*>gbBX
M^XE!9f*2UA0nx1yDgXcg

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/src/TensorFlow-OpGen-AttrVal.html b/docs/haddock/tensorflow-opgen-0.1.0.0/src/TensorFlow-OpGen-AttrVal.html
new file mode 100644
index 0000000..2cbe9fb
--- /dev/null
+++ b/docs/haddock/tensorflow-opgen-0.1.0.0/src/TensorFlow-OpGen-AttrVal.html
@@ -0,0 +1,131 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/OpGen/AttrVal.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE OverloadedStrings #-}</span>
+<a name="line-16"></a>
+<a name="line-17"></a><span class='hs-comment'>-- | Wrapping of TensorFlow attributes into Haskell entities.</span>
+<a name="line-18"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>OpGen</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrVal</span>
+<a name="line-19"></a>       <span class='hs-layout'>(</span><span class='hs-conid'>AttrDef</span>
+<a name="line-20"></a>       <span class='hs-layout'>,</span> <span class='hs-conid'>AttrCase</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-21"></a>       <span class='hs-layout'>,</span> <span class='hs-conid'>AttrTemplate</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-22"></a>       <span class='hs-layout'>,</span> <span class='hs-conid'>Template</span>
+<a name="line-23"></a>       <span class='hs-layout'>,</span> <span class='hs-varid'>attrDef</span>
+<a name="line-24"></a>       <span class='hs-layout'>,</span> <span class='hs-varid'>attrOriginal</span>
+<a name="line-25"></a>       <span class='hs-layout'>,</span> <span class='hs-varid'>attrTemplate</span>
+<a name="line-26"></a>       <span class='hs-layout'>,</span> <span class='hs-varid'>templateDefault</span>
+<a name="line-27"></a>       <span class='hs-layout'>,</span> <span class='hs-varid'>templateRestrictions</span>
+<a name="line-28"></a>       <span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-29"></a>
+<a name="line-30"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span> <span class='hs-layout'>(</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-31"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Monoid</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varop'>&lt;&gt;</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-32"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span> <span class='hs-layout'>(</span><span class='hs-conid'>Lens'</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>^.</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-33"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span> <span class='hs-layout'>(</span><span class='hs-varid'>lens</span><span class='hs-layout'>)</span>
+<a name="line-34"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>AttrValue</span>
+<a name="line-35"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>OpDef</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>OpDef</span>
+<a name="line-36"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span> <span class='hs-layout'>(</span><span class='hs-conid'>DataType</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-37"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorShapeProto</span><span class='hs-layout'>)</span>
+<a name="line-38"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>B</span>
+<a name="line-39"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Text</span>
+<a name="line-40"></a>
+<a name="line-41"></a><a name="Template"></a><span class='hs-comment'>-- | Specifies the optional default value and a set of allowed values</span>
+<a name="line-42"></a><a name="Template"></a><span class='hs-comment'>-- for the given type.</span>
+<a name="line-43"></a><a name="Template"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>Template</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Template</span> <span class='hs-layout'>{</span>
+<a name="line-44"></a>    <span class='hs-sel'>_templateDefault</span>      <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Maybe</span> <span class='hs-varid'>a</span>
+<a name="line-45"></a>    <span class='hs-comment'>-- ^ The default value (mandatory if unspecified)</span>
+<a name="line-46"></a>  <span class='hs-layout'>,</span> <span class='hs-sel'>_templateRestrictions</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span>
+<a name="line-47"></a>    <span class='hs-comment'>-- ^ The allowed set of values, empty if no restrictions</span>
+<a name="line-48"></a> <span class='hs-layout'>}</span>
+<a name="line-49"></a>
+<a name="line-50"></a><a name="templateDefault"></a><span class='hs-definition'>templateDefault</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Template</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-conid'>Maybe</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-51"></a><span class='hs-definition'>templateDefault</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_templateDefault</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>g</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>g</span> <span class='hs-layout'>{</span> <span class='hs-sel'>_templateDefault</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span> <span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-52"></a>
+<a name="line-53"></a><a name="templateRestrictions"></a><span class='hs-definition'>templateRestrictions</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Template</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span>
+<a name="line-54"></a><span class='hs-definition'>templateRestrictions</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_templateRestrictions</span>
+<a name="line-55"></a>                            <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>g</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>g</span> <span class='hs-layout'>{</span> <span class='hs-sel'>_templateRestrictions</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span> <span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-56"></a>
+<a name="line-57"></a><a name="UnusedTensor"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>UnusedTensor</span>
+<a name="line-58"></a>
+<a name="line-59"></a><a name="AttrCase"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>AttrCase</span> <span class='hs-varid'>f</span>
+<a name="line-60"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AttrBytes</span> <span class='hs-layout'>(</span><span class='hs-varid'>f</span> <span class='hs-conid'>B</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>          <span class='hs-comment'>-- bytes s = 2; // "string"</span>
+<a name="line-61"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-conid'>AttrInt64</span> <span class='hs-layout'>(</span><span class='hs-varid'>f</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>                 <span class='hs-comment'>-- int64 i = 3; // "int"</span>
+<a name="line-62"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-conid'>AttrFloat</span> <span class='hs-layout'>(</span><span class='hs-varid'>f</span> <span class='hs-conid'>Float</span><span class='hs-layout'>)</span>                 <span class='hs-comment'>-- float f = 4; // "float"</span>
+<a name="line-63"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-conid'>AttrBool</span>  <span class='hs-layout'>(</span><span class='hs-varid'>f</span> <span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>                  <span class='hs-comment'>-- bool b = 5;  // "bool"</span>
+<a name="line-64"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-conid'>AttrType</span>  <span class='hs-layout'>(</span><span class='hs-varid'>f</span> <span class='hs-conid'>DataType</span><span class='hs-layout'>)</span>              <span class='hs-comment'>-- type = 6; // "type"</span>
+<a name="line-65"></a>    <span class='hs-comment'>-- To be translated into TensorFlow.Types.Shape before use.</span>
+<a name="line-66"></a>    <span class='hs-comment'>-- Leaving as a proto to reduce dependencies.</span>
+<a name="line-67"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-conid'>AttrShape</span> <span class='hs-layout'>(</span><span class='hs-varid'>f</span> <span class='hs-conid'>TensorShapeProto</span><span class='hs-layout'>)</span>      <span class='hs-comment'>-- shape = 7; // "shape"</span>
+<a name="line-68"></a>
+<a name="line-69"></a><a name="AttrTemplate"></a><span class='hs-comment'>-- | Type-reified representation of TensorFlow AttrDef.</span>
+<a name="line-70"></a><a name="AttrTemplate"></a><span class='hs-comment'>-- Initially limited to just the types in Op descriptors.</span>
+<a name="line-71"></a><a name="AttrTemplate"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>AttrTemplate</span>
+<a name="line-72"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AttrSingle</span> <span class='hs-layout'>(</span><span class='hs-conid'>AttrCase</span> <span class='hs-conid'>Template</span><span class='hs-layout'>)</span>
+<a name="line-73"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-conid'>AttrList</span> <span class='hs-layout'>(</span><span class='hs-conid'>AttrCase</span> <span class='hs-conid'>[]</span><span class='hs-layout'>)</span>
+<a name="line-74"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-conid'>AttrTensor</span> <span class='hs-conid'>UnusedTensor</span>         <span class='hs-comment'>-- tensor = 8; // "tensor"</span>
+<a name="line-75"></a>
+<a name="line-76"></a><a name="AttrDef"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>AttrDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AttrDef</span> <span class='hs-layout'>{</span>
+<a name="line-77"></a>    <span class='hs-sel'>_attrOriginal</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-comment'>-- ^ the proto this value was created from</span>
+<a name="line-78"></a>  <span class='hs-layout'>,</span> <span class='hs-sel'>_attrTemplate</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>AttrTemplate</span>  <span class='hs-comment'>-- ^ the type of the attribute</span>
+<a name="line-79"></a>  <span class='hs-layout'>}</span>
+<a name="line-80"></a>
+<a name="line-81"></a><a name="attrTemplate"></a><span class='hs-definition'>attrTemplate</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>AttrDef</span> <span class='hs-conid'>AttrTemplate</span>
+<a name="line-82"></a><span class='hs-definition'>attrTemplate</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_attrTemplate</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>g</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>g</span> <span class='hs-layout'>{</span> <span class='hs-sel'>_attrTemplate</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span> <span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-83"></a>
+<a name="line-84"></a><a name="attrOriginal"></a><span class='hs-definition'>attrOriginal</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-conid'>AttrDef</span> <span class='hs-conid'>OpDef'AttrDef</span>
+<a name="line-85"></a><span class='hs-definition'>attrOriginal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_attrOriginal</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>g</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>g</span> <span class='hs-layout'>{</span> <span class='hs-sel'>_attrOriginal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span> <span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-86"></a>
+<a name="line-87"></a><a name="attrDef"></a><span class='hs-definition'>attrDef</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>AttrDef</span>
+<a name="line-88"></a><span class='hs-definition'>attrDef</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AttrDef</span> <span class='hs-varid'>a</span>
+<a name="line-89"></a>                  <span class='hs-varop'>$</span> <span class='hs-varid'>translate</span> <span class='hs-layout'>(</span><span class='hs-varid'>a</span><span class='hs-varop'>^.</span><span class='hs-conid'>OpDef</span><span class='hs-varop'>.</span><span class='hs-varid'>type'</span><span class='hs-layout'>)</span>
+<a name="line-90"></a>                              <span class='hs-layout'>(</span><span class='hs-varid'>a</span><span class='hs-varop'>^.</span><span class='hs-conid'>OpDef</span><span class='hs-varop'>.</span><span class='hs-varid'>defaultValue</span><span class='hs-layout'>)</span>
+<a name="line-91"></a>                              <span class='hs-layout'>(</span><span class='hs-varid'>a</span><span class='hs-varop'>^.</span><span class='hs-varid'>allowedValues</span><span class='hs-layout'>)</span>
+<a name="line-92"></a>
+<a name="line-93"></a><a name="translate"></a><span class='hs-comment'>-- | Converts the given AttrValue with the type given by the string</span>
+<a name="line-94"></a><span class='hs-comment'>-- into the AttrVal if the type is known.</span>
+<a name="line-95"></a><span class='hs-definition'>translate</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>  <span class='hs-comment'>-- ^ one of the TensorFlow type strings</span>
+<a name="line-96"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>AttrValue</span>  <span class='hs-comment'>-- ^ default value</span>
+<a name="line-97"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>AttrValue</span>  <span class='hs-comment'>-- ^ allowed values</span>
+<a name="line-98"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>AttrTemplate</span>
+<a name="line-99"></a><span class='hs-definition'>translate</span> <span class='hs-varid'>t</span> <span class='hs-varid'>defaults</span> <span class='hs-varid'>allowed</span>
+<a name="line-100"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>t</span> <span class='hs-varop'>==</span> <span class='hs-str'>"string"</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>makeVal</span> <span class='hs-conid'>AttrBytes</span> <span class='hs-varid'>maybe's</span> <span class='hs-varid'>s</span>
+<a name="line-101"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>t</span> <span class='hs-varop'>==</span> <span class='hs-str'>"int"</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>makeVal</span> <span class='hs-conid'>AttrInt64</span> <span class='hs-varid'>maybe'i</span> <span class='hs-varid'>i</span>
+<a name="line-102"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>t</span> <span class='hs-varop'>==</span> <span class='hs-str'>"float"</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>makeVal</span> <span class='hs-conid'>AttrFloat</span> <span class='hs-varid'>maybe'f</span> <span class='hs-varid'>f</span>
+<a name="line-103"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>t</span> <span class='hs-varop'>==</span> <span class='hs-str'>"bool"</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>makeVal</span> <span class='hs-conid'>AttrBool</span> <span class='hs-varid'>maybe'b</span> <span class='hs-varid'>b</span>
+<a name="line-104"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>t</span> <span class='hs-varop'>==</span> <span class='hs-str'>"type"</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>makeVal</span> <span class='hs-conid'>AttrType</span> <span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-varid'>maybe'type'</span> <span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-varid'>type'</span>
+<a name="line-105"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>t</span> <span class='hs-varop'>==</span> <span class='hs-str'>"shape"</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>makeVal</span> <span class='hs-conid'>AttrShape</span> <span class='hs-varid'>maybe'shape</span> <span class='hs-varid'>shape</span>
+<a name="line-106"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>t</span> <span class='hs-varop'>==</span> <span class='hs-str'>"tensor"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AttrTensor</span> <span class='hs-varop'>$</span> <span class='hs-varid'>error</span> <span class='hs-str'>"tensor is unimplemented"</span>
+<a name="line-107"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>t</span> <span class='hs-varop'>==</span> <span class='hs-str'>"list(string)"</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>makeList</span> <span class='hs-conid'>AttrBytes</span> <span class='hs-varop'>$</span> <span class='hs-varid'>list</span><span class='hs-varop'>.</span><span class='hs-varid'>s</span>
+<a name="line-108"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>t</span> <span class='hs-varop'>==</span> <span class='hs-str'>"list(int)"</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>makeList</span> <span class='hs-conid'>AttrInt64</span> <span class='hs-varop'>$</span> <span class='hs-varid'>list</span><span class='hs-varop'>.</span><span class='hs-varid'>i</span>
+<a name="line-109"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>t</span> <span class='hs-varop'>==</span> <span class='hs-str'>"list(float)"</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>makeList</span> <span class='hs-conid'>AttrFloat</span> <span class='hs-varop'>$</span> <span class='hs-varid'>list</span><span class='hs-varop'>.</span><span class='hs-varid'>f</span>
+<a name="line-110"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>t</span> <span class='hs-varop'>==</span> <span class='hs-str'>"list(bool)"</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>makeList</span> <span class='hs-conid'>AttrBool</span> <span class='hs-varop'>$</span> <span class='hs-varid'>list</span><span class='hs-varop'>.</span><span class='hs-varid'>b</span>
+<a name="line-111"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>t</span> <span class='hs-varop'>==</span> <span class='hs-str'>"list(type)"</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>makeList</span> <span class='hs-conid'>AttrType</span> <span class='hs-varop'>$</span> <span class='hs-varid'>list</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-varid'>type'</span>
+<a name="line-112"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>t</span> <span class='hs-varop'>==</span> <span class='hs-str'>"list(shape)"</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>makeList</span> <span class='hs-conid'>AttrShape</span> <span class='hs-varop'>$</span> <span class='hs-varid'>list</span><span class='hs-varop'>.</span><span class='hs-varid'>shape</span>
+<a name="line-113"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>t</span> <span class='hs-varop'>==</span> <span class='hs-str'>"list(tensor)"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AttrTensor</span> <span class='hs-varop'>$</span> <span class='hs-varid'>error</span> <span class='hs-str'>"list(tensor) is unimplemented"</span>
+<a name="line-114"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>t</span> <span class='hs-varop'>==</span> <span class='hs-str'>"func"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AttrTensor</span> <span class='hs-varop'>$</span> <span class='hs-varid'>error</span> <span class='hs-str'>"func is unimplemented"</span>
+<a name="line-115"></a>  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>otherwise</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-varop'>$</span> <span class='hs-varid'>show</span> <span class='hs-layout'>(</span><span class='hs-str'>"Unknown attribute type "</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-varop'>++</span>
+<a name="line-116"></a>                        <span class='hs-str'>","</span> <span class='hs-varop'>++</span> <span class='hs-varid'>show</span> <span class='hs-varid'>defaults</span> <span class='hs-varop'>++</span>
+<a name="line-117"></a>                        <span class='hs-str'>","</span> <span class='hs-varop'>++</span> <span class='hs-varid'>show</span> <span class='hs-varid'>allowed</span>
+<a name="line-118"></a>  <span class='hs-keyword'>where</span> <span class='hs-varid'>makeVal</span> <span class='hs-varid'>c</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AttrSingle</span> <span class='hs-varop'>$</span> <span class='hs-varid'>c</span> <span class='hs-varop'>$</span>
+<a name="line-119"></a>                        <span class='hs-conid'>Template</span> <span class='hs-layout'>(</span><span class='hs-varid'>defaults</span><span class='hs-varop'>^.</span><span class='hs-varid'>x</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>allowed</span><span class='hs-varop'>^.</span><span class='hs-varid'>list</span><span class='hs-varop'>.</span><span class='hs-varid'>y</span><span class='hs-layout'>)</span>
+<a name="line-120"></a>        <span class='hs-varid'>makeList</span> <span class='hs-varid'>c</span> <span class='hs-varid'>y</span>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AttrList</span> <span class='hs-varop'>$</span> <span class='hs-varid'>c</span> <span class='hs-varop'>$</span> <span class='hs-varid'>defaults</span><span class='hs-varop'>^.</span><span class='hs-varid'>y</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/src/TensorFlow-OpGen.html b/docs/haddock/tensorflow-opgen-0.1.0.0/src/TensorFlow-OpGen.html
new file mode 100644
index 0000000..7ae17e2
--- /dev/null
+++ b/docs/haddock/tensorflow-opgen-0.1.0.0/src/TensorFlow-OpGen.html
@@ -0,0 +1,468 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/OpGen.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE FlexibleContexts #-}</span>
+<a name="line-16"></a><span class='hs-comment'>{-# LANGUAGE OverloadedStrings #-}</span>
+<a name="line-17"></a><span class='hs-comment'>{-# LANGUAGE TypeFamilies #-}</span>
+<a name="line-18"></a><span class='hs-comment'>-- | Rendering of TensorFlow operations as Haskell functions.</span>
+<a name="line-19"></a>
+<a name="line-20"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>OpGen</span>
+<a name="line-21"></a>  <span class='hs-layout'>(</span> <span class='hs-conid'>OpGenFlags</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-22"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>docOpList</span>
+<a name="line-23"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>flagParser</span><span class='hs-layout'>)</span>
+<a name="line-24"></a>  <span class='hs-keyword'>where</span>
+<a name="line-25"></a>
+<a name="line-26"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Prelude</span> <span class='hs-varid'>hiding</span> <span class='hs-layout'>(</span><span class='hs-varid'>head</span><span class='hs-layout'>,</span> <span class='hs-varid'>tail</span><span class='hs-layout'>)</span>
+<a name="line-27"></a>
+<a name="line-28"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Monad</span> <span class='hs-layout'>(</span><span class='hs-varid'>guard</span><span class='hs-layout'>)</span>
+<a name="line-29"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Char</span> <span class='hs-layout'>(</span><span class='hs-varid'>toLower</span><span class='hs-layout'>,</span> <span class='hs-varid'>toUpper</span><span class='hs-layout'>)</span>
+<a name="line-30"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Foldable</span> <span class='hs-layout'>(</span><span class='hs-varid'>toList</span><span class='hs-layout'>)</span>
+<a name="line-31"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromMaybe</span><span class='hs-layout'>,</span> <span class='hs-varid'>maybeToList</span><span class='hs-layout'>)</span>
+<a name="line-32"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span> <span class='hs-layout'>(</span><span class='hs-varid'>def</span><span class='hs-layout'>,</span> <span class='hs-varid'>showMessage</span><span class='hs-layout'>)</span>
+<a name="line-33"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>List</span><span class='hs-varop'>.</span><span class='hs-conid'>NonEmpty</span> <span class='hs-layout'>(</span><span class='hs-conid'>NonEmpty</span><span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conop'>:|</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-varid'>head</span><span class='hs-layout'>)</span>
+<a name="line-34"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>List</span><span class='hs-varop'>.</span><span class='hs-conid'>NonEmpty</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>NE</span>
+<a name="line-35"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varop'>^.</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>.~</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>&amp;</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-varid'>view</span><span class='hs-layout'>)</span>
+<a name="line-36"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Options</span><span class='hs-varop'>.</span><span class='hs-conid'>Applicative</span> <span class='hs-layout'>(</span><span class='hs-conid'>Parser</span><span class='hs-layout'>,</span> <span class='hs-varid'>help</span><span class='hs-layout'>,</span> <span class='hs-varid'>long</span><span class='hs-layout'>,</span> <span class='hs-varid'>strOption</span><span class='hs-layout'>,</span> <span class='hs-varid'>value</span><span class='hs-layout'>)</span>
+<a name="line-37"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>OpDef</span>
+<a name="line-38"></a>  <span class='hs-layout'>(</span> <span class='hs-conid'>OpList</span>
+<a name="line-39"></a>  <span class='hs-layout'>,</span> <span class='hs-conid'>OpDef</span>
+<a name="line-40"></a>  <span class='hs-layout'>,</span> <span class='hs-conid'>OpDef'ArgDef</span>
+<a name="line-41"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>attr</span>
+<a name="line-42"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>description</span>
+<a name="line-43"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>inputArg</span>
+<a name="line-44"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>name</span>
+<a name="line-45"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>numberAttr</span>
+<a name="line-46"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>op</span>
+<a name="line-47"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>outputArg</span>
+<a name="line-48"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>summary</span>
+<a name="line-49"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>type'</span>
+<a name="line-50"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>typeAttr</span>
+<a name="line-51"></a>  <span class='hs-layout'>)</span>
+<a name="line-52"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span> <span class='hs-layout'>(</span><span class='hs-conid'>DataType</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-53"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>System</span><span class='hs-varop'>.</span><span class='hs-conid'>FilePath</span> <span class='hs-layout'>(</span><span class='hs-varid'>takeBaseName</span><span class='hs-layout'>)</span>
+<a name="line-54"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>OpGen</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrVal</span>
+<a name="line-55"></a>  <span class='hs-layout'>(</span><span class='hs-conid'>AttrDef</span>
+<a name="line-56"></a>  <span class='hs-layout'>,</span> <span class='hs-conid'>AttrCase</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-57"></a>  <span class='hs-layout'>,</span> <span class='hs-conid'>AttrTemplate</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-58"></a>  <span class='hs-layout'>,</span> <span class='hs-conid'>Template</span>
+<a name="line-59"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>attrDef</span>
+<a name="line-60"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>attrOriginal</span>
+<a name="line-61"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>attrTemplate</span>
+<a name="line-62"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>templateDefault</span>
+<a name="line-63"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>templateRestrictions</span>
+<a name="line-64"></a>  <span class='hs-layout'>)</span>
+<a name="line-65"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>PrettyPrint</span><span class='hs-varop'>.</span><span class='hs-conid'>Mainland</span>
+<a name="line-66"></a>  <span class='hs-layout'>(</span> <span class='hs-conid'>Doc</span>
+<a name="line-67"></a>  <span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>&lt;&gt;</span><span class='hs-layout'>)</span>
+<a name="line-68"></a>  <span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>&lt;+&gt;</span><span class='hs-layout'>)</span>
+<a name="line-69"></a>  <span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>&lt;/&gt;</span><span class='hs-layout'>)</span>
+<a name="line-70"></a>  <span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>&lt;+/&gt;</span><span class='hs-layout'>)</span>
+<a name="line-71"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>brackets</span>
+<a name="line-72"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>comma</span>
+<a name="line-73"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>commasep</span>
+<a name="line-74"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>dquotes</span>
+<a name="line-75"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>empty</span>
+<a name="line-76"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>enclose</span>
+<a name="line-77"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>flatten</span>
+<a name="line-78"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>folddoc</span>
+<a name="line-79"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>hang</span>
+<a name="line-80"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>indent</span>
+<a name="line-81"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>int</span>
+<a name="line-82"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>parens</span>
+<a name="line-83"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>sep</span>
+<a name="line-84"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>stack</span>
+<a name="line-85"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>strictText</span>
+<a name="line-86"></a>  <span class='hs-layout'>,</span> <span class='hs-varid'>tuple</span>
+<a name="line-87"></a>  <span class='hs-layout'>)</span>
+<a name="line-88"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Strict</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Map</span>
+<a name="line-89"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Set</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Set</span>
+<a name="line-90"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Text</span>
+<a name="line-91"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Semigroup</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Semigroup</span>
+<a name="line-92"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-layout'>(</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-93"></a>
+<a name="line-94"></a><a name="OpGenFlags"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>OpGenFlags</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OpGenFlags</span>
+<a name="line-95"></a>     <span class='hs-layout'>{</span> <span class='hs-varid'>outputFile</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>String</span>
+<a name="line-96"></a>     <span class='hs-layout'>,</span> <span class='hs-varid'>prefix</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>String</span>
+<a name="line-97"></a>     <span class='hs-layout'>,</span> <span class='hs-varid'>excludeList</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>String</span>
+<a name="line-98"></a>     <span class='hs-layout'>}</span>
+<a name="line-99"></a>
+<a name="line-100"></a><a name="flagParser"></a><span class='hs-definition'>flagParser</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Parser</span> <span class='hs-conid'>OpGenFlags</span>
+<a name="line-101"></a><span class='hs-definition'>flagParser</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OpGenFlags</span>
+<a name="line-102"></a>     <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>strOption</span> <span class='hs-layout'>(</span><span class='hs-varid'>mconcat</span> <span class='hs-keyglyph'>[</span> <span class='hs-varid'>long</span> <span class='hs-str'>"output"</span>
+<a name="line-103"></a>                            <span class='hs-layout'>,</span> <span class='hs-varid'>help</span> <span class='hs-str'>"File to write."</span>
+<a name="line-104"></a>                            <span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-105"></a>     <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>strOption</span> <span class='hs-layout'>(</span><span class='hs-varid'>mconcat</span> <span class='hs-keyglyph'>[</span> <span class='hs-varid'>long</span> <span class='hs-str'>"prefix"</span>
+<a name="line-106"></a>                            <span class='hs-layout'>,</span> <span class='hs-varid'>help</span> <span class='hs-str'>"Haskell package prefix to use"</span>
+<a name="line-107"></a>                            <span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-108"></a>     <span class='hs-varop'>&lt;*&gt;</span> <span class='hs-varid'>strOption</span> <span class='hs-layout'>(</span><span class='hs-varid'>mconcat</span> <span class='hs-keyglyph'>[</span> <span class='hs-varid'>long</span> <span class='hs-str'>"exclude_list"</span>
+<a name="line-109"></a>                            <span class='hs-layout'>,</span> <span class='hs-varid'>value</span> <span class='hs-str'>""</span>
+<a name="line-110"></a>                            <span class='hs-layout'>,</span> <span class='hs-varid'>help</span> <span class='hs-str'>"Comma separated Ops names to ignore"</span>
+<a name="line-111"></a>                            <span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-112"></a>
+<a name="line-113"></a>
+<a name="line-114"></a><a name="docOpList"></a><span class='hs-definition'>docOpList</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpGenFlags</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>OpList</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Doc</span>
+<a name="line-115"></a><span class='hs-definition'>docOpList</span> <span class='hs-varid'>flags</span> <span class='hs-varid'>opList</span> <span class='hs-keyglyph'>=</span>
+<a name="line-116"></a>  <span class='hs-varid'>stack</span> <span class='hs-keyglyph'>[</span> <span class='hs-str'>"{-# LANGUAGE ConstraintKinds #-}"</span>
+<a name="line-117"></a>        <span class='hs-layout'>,</span> <span class='hs-str'>"{-# LANGUAGE DataKinds #-}"</span>
+<a name="line-118"></a>        <span class='hs-layout'>,</span> <span class='hs-str'>"{-# LANGUAGE FlexibleInstances #-}"</span>
+<a name="line-119"></a>        <span class='hs-layout'>,</span> <span class='hs-str'>"{-# LANGUAGE OverloadedStrings #-}"</span>
+<a name="line-120"></a>        <span class='hs-layout'>,</span> <span class='hs-str'>"{-# LANGUAGE RankNTypes #-}"</span>
+<a name="line-121"></a>        <span class='hs-layout'>,</span> <span class='hs-str'>"{-# LANGUAGE ScopedTypeVariables #-}"</span>
+<a name="line-122"></a>        <span class='hs-layout'>,</span> <span class='hs-str'>"module"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>strictText</span> <span class='hs-varid'>moduleName</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-str'>"where"</span>
+<a name="line-123"></a>        <span class='hs-layout'>,</span> <span class='hs-varid'>empty</span>
+<a name="line-124"></a>        <span class='hs-layout'>,</span> <span class='hs-varid'>imports</span>
+<a name="line-125"></a>        <span class='hs-layout'>,</span> <span class='hs-varid'>empty</span>
+<a name="line-126"></a>        <span class='hs-layout'>,</span> <span class='hs-varid'>folddoc</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span> <span class='hs-varop'>&lt;/&gt;</span> <span class='hs-varid'>empty</span> <span class='hs-varop'>&lt;/&gt;</span> <span class='hs-varid'>y</span><span class='hs-layout'>)</span>
+<a name="line-127"></a>                  <span class='hs-layout'>(</span><span class='hs-varid'>map</span> <span class='hs-varid'>renderDef</span> <span class='hs-varop'>$</span>
+<a name="line-128"></a>                   <span class='hs-varid'>filter</span> <span class='hs-layout'>(</span><span class='hs-varid'>not</span> <span class='hs-varop'>.</span> <span class='hs-varid'>flip</span> <span class='hs-varid'>elem</span> <span class='hs-varid'>exclusions</span> <span class='hs-varop'>.</span> <span class='hs-varid'>view</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span> <span class='hs-varop'>$</span>
+<a name="line-129"></a>                   <span class='hs-varid'>toList</span> <span class='hs-varop'>$</span> <span class='hs-varid'>opList</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>op</span><span class='hs-layout'>)</span>
+<a name="line-130"></a>        <span class='hs-keyglyph'>]</span>
+<a name="line-131"></a>  <span class='hs-keyword'>where</span> <span class='hs-varid'>moduleName</span> <span class='hs-keyglyph'>=</span>
+<a name="line-132"></a>            <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>pack</span> <span class='hs-layout'>(</span><span class='hs-varid'>prefix</span> <span class='hs-varid'>flags</span><span class='hs-layout'>)</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-str'>"."</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-varid'>camelCase</span>
+<a name="line-133"></a>             <span class='hs-comment'>-- Discards the optional trailing _op_lib</span>
+<a name="line-134"></a>            <span class='hs-layout'>(</span><span class='hs-varid'>fromMaybe</span> <span class='hs-varid'>shortName</span> <span class='hs-layout'>(</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>stripSuffix</span> <span class='hs-str'>"_op_lib"</span> <span class='hs-varid'>shortName</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-135"></a>        <span class='hs-varid'>shortName</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>pack</span> <span class='hs-layout'>(</span><span class='hs-varid'>takeBaseName</span> <span class='hs-varop'>$</span> <span class='hs-varid'>outputFile</span> <span class='hs-varid'>flags</span><span class='hs-layout'>)</span>
+<a name="line-136"></a>        <span class='hs-varid'>exclusions</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>splitOn</span> <span class='hs-str'>","</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>pack</span> <span class='hs-varop'>$</span> <span class='hs-varid'>excludeList</span> <span class='hs-varid'>flags</span>
+<a name="line-137"></a>
+<a name="line-138"></a><a name="camelCase"></a><span class='hs-definition'>camelCase</span> <span class='hs-varid'>s</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>concat</span> <span class='hs-varop'>$</span> <span class='hs-varid'>map</span> <span class='hs-varid'>upCase</span>
+<a name="line-139"></a>                          <span class='hs-varop'>$</span> <span class='hs-varid'>filter</span> <span class='hs-layout'>(</span><span class='hs-varop'>/=</span> <span class='hs-str'>"ops"</span><span class='hs-layout'>)</span>
+<a name="line-140"></a>                          <span class='hs-varop'>$</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>splitOn</span> <span class='hs-str'>"_"</span> <span class='hs-varid'>s</span>
+<a name="line-141"></a>
+<a name="line-142"></a><a name="upCase"></a><span class='hs-comment'>-- | Upper-case the given text.</span>
+<a name="line-143"></a><span class='hs-definition'>upCase</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Text</span>
+<a name="line-144"></a><span class='hs-definition'>upCase</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>forceCase</span> <span class='hs-varid'>toUpper</span>
+<a name="line-145"></a>
+<a name="line-146"></a><a name="lowCase"></a><span class='hs-comment'>-- | Lower-case the given name, and prevent it from overlapping with a reserved</span>
+<a name="line-147"></a><span class='hs-comment'>-- Haskell name.</span>
+<a name="line-148"></a><span class='hs-definition'>lowCase</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Text</span>
+<a name="line-149"></a><span class='hs-definition'>lowCase</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>replaceReservedName</span> <span class='hs-varop'>.</span> <span class='hs-varid'>forceCase</span> <span class='hs-varid'>toLower</span>
+<a name="line-150"></a>
+<a name="line-151"></a><a name="forceCase"></a><span class='hs-definition'>forceCase</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>Char</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Char</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Text</span>
+<a name="line-152"></a><span class='hs-definition'>forceCase</span> <span class='hs-varid'>convert</span> <span class='hs-varid'>s</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>maybe</span> <span class='hs-str'>""</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-layout'>(</span><span class='hs-varid'>c</span><span class='hs-layout'>,</span> <span class='hs-varid'>cs</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>cons</span> <span class='hs-layout'>(</span><span class='hs-varid'>convert</span> <span class='hs-varid'>c</span><span class='hs-layout'>)</span> <span class='hs-varid'>cs</span><span class='hs-layout'>)</span>
+<a name="line-153"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>uncons</span> <span class='hs-varid'>s</span><span class='hs-layout'>)</span>
+<a name="line-154"></a>
+<a name="line-155"></a><a name="imports"></a><span class='hs-definition'>imports</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>stack</span> <span class='hs-keyglyph'>[</span>
+<a name="line-156"></a>      <span class='hs-str'>"import Data.ByteString (ByteString)"</span>
+<a name="line-157"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"import Data.Complex (Complex)"</span>
+<a name="line-158"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"import Data.Int (Int8, Int16, Int32, Int64)"</span>
+<a name="line-159"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"import Data.Word (Word8, Word16)"</span>
+<a name="line-160"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"import Lens.Family2 ((.~), (&amp;))"</span>
+<a name="line-161"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"import TensorFlow.Build"</span>
+<a name="line-162"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"import TensorFlow.BuildOp"</span>
+<a name="line-163"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"import TensorFlow.Tensor"</span>
+<a name="line-164"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"import TensorFlow.Types"</span>
+<a name="line-165"></a>      <span class='hs-keyglyph'>]</span>
+<a name="line-166"></a>
+<a name="line-167"></a><a name="renderDef"></a><span class='hs-definition'>renderDef</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Doc</span>
+<a name="line-168"></a><span class='hs-definition'>renderDef</span> <span class='hs-varid'>d</span> <span class='hs-keyglyph'>=</span>
+<a name="line-169"></a>  <span class='hs-varid'>stack</span> <span class='hs-keyglyph'>[</span>
+<a name="line-170"></a>      <span class='hs-varid'>haddocks</span>
+<a name="line-171"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>n</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-str'>"::"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>hang</span> <span class='hs-num'>0</span> <span class='hs-layout'>(</span><span class='hs-varid'>typeSig</span> <span class='hs-varid'>d</span><span class='hs-layout'>)</span>
+<a name="line-172"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>n</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>hang</span> <span class='hs-num'>0</span> <span class='hs-varid'>args</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-str'>"|"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>funcGuard</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-str'>"="</span> <span class='hs-varop'>&lt;/&gt;</span>  <span class='hs-comment'>-- args are indented</span>
+<a name="line-173"></a>            <span class='hs-comment'>-- the body needs to be indented wrt the name</span>
+<a name="line-174"></a>            <span class='hs-varid'>indent</span> <span class='hs-varid'>indentation</span> <span class='hs-varid'>functionBody</span>
+<a name="line-175"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>extras</span>  <span class='hs-comment'>-- just for debug</span>
+<a name="line-176"></a>    <span class='hs-keyglyph'>]</span>
+<a name="line-177"></a>  <span class='hs-keyword'>where</span>
+<a name="line-178"></a>    <span class='hs-varid'>n</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>strictText</span> <span class='hs-varop'>$</span> <span class='hs-varid'>fixOpName</span> <span class='hs-layout'>(</span><span class='hs-varid'>d</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span>
+<a name="line-179"></a>    <span class='hs-varid'>args</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>sep</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>hsName</span> <span class='hs-keyglyph'>|</span> <span class='hs-layout'>(</span><span class='hs-keyword'>_</span><span class='hs-layout'>,</span> <span class='hs-varid'>hsName</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>mandatoryAttrs</span><span class='hs-keyglyph'>]</span> <span class='hs-varop'>++</span> <span class='hs-varid'>tensorArgs</span>
+<a name="line-180"></a>    <span class='hs-varid'>tensorArgs</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>strictText</span> <span class='hs-varop'>$</span> <span class='hs-varid'>lowCase</span> <span class='hs-layout'>(</span><span class='hs-varid'>a</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>d</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>inputArg</span><span class='hs-keyglyph'>]</span>
+<a name="line-181"></a>    <span class='hs-varid'>fixOpName</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lowCase</span>
+<a name="line-182"></a>    <span class='hs-varid'>funcGuard</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"eqLengthGuard"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>brackets</span> <span class='hs-layout'>(</span><span class='hs-varid'>commasep</span> <span class='hs-varid'>entries</span><span class='hs-layout'>)</span>
+<a name="line-183"></a>      <span class='hs-keyword'>where</span>
+<a name="line-184"></a>        <span class='hs-varid'>entries</span> <span class='hs-keyglyph'>=</span>
+<a name="line-185"></a>            <span class='hs-keyglyph'>[</span> <span class='hs-varid'>parens</span> <span class='hs-varop'>$</span> <span class='hs-varid'>quotedText</span> <span class='hs-varid'>nAttr</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-varid'>comma</span> <span class='hs-varop'>&lt;+&gt;</span>
+<a name="line-186"></a>              <span class='hs-varid'>brackets</span> <span class='hs-layout'>(</span><span class='hs-varid'>commasep</span> <span class='hs-varop'>$</span> <span class='hs-varid'>toList</span> <span class='hs-varop'>$</span>
+<a name="line-187"></a>              <span class='hs-conid'>NE</span><span class='hs-varop'>.</span><span class='hs-varid'>map</span> <span class='hs-varid'>renderTensorName</span> <span class='hs-varid'>tensorNames</span><span class='hs-layout'>)</span>
+<a name="line-188"></a>            <span class='hs-keyglyph'>|</span> <span class='hs-layout'>(</span><span class='hs-varid'>nAttr</span><span class='hs-layout'>,</span> <span class='hs-varid'>tensorNames</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>toList</span> <span class='hs-varop'>$</span> <span class='hs-varid'>numberAttrMap</span> <span class='hs-varid'>d</span>
+<a name="line-189"></a>            <span class='hs-keyglyph'>]</span>
+<a name="line-190"></a>        <span class='hs-varid'>renderTensorName</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>parens</span> <span class='hs-varop'>$</span> <span class='hs-varid'>quotedText</span> <span class='hs-varid'>x</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-varid'>comma</span> <span class='hs-varop'>&lt;+&gt;</span>
+<a name="line-191"></a>                             <span class='hs-str'>"length"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>strictText</span> <span class='hs-varid'>x</span>
+<a name="line-192"></a>    <span class='hs-comment'>-- Uses hang 0 to align the argument vertically on multiple lines.</span>
+<a name="line-193"></a>    <span class='hs-varid'>functionBody</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>buildFunction</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>parens</span> <span class='hs-layout'>(</span><span class='hs-varid'>hang</span> <span class='hs-num'>0</span> <span class='hs-layout'>(</span><span class='hs-varid'>stack</span> <span class='hs-varid'>buildOpParts</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-194"></a>                                 <span class='hs-varop'>&lt;/&gt;</span> <span class='hs-varid'>indent</span> <span class='hs-varid'>indentation</span> <span class='hs-layout'>(</span><span class='hs-varid'>sep</span> <span class='hs-varid'>tensorArgs</span><span class='hs-layout'>)</span>
+<a name="line-195"></a>    <span class='hs-varid'>buildFunction</span>
+<a name="line-196"></a>        <span class='hs-keyglyph'>|</span> <span class='hs-varid'>null</span> <span class='hs-varid'>outputListsSizes</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"buildOp"</span>
+<a name="line-197"></a>        <span class='hs-keyglyph'>|</span> <span class='hs-varid'>otherwise</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"buildListOp"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>brackets</span> <span class='hs-layout'>(</span><span class='hs-varid'>commasep</span> <span class='hs-varid'>outputListsSizes</span><span class='hs-layout'>)</span>
+<a name="line-198"></a>    <span class='hs-varid'>outputListsSizes</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span> <span class='hs-varid'>strictText</span> <span class='hs-varid'>numberAttrName</span>
+<a name="line-199"></a>                       <span class='hs-keyglyph'>|</span> <span class='hs-varid'>o</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>d</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>outputArg</span>
+<a name="line-200"></a>                       <span class='hs-layout'>,</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>numberAttrName</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>o</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>numberAttr</span>
+<a name="line-201"></a>                       <span class='hs-layout'>,</span> <span class='hs-varid'>not</span> <span class='hs-layout'>(</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>null</span> <span class='hs-varid'>numberAttrName</span><span class='hs-layout'>)</span> <span class='hs-varop'>&amp;&amp;</span>
+<a name="line-202"></a>                         <span class='hs-varid'>numberAttrName</span> <span class='hs-varop'>`</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>member</span><span class='hs-varop'>`</span> <span class='hs-varid'>mandatoryAttrMap</span> <span class='hs-varid'>d</span>
+<a name="line-203"></a>                       <span class='hs-keyglyph'>]</span>
+<a name="line-204"></a>    <span class='hs-varid'>buildOpParts</span> <span class='hs-keyglyph'>=</span>
+<a name="line-205"></a>        <span class='hs-str'>"opDef"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>quotedText</span> <span class='hs-layout'>(</span><span class='hs-varid'>d</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span> <span class='hs-conop'>:</span>
+<a name="line-206"></a>        <span class='hs-comment'>-- Renders tensor arguments.</span>
+<a name="line-207"></a>        <span class='hs-keyglyph'>[</span> <span class='hs-str'>"&amp; opAttr"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>quotedText</span> <span class='hs-varid'>tfName</span> <span class='hs-varop'>&lt;+&gt;</span>
+<a name="line-208"></a>          <span class='hs-str'>".~ tensorType (undefined ::"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>strictText</span> <span class='hs-varid'>hsName</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-str'>")"</span>
+<a name="line-209"></a>        <span class='hs-keyglyph'>|</span> <span class='hs-layout'>(</span><span class='hs-varid'>tfName</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varid'>hsName</span><span class='hs-layout'>,</span> <span class='hs-keyword'>_</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>toList</span> <span class='hs-varid'>typeMap</span>
+<a name="line-210"></a>        <span class='hs-keyglyph'>]</span> <span class='hs-varop'>++</span>
+<a name="line-211"></a>        <span class='hs-comment'>-- Renders mandatory attributes as function parameters.</span>
+<a name="line-212"></a>        <span class='hs-keyglyph'>[</span> <span class='hs-str'>"&amp; opAttr"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>dquotes</span> <span class='hs-varid'>tfName</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-str'>".~"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>hsName</span>
+<a name="line-213"></a>        <span class='hs-keyglyph'>|</span> <span class='hs-layout'>(</span><span class='hs-varid'>tfName</span><span class='hs-layout'>,</span> <span class='hs-varid'>hsName</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>mandatoryAttrs</span>
+<a name="line-214"></a>        <span class='hs-keyglyph'>]</span> <span class='hs-varop'>++</span>
+<a name="line-215"></a>        <span class='hs-comment'>-- Renders sizes of tensor list types having number_attr.</span>
+<a name="line-216"></a>        <span class='hs-keyglyph'>[</span> <span class='hs-str'>"&amp; opAttr"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>quotedText</span> <span class='hs-varid'>nAttr</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-str'>".~"</span> <span class='hs-varop'>&lt;+&gt;</span>
+<a name="line-217"></a>          <span class='hs-str'>"(fromIntegral (length"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>strictText</span> <span class='hs-layout'>(</span><span class='hs-varid'>head</span> <span class='hs-varid'>tensorNames</span><span class='hs-layout'>)</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-str'>") :: Int64)"</span>
+<a name="line-218"></a>        <span class='hs-keyglyph'>|</span> <span class='hs-layout'>(</span><span class='hs-varid'>nAttr</span><span class='hs-layout'>,</span> <span class='hs-varid'>tensorNames</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>toList</span> <span class='hs-varop'>$</span> <span class='hs-varid'>numberAttrMap</span> <span class='hs-varid'>d</span>
+<a name="line-219"></a>        <span class='hs-keyglyph'>]</span>
+<a name="line-220"></a>    <span class='hs-varid'>mandatoryAttrs</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-varid'>strictText</span> <span class='hs-varid'>tf</span><span class='hs-layout'>,</span> <span class='hs-varid'>strictText</span> <span class='hs-varid'>hs</span><span class='hs-layout'>)</span>
+<a name="line-221"></a>                     <span class='hs-keyglyph'>|</span> <span class='hs-layout'>(</span><span class='hs-varid'>tf</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varid'>hs</span><span class='hs-layout'>,</span> <span class='hs-keyword'>_</span><span class='hs-layout'>,</span> <span class='hs-keyword'>_</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>toList</span> <span class='hs-layout'>(</span><span class='hs-varid'>mandatoryAttrMap</span> <span class='hs-varid'>d</span><span class='hs-layout'>)</span>
+<a name="line-222"></a>                     <span class='hs-keyglyph'>]</span>
+<a name="line-223"></a>    <span class='hs-varid'>haddocks</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"-- |"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>multilineComment</span> <span class='hs-layout'>(</span><span class='hs-varid'>d</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>summary</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>d</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>description</span><span class='hs-layout'>)</span>
+<a name="line-224"></a>    <span class='hs-varid'>extras</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>enclose</span> <span class='hs-str'>"{-\n"</span> <span class='hs-str'>"\n-}"</span> <span class='hs-varop'>$</span>
+<a name="line-225"></a>             <span class='hs-varid'>strictText</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>pack</span> <span class='hs-varop'>$</span>
+<a name="line-226"></a>             <span class='hs-varid'>showMessage</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varid'>def</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpDef</span><span class='hs-layout'>)</span>
+<a name="line-227"></a>                          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>inputArg</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>d</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>inputArg</span><span class='hs-layout'>)</span>
+<a name="line-228"></a>                          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>outputArg</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>d</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>outputArg</span><span class='hs-layout'>)</span>
+<a name="line-229"></a>                          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>attr</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>d</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>attr</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-230"></a>    <span class='hs-varid'>typeMap</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>opDefTypeMap</span> <span class='hs-varid'>d</span>
+<a name="line-231"></a>
+<a name="line-232"></a><a name="quotedText"></a><span class='hs-comment'>-- | Makes a quoted string doc out of the given text value.</span>
+<a name="line-233"></a><span class='hs-definition'>quotedText</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Doc</span>
+<a name="line-234"></a><span class='hs-definition'>quotedText</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>dquotes</span> <span class='hs-varop'>.</span> <span class='hs-varid'>strictText</span>
+<a name="line-235"></a>
+<a name="line-236"></a><a name="typeSig"></a><span class='hs-comment'>-- | typeSig renders the type signature of the given OpDef.</span>
+<a name="line-237"></a><span class='hs-definition'>typeSig</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Doc</span>
+<a name="line-238"></a><span class='hs-definition'>typeSig</span> <span class='hs-varid'>d</span> <span class='hs-keyglyph'>=</span>
+<a name="line-239"></a>    <span class='hs-varid'>foralls</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>constraints</span> <span class='hs-varop'>&lt;+/&gt;</span>
+<a name="line-240"></a>    <span class='hs-varid'>signatureFold</span> <span class='hs-layout'>(</span><span class='hs-varid'>mandatoryAttrInputs</span> <span class='hs-varop'>++</span> <span class='hs-varid'>tensorInputs</span> <span class='hs-varop'>++</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>outputs</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-241"></a>  <span class='hs-keyword'>where</span>
+<a name="line-242"></a>    <span class='hs-varid'>foralls</span> <span class='hs-keyglyph'>|</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>null</span> <span class='hs-varid'>typeMap</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>empty</span>
+<a name="line-243"></a>            <span class='hs-keyglyph'>|</span> <span class='hs-varid'>otherwise</span> <span class='hs-keyglyph'>=</span>
+<a name="line-244"></a>              <span class='hs-str'>"forall"</span>
+<a name="line-245"></a>              <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>sep</span> <span class='hs-layout'>(</span><span class='hs-varid'>refTypes</span> <span class='hs-varop'>++</span> <span class='hs-varid'>map</span> <span class='hs-layout'>(</span><span class='hs-varid'>strictText</span> <span class='hs-varop'>.</span> <span class='hs-varid'>fst</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>elems</span> <span class='hs-varid'>typeMap</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-246"></a>              <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-str'>"."</span>
+<a name="line-247"></a>    <span class='hs-varid'>constraints</span> <span class='hs-keyglyph'>|</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>null</span> <span class='hs-varid'>typeMap</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>empty</span>
+<a name="line-248"></a>                <span class='hs-keyglyph'>|</span> <span class='hs-varid'>otherwise</span> <span class='hs-keyglyph'>=</span>
+<a name="line-249"></a>                  <span class='hs-varid'>tuple</span> <span class='hs-layout'>(</span><span class='hs-varid'>concatMap</span>
+<a name="line-250"></a>                         <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-layout'>(</span><span class='hs-varid'>t</span><span class='hs-layout'>,</span> <span class='hs-varid'>aDef</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-251"></a>                           <span class='hs-str'>"TensorType"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>strictText</span> <span class='hs-varid'>t</span>
+<a name="line-252"></a>                           <span class='hs-conop'>:</span> <span class='hs-varid'>maybeToList</span> <span class='hs-layout'>(</span><span class='hs-varid'>oneOfRestrictions</span> <span class='hs-varid'>aDef</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-253"></a>                         <span class='hs-layout'>(</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>elems</span> <span class='hs-varid'>typeMap</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-str'>"=&gt;"</span>
+<a name="line-254"></a>    <span class='hs-varid'>tensorInputs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>zipWith</span> <span class='hs-varid'>tensorArg</span> <span class='hs-varid'>refTypes</span> <span class='hs-layout'>(</span><span class='hs-varid'>d</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>inputArg</span><span class='hs-layout'>)</span>
+<a name="line-255"></a>    <span class='hs-varid'>refTypes</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>map</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-str'>"v"</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-varid'>int</span> <span class='hs-varid'>x</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>[</span><span class='hs-num'>1</span><span class='hs-keyglyph'>..</span><span class='hs-varid'>length</span> <span class='hs-layout'>(</span><span class='hs-varid'>d</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>inputArg</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span>
+<a name="line-256"></a>    <span class='hs-varid'>tensorArg</span> <span class='hs-varid'>refType</span> <span class='hs-varid'>arg</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>wrapArg</span> <span class='hs-varid'>refType</span> <span class='hs-varid'>arg</span> <span class='hs-varop'>&lt;+&gt;</span>
+<a name="line-257"></a>                            <span class='hs-varid'>hang</span> <span class='hs-num'>0</span> <span class='hs-layout'>(</span><span class='hs-str'>"-- ^"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>argComment</span> <span class='hs-varid'>arg</span><span class='hs-layout'>)</span>
+<a name="line-258"></a>    <span class='hs-comment'>-- Argument type is a list of tensors if number_attr is set;</span>
+<a name="line-259"></a>    <span class='hs-comment'>-- otherwise it's a single Tensor.</span>
+<a name="line-260"></a>    <span class='hs-varid'>wrapArg</span> <span class='hs-varid'>refType</span> <span class='hs-varid'>arg</span> <span class='hs-keyglyph'>=</span>
+<a name="line-261"></a>        <span class='hs-keyword'>if</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>null</span> <span class='hs-layout'>(</span><span class='hs-varid'>arg</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>numberAttr</span><span class='hs-layout'>)</span> <span class='hs-keyword'>then</span> <span class='hs-varid'>typ</span> <span class='hs-keyword'>else</span> <span class='hs-varid'>brackets</span> <span class='hs-varid'>typ</span>
+<a name="line-262"></a>      <span class='hs-keyword'>where</span> <span class='hs-varid'>typ</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tensorType</span> <span class='hs-varid'>refType</span> <span class='hs-varid'>arg</span>
+<a name="line-263"></a>    <span class='hs-varid'>tensorType</span> <span class='hs-varid'>refType</span> <span class='hs-varid'>arg</span> <span class='hs-keyglyph'>=</span>
+<a name="line-264"></a>      <span class='hs-str'>"Tensor"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>refType</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>maybe</span> <span class='hs-varid'>directType</span> <span class='hs-varid'>strictText</span> <span class='hs-varid'>indirectType</span>
+<a name="line-265"></a>      <span class='hs-keyword'>where</span>
+<a name="line-266"></a>        <span class='hs-varid'>indirectType</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>fmap</span> <span class='hs-varid'>fst</span> <span class='hs-layout'>(</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>lookup</span> <span class='hs-layout'>(</span><span class='hs-varid'>arg</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>typeAttr</span><span class='hs-layout'>)</span> <span class='hs-varid'>typeMap</span><span class='hs-layout'>)</span>
+<a name="line-267"></a>        <span class='hs-varid'>directType</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>dtTypeToDoc</span> <span class='hs-layout'>(</span><span class='hs-varid'>arg</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>type'</span><span class='hs-layout'>)</span>
+<a name="line-268"></a>    <span class='hs-varid'>outputs</span> <span class='hs-keyglyph'>=</span>
+<a name="line-269"></a>      <span class='hs-keyword'>case</span> <span class='hs-varid'>d</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>outputArg</span> <span class='hs-keyword'>of</span>
+<a name="line-270"></a>        <span class='hs-conid'>[]</span>  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-str'>"ControlNode"</span>
+<a name="line-271"></a>        <span class='hs-keyglyph'>[</span><span class='hs-varid'>o</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>wrappedOutput</span> <span class='hs-varid'>o</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-str'>"-- ^"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>argComment</span> <span class='hs-varid'>o</span>
+<a name="line-272"></a>        <span class='hs-varid'>os</span>  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>renderTupleResult</span> <span class='hs-varid'>os</span>
+<a name="line-273"></a>    <span class='hs-varid'>wrappedOutput</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>wrapArg</span> <span class='hs-str'>"Value"</span>
+<a name="line-274"></a>    <span class='hs-comment'>-- Tuple result case is rendered differently to give</span>
+<a name="line-275"></a>    <span class='hs-comment'>-- individual elements their own comments.</span>
+<a name="line-276"></a>    <span class='hs-varid'>renderTupleResult</span> <span class='hs-varid'>os</span> <span class='hs-keyglyph'>=</span>
+<a name="line-277"></a>        <span class='hs-varid'>stack</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>[</span> <span class='hs-varid'>tuple</span> <span class='hs-layout'>(</span><span class='hs-varid'>map</span> <span class='hs-varid'>wrappedOutput</span> <span class='hs-varid'>os</span><span class='hs-layout'>)</span>
+<a name="line-278"></a>                <span class='hs-layout'>,</span> <span class='hs-varid'>flatten</span> <span class='hs-varid'>commentSummary</span>
+<a name="line-279"></a>                <span class='hs-keyglyph'>]</span> <span class='hs-varop'>++</span> <span class='hs-varid'>map</span> <span class='hs-varid'>commentDetails</span> <span class='hs-varid'>os</span>
+<a name="line-280"></a>      <span class='hs-keyword'>where</span>
+<a name="line-281"></a>        <span class='hs-varid'>commentSummary</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"-- ^"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>tuple</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>bold</span> <span class='hs-layout'>(</span><span class='hs-varid'>o</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>o</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>os</span><span class='hs-keyglyph'>]</span>
+<a name="line-282"></a>        <span class='hs-varid'>commentDetails</span> <span class='hs-varid'>o</span> <span class='hs-keyglyph'>=</span>
+<a name="line-283"></a>          <span class='hs-varid'>stack</span> <span class='hs-keyglyph'>[</span> <span class='hs-str'>"--"</span>
+<a name="line-284"></a>                <span class='hs-layout'>,</span> <span class='hs-str'>"-- *"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>argComment</span> <span class='hs-varid'>o</span>
+<a name="line-285"></a>                <span class='hs-keyglyph'>]</span>
+<a name="line-286"></a>    <span class='hs-varid'>signatureFold</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>folddoc</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span> <span class='hs-varop'>&lt;/&gt;</span> <span class='hs-str'>"-&gt;"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>y</span><span class='hs-layout'>)</span>
+<a name="line-287"></a>    <span class='hs-varid'>mandatoryAttrInputs</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span>
+<a name="line-288"></a>      <span class='hs-varid'>dtTypeToDoc</span> <span class='hs-varid'>dtType</span> <span class='hs-varop'>&lt;+&gt;</span>
+<a name="line-289"></a>          <span class='hs-varid'>hang</span> <span class='hs-num'>0</span> <span class='hs-layout'>(</span><span class='hs-str'>"-- ^"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>argComment'</span> <span class='hs-varid'>tfName</span> <span class='hs-varid'>descr</span><span class='hs-layout'>)</span>
+<a name="line-290"></a>      <span class='hs-keyglyph'>|</span> <span class='hs-layout'>(</span><span class='hs-varid'>tfName</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-keyword'>_</span><span class='hs-layout'>,</span> <span class='hs-varid'>dtType</span><span class='hs-layout'>,</span> <span class='hs-varid'>descr</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>toList</span> <span class='hs-varop'>$</span> <span class='hs-varid'>mandatoryAttrMap</span> <span class='hs-varid'>d</span><span class='hs-keyglyph'>]</span>
+<a name="line-291"></a>    <span class='hs-varid'>typeMap</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>opDefTypeMap</span> <span class='hs-varid'>d</span>
+<a name="line-292"></a>
+<a name="line-293"></a><a name="oneOfRestrictions"></a><span class='hs-comment'>-- | Returns the type restriction for the given tensor type if the</span>
+<a name="line-294"></a><span class='hs-comment'>-- set of allowed types is not empty (i.e. restricted).</span>
+<a name="line-295"></a><span class='hs-definition'>oneOfRestrictions</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>AttrDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Maybe</span> <span class='hs-conid'>Doc</span>
+<a name="line-296"></a><span class='hs-definition'>oneOfRestrictions</span> <span class='hs-varid'>aDef</span> <span class='hs-varid'>tName</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-297"></a>    <span class='hs-varid'>typs</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>onAttrType</span> <span class='hs-layout'>(</span><span class='hs-varop'>^.</span> <span class='hs-varid'>templateRestrictions</span><span class='hs-layout'>)</span> <span class='hs-varid'>aDef</span>
+<a name="line-298"></a>    <span class='hs-varid'>guard</span> <span class='hs-varop'>$</span> <span class='hs-varid'>not</span> <span class='hs-varop'>$</span> <span class='hs-varid'>null</span> <span class='hs-varid'>typs</span>
+<a name="line-299"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>typeList</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>commasep</span> <span class='hs-varop'>$</span> <span class='hs-varid'>map</span> <span class='hs-varid'>strictText</span> <span class='hs-varop'>$</span>
+<a name="line-300"></a>                   <span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>toList</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span> <span class='hs-varop'>$</span>
+<a name="line-301"></a>                   <span class='hs-varid'>map</span> <span class='hs-varid'>dtTypeToHaskell</span> <span class='hs-varid'>typs</span>
+<a name="line-302"></a>    <span class='hs-varid'>return</span> <span class='hs-varop'>$</span> <span class='hs-str'>"OneOf"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-str'>"'"</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-varid'>brackets</span> <span class='hs-varid'>typeList</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>strictText</span> <span class='hs-varid'>tName</span>
+<a name="line-303"></a>
+<a name="line-304"></a><a name="numberAttrMap"></a><span class='hs-comment'>-- | Identifies the attributes used as tensor cardinalities. In such</span>
+<a name="line-305"></a><span class='hs-comment'>-- cases a list of tensors is supplied as an input_arg. The number of</span>
+<a name="line-306"></a><span class='hs-comment'>-- such inputs is communicated as a separate opAttr.</span>
+<a name="line-307"></a><span class='hs-comment'>-- The result key is TensorFlow attribute name and the value is the</span>
+<a name="line-308"></a><span class='hs-comment'>-- tensor names which have number_attr set to the result key.</span>
+<a name="line-309"></a><span class='hs-definition'>numberAttrMap</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-layout'>(</span><span class='hs-conid'>NonEmpty</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-310"></a><span class='hs-definition'>numberAttrMap</span> <span class='hs-varid'>d</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromListWith</span> <span class='hs-layout'>(</span><span class='hs-conid'>Semigroup</span><span class='hs-varop'>.&lt;&gt;</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>[</span>
+<a name="line-311"></a>    <span class='hs-layout'>(</span><span class='hs-varid'>nAttr</span><span class='hs-layout'>,</span> <span class='hs-varid'>replaceReservedName</span> <span class='hs-layout'>(</span><span class='hs-varid'>inp</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span> <span class='hs-conop'>:|</span> <span class='hs-conid'>[]</span><span class='hs-layout'>)</span>
+<a name="line-312"></a>    <span class='hs-keyglyph'>|</span> <span class='hs-varid'>inp</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>d</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>inputArg</span>
+<a name="line-313"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>nAttr</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>inp</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>numberAttr</span><span class='hs-keyglyph'>]</span>
+<a name="line-314"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>not</span> <span class='hs-layout'>(</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>null</span> <span class='hs-varid'>nAttr</span><span class='hs-layout'>)</span>
+<a name="line-315"></a>    <span class='hs-keyglyph'>]</span>
+<a name="line-316"></a>
+<a name="line-317"></a><a name="argComment"></a><span class='hs-definition'>argComment</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Doc</span>
+<a name="line-318"></a><span class='hs-definition'>argComment</span> <span class='hs-varid'>arg</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>argComment'</span> <span class='hs-layout'>(</span><span class='hs-varid'>arg</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>arg</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>description</span><span class='hs-layout'>)</span>
+<a name="line-319"></a>
+<a name="line-320"></a><a name="argComment'"></a><span class='hs-definition'>argComment'</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Doc</span>
+<a name="line-321"></a><span class='hs-definition'>argComment'</span> <span class='hs-varid'>argName</span> <span class='hs-varid'>argDesc</span> <span class='hs-keyglyph'>=</span>
+<a name="line-322"></a>    <span class='hs-varid'>bold</span> <span class='hs-varid'>argName</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-varid'>splitMultilineText</span> <span class='hs-layout'>(</span><span class='hs-str'>":"</span> <span class='hs-varop'>&lt;+&gt;</span><span class='hs-layout'>)</span> <span class='hs-varid'>argDesc</span>
+<a name="line-323"></a>
+<a name="line-324"></a><a name="bold"></a><span class='hs-definition'>bold</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Doc</span>
+<a name="line-325"></a><span class='hs-definition'>bold</span> <span class='hs-varid'>n</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>strictText</span> <span class='hs-layout'>(</span><span class='hs-str'>"__"</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-varid'>n</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-str'>"__"</span><span class='hs-layout'>)</span>
+<a name="line-326"></a>
+<a name="line-327"></a><a name="opDefTypeMap"></a><span class='hs-definition'>opDefTypeMap</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-layout'>(</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span> <span class='hs-conid'>AttrDef</span><span class='hs-layout'>)</span>
+<a name="line-328"></a><span class='hs-definition'>opDefTypeMap</span> <span class='hs-varid'>d</span> <span class='hs-keyglyph'>=</span>
+<a name="line-329"></a>    <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-varid'>n</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varid'>lowCase</span> <span class='hs-varid'>n</span><span class='hs-layout'>,</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>|</span> <span class='hs-layout'>(</span><span class='hs-varid'>n</span><span class='hs-layout'>,</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>attrList</span> <span class='hs-varid'>d</span><span class='hs-layout'>,</span> <span class='hs-varid'>isType</span> <span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span>
+<a name="line-330"></a>
+<a name="line-331"></a><a name="attrList"></a><span class='hs-definition'>attrList</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span> <span class='hs-conid'>AttrDef</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span>
+<a name="line-332"></a><span class='hs-definition'>attrList</span> <span class='hs-varid'>d</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-varid'>a</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>name</span><span class='hs-layout'>,</span> <span class='hs-varid'>attrDef</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>d</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>attr</span><span class='hs-keyglyph'>]</span>
+<a name="line-333"></a>
+<a name="line-334"></a><a name="isType"></a><span class='hs-definition'>isType</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>AttrDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Bool</span>
+<a name="line-335"></a><span class='hs-definition'>isType</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>fromMaybe</span> <span class='hs-conid'>False</span> <span class='hs-varop'>.</span> <span class='hs-varid'>onAttrType</span> <span class='hs-layout'>(</span><span class='hs-varid'>const</span> <span class='hs-conid'>True</span><span class='hs-layout'>)</span>
+<a name="line-336"></a>
+<a name="line-337"></a><a name="onAttrType"></a><span class='hs-comment'>-- | Applies the given function to the data type. Is this a Prism?</span>
+<a name="line-338"></a><span class='hs-definition'>onAttrType</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>Template</span> <span class='hs-conid'>DataType</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>AttrDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Maybe</span> <span class='hs-varid'>a</span>
+<a name="line-339"></a><span class='hs-definition'>onAttrType</span> <span class='hs-varid'>f</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>case</span> <span class='hs-varid'>x</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>attrTemplate</span> <span class='hs-keyword'>of</span>
+<a name="line-340"></a>    <span class='hs-conid'>AttrSingle</span> <span class='hs-layout'>(</span><span class='hs-conid'>AttrType</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Just</span> <span class='hs-layout'>(</span><span class='hs-varid'>f</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-341"></a>    <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Nothing</span>
+<a name="line-342"></a>
+<a name="line-343"></a><a name="mandatoryAttrMap"></a><span class='hs-comment'>-- | mandatoryAttrMap contains the attributes chosen by</span>
+<a name="line-344"></a><span class='hs-comment'>-- isMandatoryAttr, excluding those which are derived from list of</span>
+<a name="line-345"></a><span class='hs-comment'>-- tensor arguments. The key is the TF name of the attribute. The</span>
+<a name="line-346"></a><span class='hs-comment'>-- value tuple is (haskell name, TF type, attribute comment).</span>
+<a name="line-347"></a><span class='hs-definition'>mandatoryAttrMap</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-layout'>(</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span> <span class='hs-conid'>DataType</span><span class='hs-layout'>,</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-348"></a><span class='hs-definition'>mandatoryAttrMap</span> <span class='hs-varid'>d</span> <span class='hs-keyglyph'>=</span>
+<a name="line-349"></a>    <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span> <span class='hs-keyglyph'>[</span> <span class='hs-layout'>(</span><span class='hs-varid'>n</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varid'>lowCase</span> <span class='hs-varid'>n</span><span class='hs-layout'>,</span> <span class='hs-varid'>dtType</span><span class='hs-layout'>,</span> <span class='hs-varid'>a</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>attrOriginal</span><span class='hs-varop'>.</span><span class='hs-varid'>description</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-350"></a>                 <span class='hs-keyglyph'>|</span> <span class='hs-layout'>(</span><span class='hs-varid'>n</span><span class='hs-layout'>,</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>attrList</span> <span class='hs-varid'>d</span>
+<a name="line-351"></a>                 <span class='hs-layout'>,</span> <span class='hs-conid'>Just</span> <span class='hs-varid'>dtType</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>isMandatoryAttr</span> <span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span>
+<a name="line-352"></a>                 <span class='hs-comment'>-- Excludes the attributes rendered as list lengths.</span>
+<a name="line-353"></a>                 <span class='hs-layout'>,</span> <span class='hs-varid'>n</span> <span class='hs-varop'>`</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>notMember</span><span class='hs-varop'>`</span> <span class='hs-varid'>numberAttrMap</span> <span class='hs-varid'>d</span>
+<a name="line-354"></a>                 <span class='hs-keyglyph'>]</span>
+<a name="line-355"></a>
+<a name="line-356"></a><a name="isMandatoryAttr"></a><span class='hs-comment'>-- | Inspects the attribute and if it is one of the implemented</span>
+<a name="line-357"></a><span class='hs-comment'>-- non-tensor values lacking default, then returns Just the TF type.</span>
+<a name="line-358"></a><span class='hs-definition'>isMandatoryAttr</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>AttrDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Maybe</span> <span class='hs-conid'>DataType</span>
+<a name="line-359"></a><span class='hs-definition'>isMandatoryAttr</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>=</span>
+<a name="line-360"></a>   <span class='hs-keyword'>case</span> <span class='hs-varid'>x</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>attrTemplate</span> <span class='hs-keyword'>of</span>
+<a name="line-361"></a>     <span class='hs-conid'>AttrSingle</span> <span class='hs-layout'>(</span><span class='hs-conid'>AttrBool</span> <span class='hs-varid'>y</span><span class='hs-layout'>)</span>  <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>noDefault</span> <span class='hs-conid'>DT_BOOL</span> <span class='hs-varid'>y</span>
+<a name="line-362"></a>     <span class='hs-conid'>AttrSingle</span> <span class='hs-layout'>(</span><span class='hs-conid'>AttrInt64</span> <span class='hs-varid'>y</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>noDefault</span> <span class='hs-conid'>DT_INT64</span> <span class='hs-varid'>y</span>
+<a name="line-363"></a>     <span class='hs-conid'>AttrSingle</span> <span class='hs-layout'>(</span><span class='hs-conid'>AttrFloat</span> <span class='hs-varid'>y</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>noDefault</span> <span class='hs-conid'>DT_FLOAT</span> <span class='hs-varid'>y</span>
+<a name="line-364"></a>     <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Nothing</span>
+<a name="line-365"></a>   <span class='hs-keyword'>where</span>
+<a name="line-366"></a>     <span class='hs-varid'>noDefault</span> <span class='hs-varid'>typ</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>maybe</span> <span class='hs-layout'>(</span><span class='hs-conid'>Just</span> <span class='hs-varid'>typ</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>const</span> <span class='hs-conid'>Nothing</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>y</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>templateDefault</span><span class='hs-layout'>)</span>
+<a name="line-367"></a>
+<a name="line-368"></a><a name="dtTypeToDoc"></a><span class='hs-definition'>dtTypeToDoc</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>DataType</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Doc</span>
+<a name="line-369"></a><span class='hs-definition'>dtTypeToDoc</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>strictText</span> <span class='hs-varop'>.</span> <span class='hs-varid'>dtTypeToHaskell</span>
+<a name="line-370"></a>
+<a name="line-371"></a><a name="dtTypeToHaskell"></a><span class='hs-comment'>-- NOTE: The cases of this function should be kept in sync with</span>
+<a name="line-372"></a><span class='hs-comment'>-- TensorFlow.Types.AllTensorTypes.</span>
+<a name="line-373"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>DataType</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-374"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_BOOL</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Bool"</span>
+<a name="line-375"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_BFLOAT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Data.Word.Word16"</span>
+<a name="line-376"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_COMPLEX128</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"(Data.Complex.Complex Double)"</span>
+<a name="line-377"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_COMPLEX64</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"(Data.Complex.Complex Float)"</span>
+<a name="line-378"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_DOUBLE</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Double"</span>
+<a name="line-379"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_FLOAT</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Float"</span>
+<a name="line-380"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_INT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Data.Int.Int16"</span>
+<a name="line-381"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_INT32</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Data.Int.Int32"</span>
+<a name="line-382"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_INT64</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Data.Int.Int64"</span>
+<a name="line-383"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_INT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Data.Int.Int8"</span>
+<a name="line-384"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_QINT32</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Data.Int.Int32"</span>  <span class='hs-comment'>-- TODO(gnezdo): make unique</span>
+<a name="line-385"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_QINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Data.Word.Word8"</span>  <span class='hs-comment'>-- TODO(gnezdo): make unique</span>
+<a name="line-386"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_QINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Data.Int.Int16"</span>  <span class='hs-comment'>-- TODO(gnezdo): make unique</span>
+<a name="line-387"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_QUINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Data.Word.Word16"</span>  <span class='hs-comment'>-- TODO(gnezdo): make unique</span>
+<a name="line-388"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_QUINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Data.Word.Word8"</span>  <span class='hs-comment'>-- TODO(gnezdo): make unique</span>
+<a name="line-389"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_STRING</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Data.ByteString.ByteString"</span>
+<a name="line-390"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_UINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Data.Word.Word16"</span>
+<a name="line-391"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_HALF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Data.Word.Word16"</span>  <span class='hs-comment'>-- TODO(gnezdo): make unique</span>
+<a name="line-392"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-conid'>DT_UINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"Data.Word.Word8"</span>
+<a name="line-393"></a><span class='hs-definition'>dtTypeToHaskell</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>=</span>
+<a name="line-394"></a>    <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>pack</span> <span class='hs-varop'>$</span> <span class='hs-str'>"Unsupported type in dtTypeToHaskell: "</span> <span class='hs-varop'>++</span> <span class='hs-varid'>show</span> <span class='hs-varid'>x</span>
+<a name="line-395"></a>
+<a name="line-396"></a><a name="haddockComment"></a><span class='hs-comment'>-- | haddockComment escapes TensorFlow doc strings into haddock.</span>
+<a name="line-397"></a><span class='hs-comment'>-- TODO(gnezdo): deal with the markup.</span>
+<a name="line-398"></a><span class='hs-definition'>haddockComment</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Doc</span>
+<a name="line-399"></a><span class='hs-definition'>haddockComment</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>strictText</span>
+<a name="line-400"></a>
+<a name="line-401"></a><a name="multilineComment"></a><span class='hs-definition'>multilineComment</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Doc</span>
+<a name="line-402"></a><span class='hs-definition'>multilineComment</span> <span class='hs-varid'>summary'</span> <span class='hs-varid'>detail</span> <span class='hs-keyglyph'>=</span>
+<a name="line-403"></a>    <span class='hs-varid'>haddockComment</span> <span class='hs-varid'>summary'</span> <span class='hs-varop'>&lt;/&gt;</span>
+<a name="line-404"></a>    <span class='hs-varid'>splitMultilineText</span> <span class='hs-varid'>insertParagraphAndComment</span> <span class='hs-varid'>detail</span>
+<a name="line-405"></a>  <span class='hs-keyword'>where</span> <span class='hs-varid'>insertParagraphAndComment</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"--"</span> <span class='hs-varop'>&lt;/&gt;</span> <span class='hs-str'>"--"</span> <span class='hs-varop'>&lt;+&gt;</span> <span class='hs-varid'>x</span>
+<a name="line-406"></a>
+<a name="line-407"></a><a name="splitMultilineText"></a><span class='hs-comment'>-- | Converts the given multi-line detail string into</span>
+<a name="line-408"></a><span class='hs-comment'>-- a multi-line haddock. Applies the given lead to the</span>
+<a name="line-409"></a><span class='hs-comment'>-- first line. Returns an empty document for empty detail.</span>
+<a name="line-410"></a><span class='hs-definition'>splitMultilineText</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>Doc</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Doc</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Doc</span>
+<a name="line-411"></a><span class='hs-definition'>splitMultilineText</span> <span class='hs-varid'>lead</span> <span class='hs-varid'>detail</span> <span class='hs-keyglyph'>=</span>
+<a name="line-412"></a>  <span class='hs-keyword'>case</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>lines</span> <span class='hs-varid'>detail</span> <span class='hs-keyword'>of</span>
+<a name="line-413"></a>    <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>empty</span>
+<a name="line-414"></a>    <span class='hs-layout'>(</span><span class='hs-varid'>l</span> <span class='hs-conop'>:</span> <span class='hs-varid'>ls</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>stack</span> <span class='hs-varop'>$</span> <span class='hs-varid'>lead</span> <span class='hs-layout'>(</span><span class='hs-varid'>haddockComment</span> <span class='hs-varid'>l</span><span class='hs-layout'>)</span>
+<a name="line-415"></a>                      <span class='hs-conop'>:</span> <span class='hs-varid'>map</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-str'>"--"</span> <span class='hs-varop'>&lt;+&gt;</span><span class='hs-layout'>)</span> <span class='hs-varop'>.</span> <span class='hs-varid'>haddockComment</span><span class='hs-layout'>)</span> <span class='hs-varid'>ls</span>
+<a name="line-416"></a>
+<a name="line-417"></a><a name="replaceReservedName"></a><span class='hs-definition'>replaceReservedName</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Text</span>
+<a name="line-418"></a><span class='hs-definition'>replaceReservedName</span> <span class='hs-varid'>n</span>
+<a name="line-419"></a>    <span class='hs-keyglyph'>|</span> <span class='hs-varid'>n</span> <span class='hs-varop'>`</span><span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>member</span><span class='hs-varop'>`</span> <span class='hs-varid'>reservedKeywords</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>n</span> <span class='hs-varop'>&lt;&gt;</span> <span class='hs-str'>"'"</span>
+<a name="line-420"></a>    <span class='hs-keyglyph'>|</span> <span class='hs-varid'>otherwise</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>n</span>
+<a name="line-421"></a>
+<a name="line-422"></a><a name="indentation"></a><span class='hs-definition'>indentation</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>4</span>
+<a name="line-423"></a>
+<a name="line-424"></a><a name="reservedKeywords"></a><span class='hs-definition'>reservedKeywords</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-conid'>Set</span> <span class='hs-conid'>Text</span>
+<a name="line-425"></a><span class='hs-definition'>reservedKeywords</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span> <span class='hs-varop'>$</span>
+<a name="line-426"></a>    <span class='hs-comment'>-- Haskell2010 keywords:</span>
+<a name="line-427"></a>    <span class='hs-comment'>-- https://www.haskell.org/onlinereport/haskell2010/haskellch2.html#x7-180002.4</span>
+<a name="line-428"></a>    <span class='hs-comment'>-- We don't include keywords that are allowed to be variable names,</span>
+<a name="line-429"></a>    <span class='hs-comment'>-- in particular: "as", "forall", and "hiding".</span>
+<a name="line-430"></a>    <span class='hs-keyglyph'>[</span> <span class='hs-str'>"case"</span>
+<a name="line-431"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"class"</span>
+<a name="line-432"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"data"</span>
+<a name="line-433"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"default"</span>
+<a name="line-434"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"deriving"</span>
+<a name="line-435"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"do"</span>
+<a name="line-436"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"else"</span>
+<a name="line-437"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"foreign"</span>
+<a name="line-438"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"if"</span>
+<a name="line-439"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"import"</span>
+<a name="line-440"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"in"</span>
+<a name="line-441"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"infix"</span>
+<a name="line-442"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"infixl"</span>
+<a name="line-443"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"infixr"</span>
+<a name="line-444"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"instance"</span>
+<a name="line-445"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"let"</span>
+<a name="line-446"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"module"</span>
+<a name="line-447"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"newtype"</span>
+<a name="line-448"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"of"</span>
+<a name="line-449"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"then"</span>
+<a name="line-450"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"type"</span>
+<a name="line-451"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"where"</span>
+<a name="line-452"></a>    <span class='hs-keyglyph'>]</span>
+<a name="line-453"></a>    <span class='hs-varop'>++</span>  <span class='hs-comment'>-- Nonstandard extensions</span>
+<a name="line-454"></a>    <span class='hs-keyglyph'>[</span> <span class='hs-str'>"mdo"</span>   <span class='hs-comment'>-- RecursiveDo</span>
+<a name="line-455"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"rec"</span>   <span class='hs-comment'>-- Arrows, RecursiveDo</span>
+<a name="line-456"></a>    <span class='hs-layout'>,</span> <span class='hs-str'>"proc"</span>  <span class='hs-comment'>-- Arrows</span>
+<a name="line-457"></a>    <span class='hs-keyglyph'>]</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/src/hscolour.css b/docs/haddock/tensorflow-opgen-0.1.0.0/src/hscolour.css
new file mode 100644
index 0000000..c15919e
--- /dev/null
+++ b/docs/haddock/tensorflow-opgen-0.1.0.0/src/hscolour.css
@@ -0,0 +1,5 @@
+.hs-keyglyph, .hs-layout {color: red;}
+.hs-keyword {color: blue;}
+.hs-comment, .hs-comment a {color: green;}
+.hs-str, .hs-chr {color: teal;}
+.hs-keyword, .hs-conid, .hs-varid, .hs-conop, .hs-varop, .hs-num, .hs-cpp, .hs-sel, .hs-definition {}
diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/synopsis.png b/docs/haddock/tensorflow-opgen-0.1.0.0/synopsis.png
new file mode 100644
index 0000000000000000000000000000000000000000..85fb86ec84907bcc86531dc82871948ff4d471fa
GIT binary patch
literal 11327
zcmV-FEWp!=P)<h;3K|Lk000e1NJLTq004jh003wR1^@s6IVt!g000+IX+uL$Nkc;*
zP;zf(X>4Tx0C)k_S!GyNTeqHT_l8Y(cXyX`gGi?cY`Qxn1VID|MJXwjPC)?)F$h6K
zMMOd+6hs7sqbPzXbr*U(-*=zy-hcPcUC*=TdiNM(jyd-lv&OpsU|J&v2m2!^0SE{T
z54F(O;<U4O;KWV=44?vZKnREv-NJ%3&CHI1|Ml9M1kj;!OzhCt|7!L>E2!K(!rTCW
z%wV;vdzf1QjBf#e&~gh74F>?Z4a=WLg$KhJ^$5nap>PLbJadS>e&h8+?D`9%QNL`g
zEVKbYGXj7k5Q(8)0Fd#*a?VIMFW3*64geVHKzE-&0BG!BtmfuTbO(T`0Jaeg2nagF
z{V*1E{Wm{e|AvV~*MEExiC+KU-~R=!2{)|c6Bg`GjQ;iG|FQ`1kAUCTuZtQk34#8{
z4r4<P-^=bq9b@>(3g7#|{=Z@d+d#}7f!3C=>=26vx*jwA8>@MS>RG@Tt_zt3hie^T
z_?<zX!Qp0Bf6L=zytR+Pm;lRjJ;HSV_$l=z8Xd-r9G16)Mp#(Gm>0%9VUd=)Fos7I
z^ghPh%Jy%YZ|)vCf6EaFPai$Q-!=$ppK!y&wrJs)bNdAuANB!m3n34Tfj{s75g-&U
z1A!Pg3bcXF-=!Gv1VmU93G2duANT;{0JugFTqg*|oPXPC|A$2HS3NJd-hcPV3EW`Y
zh=1Dr-5Mv{<{zIvz#Yba<V3iQL)~HT!km4Q{?+Cmi$j0EdL!8&xgxpyr%xz;lr%~X
zrG-*OsiBkrCyE}$48%~fC`FVeN*$#F>y&^Vcn^E_`qRfl{{bzYkp)4~$~NAx_VB;E
z{?P)PU)DbV{Qi#~0H0@T9czDj06@6MNq8OrpdAz(9qQxd9nPr<&s+~tPQySqaZyfb
zNh!%g_5YjeaLxMN*$s<dYY{|0@-La=c>v_p;d%b#U$Wpz0Geb0U>E+EOsEQ;I<Re%
zqFY30<Zpkd!v`9W019}7G68nL0|epsEDhv=3eW($zzCQED_{?t;PK-PkHJt71>!&=
zNC6q(BFF<p;2Nj^wV)BSfDUjM^n!=r33v`Bz%-Zxi(nP3gDtQFK@b)qgJ>Wohy&t-
zL?CHM5mJM6p`(xmWDmJOUQi$u0mVUQpbRJ*DuT+OI;a`C4fR4p&?xj8nuk`Puh35f
z5<x)FBG?f82nmEDLK9((utJ<bcp*X%F^FWu1w;X&9C01dj_5@^K};a#5UYqyBtViP
z>5*JWF{C0=8)=GkKzbrWk@3iMWInPS*@Wyu4kE{pbI3L14-^JPgW^Pq!Q<2bWsPz}
zg`nb5nW!REEvg;Wj~YYGqt;RTXfiY_S_G|(HbmQ@z0gtU6m&ki8r_B-Ku@3-(OVb{
zh8`n;QNS2r>@mKWSWG773g!l;2Q!LUz-(f%SSG9pRuyZCC1S&|DcC~nb!<2G1$Gg;
zjU&Zz;G}VSI0sxHE(w>9tH<5Py}&KucJP#VKD;vC6z`6Y#%JLx@m=^4{33pbgo;Ff
zM3uyf#Fr$Iq=2M}WPoIbWP_BHl$%tE)ST3Z^fYM!=}po{r1PXd2-E~&f;PdC5J9*=
zs3G(aUK2LR$jJD~G{_vt!pSa>)sa0QdqcKOPD3tEZb<G<o<Lqq-cJ64{1XM9f{#Lz
z!ige=BA=p_VuWIi5>LrbsZB|wjHfK7yiNI%a+8XNN{Y&qDu61Js-9|yYMB~K%}=dM
z?M|IcT|xbTdVvN>!$YG@<3@9arjllWW|0;{D?n>V>r0zK+erJ2cAbuzPL|Gw?j&6?
z-95TFdL%tRy&=6neHMKS{UrTQ1~vvw1`mcbh9-s=4Br`97&RC@7}FVVFitT3Wa4Df
zW%6UX#MHqw%Zy?cW;SPzV!p~ez`Vvn%c8>K#*)s`!ZO8*U=?PyV2x$1V13HE$;Qs6
z&lb#9$o7D3jh&udgWZ=sm;FBb3I`2`8ix-@E=M=VM@~9UO-_H#0?vNUbuLye1Fi_J
zGOlM_JKO@?*4#<lH@IhcNO%-^Jb7|?`gqoPIe1NY<9Hi*r}*%EihSODm-z<yzVi$4
z+w!OKcknL@FbWt7#0WG9%m|VRstbk)mJ5yxVT2Tg{DiIuJr_m@%L#i67YmPwAVlOv
zd_=B@jEbT~l|%zY%SB&`5yUjbBE=fS=Edp7kBTRWcZh$I;FWNYxFj(su`ekn=`UF+
zIW0vkWh9j()g|>+T3Fgmx>$N#hD=6JCPAiC=8LR|tcUDX*;jHjawc-Aa(!}p@(S{y
z@=fw93cLy~3MC3J6=@aC6f+ecDWR3LloFKgD*aHFR}NQhQU0tVrsAhkud;kZ;E2bO
z$|DP^+^R&?GSxXXPBj;`QnfjCE_I@Mx%xW|9u0SmYK<jLAx&S+>zbdmB(*}d+O)oF
zD{G(9?$JT&=<B5GJkzDtwbm`teWS;%=cU)6x2`XzAFJPEfHp8TxMc9skj>D|u+DJZ
zNWtioQNJ<4*wVPj_}x+AqoGH;Ob{kUCOIZE$M}u~9_ug#riP|Drn6=OW+7&G%rWL>
z=Ede8ETk;rECwxUES)XuEw`++tg@`8tp%+ktov*zY#eRsY`)v-*k;?#*-6-)vU_6B
zZ0}>=>40^xaj16KJg$2@@A#sloMVdP<OzWjCr=DHF+2G?-E*dJCOWsepj~WT>Ron;
zro?jMrmLZAiR-$Xw%cX5Rd)^dT=x|ZRgY|sB~Mk)Y|mvcRj(Yc6>oL#eD5_MZJ#2a
zFTMu8*L=VGnflfE9r)Y&-w413xCGn|qz?28><?lOIvF$?EE1d&JQt!Ik{7ZbY8+Y}
z28Epn>kOxb4~I`91S8Hy%txw47DsMJ*+jLTq&gXR@@ceibXxRMj9yGtEGpJ5wl9t=
zE-`NYl;)|jcqraAzAu3%Avt03wEpSZM3O|m#Ni~#r0k?`XKc@OC9@@;PF^^xf3_io
zJS8;cWvWW*wR5O*KIfjL$)pvg?Wen^KhBWM$j{i#bjy5vUg~_o`GX6d7oKIwXI;IB
zxfpnH@{;<c+H9)qnCyidlbrTk&fN6eZ+XsnL;14#*DjM>j<`HmaI~Pakhkz+;ck(4
z(L}LU@r@GJlC+ZVSKP0>xT<lr=^ER$jB7tjeM(=I>6f*a^OxsWU@9UjK2+LN4pu2v
z)m1ZBXH@Ui1lG*eTGaN}Db&@~v({%dAQ~bXR<1i<f8MCm*x4l3RCa^@M#fEWGvemQ
zW|!v47UPz_R>jt)TYR@l+GyI++oAU8_Vo_$j=4_z&e7XOxBI$Oy4voD->JFFb+`B)
z-My^)B=?i=A9TlbZ}tTDto3^JF7!F~O+T=EFy3$8|7^f`;L$_9hYtod2fH7sKDs-k
zJaqf9;^U4d@=w~I$~|oxmK$z+CjYE`L}8@!xzh8l(IcbxU#P$69n%?mIBq!pWa8Mw
z=%n@JtCx;1=U%zLT7K>S`pZ=0)Xwzj8T3s0Eahze8`d}FZ-w68n3JEoH?K4Q^qu9q
z=>@li)%RiVcNddCkbTHs;#jI%mR`QQqPOz=<MEG+tHG-~Ybk$G{Z;Zw@YAi&8lRu8
z+pRBt3H@^LHFJY$qvo6Jx4uoY%{N;CTRY#=x0$wUe<=PK{Av4hX(wtIyIZg)wAZz7
zygzdgaBvVz3?=?L3n1YV;pqv$*BUs7u!nPp=KzrO{LU$US-|NE1lz#a2I7HkB7Bib
zC_HKj-GeE>CgGy+9whdp4g`BLCvp!8U&;uov(!a2t+bEnRv6HXyi9t`-YglcEo`$K
zI8GTZXYLH1F5YE+b^&9-c%dfYc~N>X1MygiCdpZ8N*OKLV7W5+5rusvVP$KTgd_E;
zV`@J%*flk^Jhjj1)aX9cTQC5ItVZ(2W=FkE;*aH-)|+*kk6SET?pjmWaNEk+>D${o
z_#cmV%sNr-bj$gX%QW$m8{|&wA?SI;%go!uC))SCU%7vKz~jI-L0?1Ap^RZ7;i?hG
zB3+__P9{WW#uUa@#oavB8Q+`m==5;nXwvwZiR6j1<0+%5!{;8Q^`_s>XwIxTUvlAM
z)|rdpmprp=bM$iM@_6#8@((Vr7Q8HcP;{fXs3iGH;8nY8TBRaov}JqcixtC_ZB<vQ
z&(`?YTGlDob2eZb{<=Qi*ws{eBju)Fvqg(SE88vPt>w07?YBCLI#1vB=rX<|d6)j~
z?!9;SA9XkN<n~7QIX}?p7apJ-*nBuW*z>4rDD83J6N{$`!z{xG&lW}=KCd6md=WHe
zF)la3F!5t@`sLkMS6?Sg5vR3gcxTbGOK%>(y*_twKH{Cjg64anMViI^4{J-a%g<LH
ze(YYoyLRud-cLiHC*aZaV}tse#HQ(1*!QCCfgc+?yt~$WY5Tng2XOpFKtGg+ut(A%
z-=j*=VHi8C5{?;<#;=l0k&Y4u$@<72Q1np_Pz_U$(Jatz(CstOF!D30Fk7<(v!=0C
zv-fel=R$LH^XTz<^JVZi3XBSF3Nwow5p@<z60edNko+Kxkr9;Dk#m)gS143$Q5sTy
zqw-A^tHz`*q@kp#r)8miT*qBEKrc)`${^M-!6?x<@o1vS>0=3|@n*5+(H4=G;Z`Bm
z0XDw2UUnY#t`5ZG&W<NeoN#h<c62#GbaHieBf7hKczXJJ`FjWZMEFMgo$^l#ND0gg
zx)_`nQW$zAtSr1TqB^oJs{Uj{bVE#iY<*nqsjB$0ge#{D6LXTzpE;Wxe>ObDFO_)C
zCe0{aEki1k_dNXt+=U-mA1_W_8p^(%Q<IyQmz*DR*{Q&|P@#yYn4);6Wa-M-)$VKc
zrTJw^<pCA;mAX|@)oe8+HQ#IB*NxV9H#A-^X-sd5zTtDzzS*!vsa50_Ya2Pdf^Bpx
zbxz-Y-ZgON&fVsFb@$7<i+XZ;v-&a~r1qx_oPC%wnEEJVDC=?lldDf_hFhL>j|@Mb
z9sM+h7-yIepVWIvd=>Y)XzKR#)XeT1j<?!#-t!mU-CCGlggyu?nJkB_<bCW|eYv*z
ziR!cPy2cmFuWlPb-=a6;w@!ag*pB@X{?m8o#IEU{#=iIg)35lmfin;%)QHeRj3K>H
zI8-@&65hs?W6g0$Tn9b?K9MevmJ{6JljSOT6GbGYHWfM5G<6M41g#z&E8Qx6H$yI?
z50eHn6Z1ODBi1suSavH8F-{EUJXaTYHjh8AJ|73)7XPq7g<!OhsW6Z5mdKE3k(jTz
zvII_ILb6OMKw41-$PCI}kh79!lAl#5RdiS4Q~ID>t>OirQ5IDz)!g7S$y<#pnvPn`
zTCcP(>sag3>W=B<=vx}l7>pa{8`&AN7|$LpGx0noeC)GnyV)so9SefRgyl6WA8Q%w
zeVfO&`F8I1(hk7k+3~B6fhW|RD4pIpx4EPekGo2^q1>k2n?25Xx_BviQ+coYJoGK~
zi}SY&kPV~?{2VkK+z^r;>Jw%VE)ao<Sd8qCsyuln+BHTamLqmAZW>-y@)AN%A4?QY
z!X(X~xtpASHaNvFl_z!g+(cSqdP;^<ru=#G^Gg@HvkESTTr$e$&EC&>mD`$^mG5`i
zpn$&+Rk%>pUtCp^dd2Um*){o6wlZ|t=klqF!OHfk>gs};%-W>7nEHr@(CeX%5lwM7
zQg7xp*S7SwzHLLbOLn+*Uc0?`NAB*$d)wWCJsW)~{h|X4gV%@BpPU*_8L1qd8t0!(
zdySmVd!st{bK%K{=9Rj&=Ffv)KX1|hFxkC)82{hg(&3(fkq6-NB>?O?0kGBtAd?QJ
zm0$~|LIBLj0I*U5i1iA9XzK$|?dCuG2lOlFq=GX}9v}f{nuc(O=>uZH1yBw;!3bD_
zU{(i`gLA_m=mOLPjX+-zbO<GcGvX|w5itp8bfQQ*WD2qkUVrG}Of3%84DUniXj}9p
z^ivET&bv}EkFcb0hLww*!SUjPaCh){ydAy{ze{3HQcDVv9w+S}&=VpFugO%&D#;1t
zQRE9078DOC<tZDexT%V%>8W#QsA+O&>1m7Uxak_`<>>nu%o*kx!<ev4CCswS!z|vc
zXx2(L9kvhb=^T<AuQ(IA#JHxo(|J^QKJ!-bo#1ES9~a0Gv=U+xdMjKf5+SN1MlZG|
z-Y;=YGG59>T2DqomQ{`*59GHMHWa@qZ7S~^!Kl)z@vEz7SZjuAWovinywxMoS2FN7
zEH|1t%4A}H?2754xrD_j%Moi{n>gE7_6iP##}7_;J59Lg5Ifz(-D^B~y{dc!eQ)?H
z1`GsQ2d{)Cgfm98MOmHv9&;s5@6?xs(nO0hxa6LcxN|CLdl`M_GqP+i31t7w9nHU9
zkY40hVt!S*RG^%pl2DDR1@+)Ms)_U_Lks^c#r9*J-d)LeEAIFAEIl9{kQ}rbihXiz
zxOZfJbZ?wtQtXx5l+ld&8>=~scSi5kK8P(dtn9DO{nh=s_)Emb(M`^+uiKA)7VrA)
zEB#tO5ODlSVZM$P@WWh#2Fx+Iz|6u~m`%6|24UXdCqxG`1g0=2kOkd@#-Q&AR(P%P
zMdTpvAy(jBM;jT2tUyk{D~~EF3{{U>K(nFk;T(JdLx-`&6l3PF0@xsI7Y>87!d2q7
z@J9GD{0|aKlAELyq`{<P1SLWl8G$UCY?<7Fe1t-eqMK5MvWrTEsvFK%hiUBK+%$p?
zLsvvEML)#g&IlN<F{v}XV@_oeW0_<<%_hn=1;_jd$2w;fmm@bl_bAT=UK2hFz8Cxj
z0?vX$f}2A9!g(UzqH1CcVqe9dOEgR7NySNf$XLi~$;ru!DDWzBDsd|Fs0baAQPog0
zQ+L;h)y&gs)gIT`*5lMSHi$92ZZv<C*~A7u*Ss=kxA3yOWsS12wQaP=IJh3~J;CFY
z>in5#@A}YP&ZEYQ#XH-V)Gsvv6_^~14ao?j4lj=6k7|w9iW!UZJhhvUlPHq(FxfQ)
zq?V>>q`%8dxgeZ1aw#H*HTOZjUjc35y<*QR6jwV-iRB~}tyPXS=-S45n}+?ysv9OZ
zzqJ(K(rR1j$hs}xHG4PtzG(M&@2Lj@{VyISJQ5#z^W@U7{<DMUcV5Jft50HIPQ0$2
zPMURnYc#L4Ah{^9q_ASN>hV|l=i6Vte3RLV-yYuK+dKCw{z!laG%#N$3ABJM%p<0O
zYA^skKqQbP%m$r-WBwLFh0ujLomRwONMWQ8vL5*f<`CmhgJ?Rm2f718hVj63W7)9r
z*mpQXTq~XnpG|@xNg&xFjU_!Gq>|CVvs#J#1w}9=HDxE2J2egUAWZ`85!yYvKKcv>
zJ4PYKJ*G+KW|m8=VQlv7TJY|}%0<qN;Qq<;jdz`Io&T%Amf)T+R)kiRM@&}SK*Cw_
zq*ShSi_DnpraX&+wqk%%q4J>0wyKDli~41a=UN19Bb{{JVSQ=?d&3H&&qviwE*<+|
zre<Dfv2EpG-EAvqm*;>!9^?4cDF}{Txa*#Kx+jZQvyZXwvVVG@WYFu7)G)>HwaCho
zPBE;pGpDX4cqED@Z6)`nTsY^LE}F4-ek7|Lj+#LpTmF}Vfuf?4z^j_2v}GSEI;v7@
ztn0YySFg7=Mcq_r{?^*qM(m*I?Cd&z=li|$-7G!jeOwO;25=992S<m5AICiD8%90T
z7zuw~Iy(4bWgIcVFe&g-@|E0cxhctM!5QXRlG$%>X5MzsmCeV$vtN*Wk9q%cvGzm6
zlGZYQ`Nc~9M~79`)tR-DzwAEIeH!_EZe4SI`^$~5?i-97Prt=)N^Q<<UHq>3ePg@o
zht*Hi&(<B6UGm+oJ@<X`{k8*#!}y1J%19sX9snX}!Sv&sgM$rz0I=Br>|HuI*eO3a
z*sFk(4fq>KkN@xQ6^F(cm~$_2K14li9;XkV|9<@!M&f%8Nam8p00009a7bBm000XU
z000XU0RWnu7ytkil}SWFRCodHT?u#;Rkr@KbUNvfeG_5`YY-wNfPp{+o{ADgGcxep
z5O;8ydCWk3pWowCbe1RjK4lzy;4&jKqk}U-a1=+ud7z@;LLwlFC>S)v1jwFrI_XY2
zop;WyuIf%_F<t2n-Ic!IS6#QNZr{59Ip>~x?x|CCgE~7q5lBOq0>MKUdH^|7ARquk
zTn+*P5D<VNPJtl~KcE2u0x-lWFvQ`9_7{IpP!Ma2Ag8UrfAvN=8&%31Xc{{x^H{Co
z5Bf3s5%3VAbLh&IH$@c|uDl{TIw}*k&TyNplZ9bWQ*bafH>lHMG@8ELxbaVWHf?&T
znHpfF&E_pZ&^rD;1;7qozi0Q$(`V)7{8<+kI>wdbHk%E>!9AN2eO+^{$KB)hHtVU6
z4;0@%KYw`%{kM%aj|)L>`1``u*EM%B_Ep|f_<QCY-I>7iHT~t6&rZsneaT;XVt##n
z3*O&%0=#!k4Gq$@x_XoAC663)d$?Wm=U<vc73yj8s7C4ye?8%O?|sJH9hZdp8`Rda
zV)AbC6#gCsfWR{`**=IUEL^cLOs^a53WvcKmwX4axn1m?g`lRU8J`}Mu9cL0kEW(p
z*S~w@nA^2_^<7tUJ89F|+dbgAd&6H>XTrha?_sgD)BZa!4dhf)W5g$)o+5f!@!6p=
z7>#E6lGpa0z~7?)*juclePn!mT$U>W2F?VqT7?}(LqHHhL#3+DoNXk5_#Pb{(lwSP
zZ<=X|iSbjYeFoatR`H}3=!RdX3qeSTbc>FTPC<i4yKox}dW8l|7rAq|($3Sz1I6E)
z0PuMxfT6vuJ<7S!uFMY$3&G48V-OV;j??9JIDGgtTjt^~;o&;W${mB481A6nNfvWk
zG}X)k!0qh~kGnVT5kF&`L0DN3I5DFSlBQz~XE21}l1oOBfJjYXO~BgEYVC0M-fop2
zDE{6E#$wJW!`duuVRDQ5Vt#39HNO0^LN1EPNWEbGuwmiKId5c0K$-Du1N3`J+mCki
zh@bx#O+t{9<6v}qdxy{j&CM3X#uzYdYMN4g6Ac50s6((}=&dwR{Cx?4=mpk}4k7Lh
zja}8gbI(8i^%;C~;;e!(COSeeKO#br9p##*Gxqi|^SQmA8UnbUF8H0>&5WKoW3<xS
z+uAzF<~aqE$%3S$Xk2`8s#2WoA?J>vT<}n4p!j<e0B)WjD5-NnRG#WdwxQH=@&k}$
zOR8R0dom`Lc=?)q&RcuB7q?Ti)x-Dbg5R}n#wZVUe2wiO-Z*u-^!Q-=uplma_-CN_
z8Dw8c=~1luSn8d}KeqH68(WbvJ_VB|4Oi+hk~;kO6RAfiC;c;{;CF65PXUWd$`m2j
z>ve)Qtntp05&Y$`N~L&mauhNrjZlt#E%Rdnz*4RdA(~WsS0P~4Cker*^h9K3rID79
zAhx!)2_f*-6tD+E@|~5o_HbR*DQEm#fix64W;xPOIEsuwz3>ej`Mg}wlx+M?%^s;7
zt7<_1|D+24j|zb6{d*Duo)R*nQ%A&N`m}UK6}Gim#oV|jr-^I5{&3u6Y!z0&JjK=N
zf~iA{0UNr_&1RH*=FkdaRxmwXu@ih1pW6b!KwO1@&&hNB<m70YDveg=#B4Xn!v>f0
z=VYU~zns|bF>|Ig{pE8Oi&e4q8Sf>;d>$HnJ*g4^2E{@!BWJXj|MK2>t{)#4iCiKM
z_X3_Wd3!22SVWGECF_5t9Wx1ebd<xtk&rM9vu2GY%sRn*f5AMUxid3;-rS5H$2b;{
z-mVQ6!@O~ym}o<<xL6?O&7IIAJ-pEtPt>Ve1IRabo*K&M<b7Y)<TC!a?+aI7*L)G{
zfCa!8K?T~>e+mp(08G`jsI~A7O*rz=A?*I(Ym_y4*Z<vrc-}ABZsXaPWE_55X~M~}
zTIJaLcOUp224n-2zVikie7yn%YnEfm!5#SRe*5FmzkQB}p11+o{Fp|=>BHj<`2EIL
z@XCfeu<hfsczydWJ6bb}o?S)x0g-o5D5qWnbezFoiTRwunk%qdX1+L6i)|mA5jidY
z{0}jolhGWEzbnk=GvPS&`Ekzpd{Qp}jg2@(@Fp)^ip7aWa9kF~U5c$wF9JUgt8d{b
zE$t=?X?=dwwZ3ox^kn_>GtW8G6RGFlFM<@CjE-OtU#5a;0kB%yXw(N%<3n(~<Qsu8
zZ|+#=eP8Q&gdZK`A?MLQF(CmT9+sYjF8fKIN9rETN>sBeG(H{~)Y9EAyo%kT#Rg2j
zpdOnacnjrpoDswQL%S&=xD)LJZ^c?^7~tUKxVSW2U-+UJ`I8c2{Q|sd4FLUcTr-0M
zaqMa26wFKpz7U~s3AlNV<LRJWL`!?_{RP)vp+K58DGni_Hng{J{G%C}6UQspd^w|9
z18~65>^qhrHMbm9<`9gTLcVV_VCkY<jGCK;u}-q#%LEum+wQu00v693hqg8|E+F?e
zr#1I3!kny;lw%R`{p7X)SelGzum&R`J{*=hGlr!kVHo3VG@_P-B6~JTzO@CWjwnQm
zSR&sif;%k~Gk*Hv$KL#GDAqtA4UzOxf>cW$bp+1aV?*4j`n;5NQvl5P$NHC1)DVqF
ze?14Uta}S5dTDmrRR#Fn;tPAZ>c6M&cw`%zt17X5(`x+mXPZPMYENh$xHA{IIn#Q&
z<WuL|wgH?ccYQ_s*J*Sb=&abC^R8TNFm``&#I=ks9J{Fj!m;GPmSYjgeN40tkN)>^
zG}YF_5*3HIuofIEDMeLB1jc8M#;C+D(d52>)gx`#@~i9ZqkAV_+e~x*&R~QFvHtHw
zX=O8P?QIyJ9<S9QfZh~|rm@{l%9`SLom|1Yuu=8~?mQv6f;~cV->Ss9*B|&g;0hMp
z3Alm-uHb+xn7Ts16&!E{`__2XkJh+p1UhOAxPk+&;D9SQ;0g}7f`^<d*f9ujPKWd{
zNwA(ei(>~4p*<oBsnf?}Y_tY_z+HhWM{(ruZlLc|d{fqlC~^zWp3Kn@&XYx@s8428
z+W?HF@nkVpZ73FT0nX?mGjp-|=hGB_f%7#j<?C8Quwu<K-PpY#ZzNXSa*ba=b=wuJ
zprVklr=6G|NiR*v4kFXBPw?d4Dy67aGq*W0e|U1NJv0<}=fG*4{I=9DV5`;u*19?o
z5{B?pJoNC@$d2dNpxSZdpKs&0Z&je7$@!^)U(mv>Mp`Hum_uHM8Ep9TllPO>m-^Cs
zpVwg1bK6i`-w1z*2vDs7WXVaJJHyU=rk@Vk3#W^iKzdl}7D4^3u#E2B8*>%rGlt8u
z5=Bg)^vMF>N2OW-kTeo=C=#;#Uwg6hiz=At%UPznGuZL$9uX3jIcgXzEoL+}ne7De
zePX!NLIZ__1sfvp<d*$}QF`*n@c|qD`l$#Bn=o#aJJp@HvT{duzX6s;924PRO?YW@
z5nghs!8um&p{H&~w#Fia`1v1)e}&ROaMy!>aY5fTR<P*VpVM{qeXU#N9z3DGRUK^<
z`8ISD3xNMj_Il@&dJSMm!=vkNfHgFv_xE4u6KU4aaHMFg_cJ-U=WN?_Kka)L1Nw>(
zUH5HKQ7-^w@TCk-ATqS$+;^2Y-9Yg{p~En8>~LcE&~OCN2SO-y!qgT7qsff0kWR!$
z^D81!lBm$TfXL;}=Y9YJK+SF{!{d*<g%X#sAaChntiD3qVd2a){!IDu!avzV!TF4o
z9xGgkjB%soL!aKOw*koiw_`(h<}=y@MgPFg@A$O|ZKq#6{$@AkGvc*R?Si#UlQDr#
z3~i~w{&Resf;DUNF*}X?fEFV@D0-Vh!)meXqjD6C9^>=}ZDsk}pA}{0WdF3_)n|T5
zFNK7P(SF;zrP#jx9qieE2>F-K@p;gyHGt(@rI_!hEt)McpP}lbFn3v=a0JCAI=-Ld
z^HfmLKw}#PgVO)j-n&3BpR3@}{)WrPilHHGIK3w22T8R6=u<`rMwjnBh~jFy5zt}A
zN81hv!KkMXNNPDnh1mq7H@>uwma1@k3;2!wtQCOj+9tn%uigkWBw{AL|5)BofhX2&
zA+XZ302%fCsUzg9CimQPVv`f;C6O8|{n>ML#6sZcPqU_9<QKRdEq1%Pk+|_c^D$ZZ
zpTp?8<hlYZ8Sj`5DMd7vdH&JsA-;|ya5+f%=&CEBr`QO8Na{5Jql+JdN1k1QvXkYg
zs3w&k5rOovX-K7)MutMt)1dz*eGsClwG%b<Mnp!CZzRsYY@Y!MK;y}yczWw0oN9FR
ze!+NJMke*DMR;s+x~Aozt~D6zfBzb)s_VMoY^H}`58OE$vvRV#ocF~8^#ULQLiXc}
zUQkuE$PY|=DWHMVx2^eKu~LOA6rK$<8jY@}eezhf09Y%}*qLuY{&mxkmF%SY=ox@6
zF~uhVd#X`JdPSa>DPe!$!>g7c<pOq#ONp0~Ns93gcerW+2#u6Nz)abBxO;&=#J<Sv
z@ytFfgdg<)OXiP7*2GlLRC*q2ahR^|-aI}H3;8oRM~>oyleK6R!5=0O9Kit+4(r(6
ziv6QJ8-P(X4Sa3SakRGjFIv?a0G4_jZD3}d!^RD-cH>&cq5?d2jrKkeAp_;!Ur#;&
z9W7Y4e9epUX=T6m-g%gom8l&2YDT>Vpn#D2K2TLOYC9;D1)wkDRn>N#8T3J_^Lk0W
z2GEDo5^3Wxdgdfd9w7&WOIUcVywJ$#^9sz{H)rNATQUdN%*}+3f?}K#TL)6Cfb&`3
z%&Qjw3IaWJ_$1z;4dDsM&%YQ~=42pUgopbkSWmW!9lu+5e2Bl(Hp~!=)psw#l#5d7
z<59t4!9`Er%bRtn7l4p3WRMY9&31sf7Q0{HC$^-K>G(;07G_Pk5PmWfQbk{$>nD;C
z$aX+;iw(co_@<~Qn^p+B=a%_MiWA>XQ&sn1{z<(6(1#*dufHEF>#Fe8m!&8!F2%dw
zHlg}-8UFYJZG<8tdn)d^eHPNC3G-m$^7_440RBMV3*u1l6Q_-M<NrRRMWepLk*_Lo
z-`bbN{R`%2;?^Z|DbJ;SGxcT1^sB{@@yW=TI0_Rc#C0ow;>ckXuK!rmQ$k)#dR$sG
z@^U71!@qOSF|2)@pOpG;Qm+AE#NKTmpy<6aRJ-8I$ex7UR10>zRSMI&Dx4*+aC%oe
z$>ksZdHCl3@33X-u5M#~!F>8s>bP;(@Z1iZ5DQ57E<W`F&{Exn`H?9pD7b1W{*9JP
zOVBg7&1l$Lik;?q)RUhiU1Oo)Yy>(pe>^RmdH=2Rkv1Y;;r0f4a|kUQI?AO7tZbEf
zJ(*E203jiWBR5FKRnt*$=_L9l06hS)bRb+XpPQ(|6)W>G1u?i-W6WoCJgU<Q$tb+@
zD$Jffj@}LJ{)WH{zrK~CgV-P9o6Kh%ssy0+_)+}f)#J#erD6&w@S}9^J6Kv&%Joq)
zVCl;1aO2ciq)r?I|MikfXnOjsHU;7FXSU<j4Ywhj0zZnsHR6e<UbYvt;`zsJK|)+S
z62!-nogR|)RM~kx=eB4z9^a39Htt8sxmx;YG^zT!qj*qZKBH+t(F@!0k$*VC+$MjO
zDHQiTunA>lRkTWYJ9y;~2lKhQP~5|72z2_#^8q&npdI^OKWZnM4)jd~l<t#{4Y+Cf
zJmk?iep_h!z8~!6w6%M3%WgX$8wEf`;a??H_G2r4_wX%<Bi{h4{7*`b`2}z{1;B(K
z1YRBPTX!e&hj*a(pcMU@bm?3ynVpH1H&2p)w`(&X<ooZ-37-u`n^q8?b=dLG_XXa&
z|9vVMfYDf2c#}A8B~Ph8B=-cMr4A=W1b8$?C6oGZsl=!338XY{hCK?{ZdF8R@clrx
z<=e!HAY|AD>xFIKK%PKOm(9u+`!IG4P>PAtq9@Rh0<!HW6ucJiefk6H>JE!{0DuH!
zkK`y|6ZXDM&ju*fYcM2?dkd?0BQd?AvKl9=rI$l^%Bzo%82<d<-V^wnjLl;2>pwp_
z3!t@d`N^j}MPee&>2}gr!FRvB)4o^~UCPYDMfxiI>b@c+MsVI_ZG?n%#SdILF9)yD
z8iBv~&32h6$j=)^`5;_--)1F7aK==Pycf`JwRRcIa&EjD`NGhX@h9M+TM4YCmA;oJ
zrO3=nv3MeD1n(z%`&dZj&7(JU#eehVv~0XE^yJ%^arZ3+;^s6cinJi_LRv*8MlRsh
z{Xp^er2%-zvwii|iPQND<~cxwB;)S&_u$&{D%8_7aQMh%>8YP30yAe!z=De>;j*0J
zN>6b7(K|VAAJyy)=J$-BZpMp7n5{I{+sN@1<}jm{UYm<6az<l#QupT~-a!ux6n{`>
zC)2KLBDKeY!To$ha&qG2BZqfAotPNM^BbQ^H8u4$*;5z(vZ|_v=c1LgH4&aJ8cR)s
zhZ25=_;#ffO9d0sLd30K^&jiDoI6+3R|Htse-FYDw`bL=buUu;*yY6jR@v$9iMtOO
z{Jm)a77X@ba<Dr`c8-aSjUKyf*>%$<P8>f%7edh>l!!{woQDqvAyLn?wOiY*$B%zo
zv32X~pEWczvH$rLZ56cfy6vr`0a$epDA9d}4E`PkfT>4BU?%e$j!CrfB%e1P1~}M{
zuQ8DZRRHLI>|J6XE5CNbPoY`u^Tv~L_DESt0J@K9biv&;RPgs@1TwMtC4bqg&n_U&
z^RqpU@fmCZV8(Krcxd8Db|Y=v9v+%_sqO*ye5%7a4GH|cY5=AL^#T?U?(IAraOf}Z
znfd(s?_l?Sx}{(;kM%5!ES&ry9?r8?uz9NYQ(Ynr1^j&q08@d8z|&jaWMSaE-1`Sx
z2*lKk?$1KN8*2mJGw(g3`l+riN$dE3Q~;P7L<D9yC)d4ncJO{>Cd=wx?7hW&8J3pu
z_e%g|LIn2Oqk!C_wTCQ#s9zKa2tdEcq}@UR0njdQ`-LnZ0R1A9b_)drK)bx{7qWl=
z^ovZ|Eff#{?eex?$N~b;FEVMjP(T2*%iDe-`+v|7m{y$1dn*6{002ovPDHLkV1lnB
B5rhB$

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-opgen-0.1.0.0/tensorflow-opgen.txt b/docs/haddock/tensorflow-opgen-0.1.0.0/tensorflow-opgen.txt
new file mode 100644
index 0000000..a81a0f9
--- /dev/null
+++ b/docs/haddock/tensorflow-opgen-0.1.0.0/tensorflow-opgen.txt
@@ -0,0 +1,48 @@
+-- Hoogle documentation, generated by Haddock
+-- See Hoogle, http://www.haskell.org/hoogle/
+
+
+-- | Code generation for TensorFlow operations.
+--   
+--   Please see README.md
+@package tensorflow-opgen
+@version 0.1.0.0
+
+
+-- | Wrapping of TensorFlow attributes into Haskell entities.
+module TensorFlow.OpGen.AttrVal
+data AttrDef
+data AttrCase f
+AttrBytes :: (f ByteString) -> AttrCase f
+AttrInt64 :: (f Int64) -> AttrCase f
+AttrFloat :: (f Float) -> AttrCase f
+AttrBool :: (f Bool) -> AttrCase f
+AttrType :: (f DataType) -> AttrCase f
+AttrShape :: (f TensorShapeProto) -> AttrCase f
+
+-- | Type-reified representation of TensorFlow AttrDef. Initially limited
+--   to just the types in Op descriptors.
+data AttrTemplate
+AttrSingle :: (AttrCase Template) -> AttrTemplate
+AttrList :: (AttrCase []) -> AttrTemplate
+AttrTensor :: UnusedTensor -> AttrTemplate
+
+-- | Specifies the optional default value and a set of allowed values for
+--   the given type.
+data Template a
+attrDef :: OpDef'AttrDef -> AttrDef
+attrOriginal :: Lens' AttrDef OpDef'AttrDef
+attrTemplate :: Lens' AttrDef AttrTemplate
+templateDefault :: Lens' (Template a) (Maybe a)
+templateRestrictions :: Lens' (Template a) [a]
+
+
+-- | Rendering of TensorFlow operations as Haskell functions.
+module TensorFlow.OpGen
+data OpGenFlags
+OpGenFlags :: String -> String -> String -> OpGenFlags
+[outputFile] :: OpGenFlags -> String
+[prefix] :: OpGenFlags -> String
+[excludeList] :: OpGenFlags -> String
+docOpList :: OpGenFlags -> OpList -> Doc
+flagParser :: Parser OpGenFlags
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/TensorFlow-EmbeddingOps.html b/docs/haddock/tensorflow-ops-0.1.0.0/TensorFlow-EmbeddingOps.html
new file mode 100644
index 0000000..5e59fe3
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/TensorFlow-EmbeddingOps.html
@@ -0,0 +1,16 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.EmbeddingOps</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-EmbeddingOps.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-EmbeddingOps.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-ops-0.1.0.0: Friendly layer around TensorFlow bindings.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.EmbeddingOps</p></div><div id="description"><p class="caption">Description</p><div class="doc"><p>Parallel lookups on the list of tensors.</p></div></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><a href="#v:embeddingLookup">embeddingLookup</a> :: <span class="keyword">forall</span> a b v. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>]` b, <a href="../base-4.8.2.0/Prelude.html#t:Num">Num</a> b) =&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> b -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a)</li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><a name="v:embeddingLookup" class="def">embeddingLookup</a> <a href="src/TensorFlow-EmbeddingOps.html#embeddingLookup" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>]` b, <a href="../base-4.8.2.0/Prelude.html#t:Num">Num</a> b)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a]</td><td class="doc"><p>A list of tensors which can be concatenated along
+ dimension 0. Each <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> must be appropriately
+ sized for <code><a href="../base-4.8.2.0/Prelude.html#v:mod">mod</a></code> partition strategy.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> b</td><td class="doc"><p>A <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> with type <code>int32</code> or <code>int64</code>
+ containing the ids to be looked up in <code>params</code>.
+ The ids are required to be flat on entry and have
+ fewer than 2^31 entries.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a)</td><td class="doc"><p>A dense tensor with shape `shape(ids) + shape(params)[1:]`.</p></td></tr></table></div><div class="doc"><p>Looks up <code>ids</code> in a list of embedding tensors.</p><p>This function is used to perform parallel lookups on the list of
+ tensors in <code>params</code>.  It is a generalization of <code><a href="TF.html#v:gather">gather</a></code>, where
+ <code>params</code> is interpreted as a partition of a larger embedding
+ tensor.</p><p>The partition_strategy is &quot;mod&quot;, we assign each id to partition
+ `p = id % len(params)`. For instance,
+ 13 ids are split across 5 partitions as:
+ `[[0, 5, 10], [1, 6, 11], [2, 7, 12], [3, 8], [4, 9]]`</p><p>The results of the lookup are concatenated into a dense
+ tensor. The returned tensor has shape `shape(ids) + shape(params)[1:]`.</p></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/TensorFlow-Gradient.html b/docs/haddock/tensorflow-ops-0.1.0.0/TensorFlow-Gradient.html
new file mode 100644
index 0000000..952f801
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/TensorFlow-Gradient.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Gradient</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-Gradient.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-Gradient.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-ops-0.1.0.0: Friendly layer around TensorFlow bindings.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.Gradient</p></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><a href="#v:gradients">gradients</a> :: <span class="keyword">forall</span> a v1 v2. (<a href="../base-4.8.2.0/Prelude.html#t:Num">Num</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 a), v1 ~ <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a>, GradientCompatible a) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 a -&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 a] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a]</li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><a name="v:gradients" class="def">gradients</a> <a href="src/TensorFlow-Gradient.html#gradients" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../base-4.8.2.0/Prelude.html#t:Num">Num</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 a), v1 ~ <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a>, GradientCompatible a)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 a</td><td class="doc"><p>The output of the graph.</p></td></tr><tr><td class="src">-&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 a]</td><td class="doc"><p>Tensors for which gradients are computed.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a]</td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Gradient of <code>y</code> w.r.t. each element of <code>xs</code>.</p></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/TensorFlow-Ops.html b/docs/haddock/tensorflow-ops-0.1.0.0/TensorFlow-Ops.html
new file mode 100644
index 0000000..919f1fe
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/TensorFlow-Ops.html
@@ -0,0 +1,122 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Ops</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-Ops.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-Ops.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-ops-0.1.0.0: Friendly layer around TensorFlow bindings.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.Ops</p></div><div id="description"><p class="caption">Description</p><div class="doc"><p>This module contains definitions for some built-in TensorFlow operations.</p><p>Note that certain, &quot;stateful&quot; ops like <code><a href="TensorFlow-Ops.html#v:variable">variable</a></code> and <code><a href="TensorFlow-Ops.html#v:assign">assign</a></code> return a
+ <code><a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a></code> action (e.g., <code>Build (Tensor Ref a)</code> instead of a pure value; the
+ returned <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code>s are always rendered in the current <code><a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a></code> context.  This
+ approach helps us avoid problems with inlining or common subexpression
+ elimination, by writing</p><pre>do
+    v &lt;- variable []
+    w &lt;- assign v 3
+    render $ w * w</pre><p>instead of</p><pre>let
+   v = variable []
+   w = assign v 3
+in w * w</pre><p>since the latter could be reasonably transformed by the compiler into (or
+ vice versa)</p><pre>let
+   v = variable []
+   w = assign v 3
+   w' = assign v 3
+in w * w'</pre><p>Ops should return a <code><a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a></code> action if their original <code>OpDef</code> marks them as
+ stateful, or if they take any Refs as input.  (This mirrors the rules that
+ TensorFlow uses to avoid common subexpression elimination.)</p></div></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><a href="#v:add">add</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))))))))))) t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:abs">abs</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))))) t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:addN">addN</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))))))) t) =&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:argMax">argMax</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))))))) t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></li><li class="src short"><a href="#v:assign">assign</a> :: <span class="keyword">forall</span> a v. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a)</li><li class="src short"><a href="#v:broadcastGradientArgs">broadcastGradientArgs</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</li><li class="src short"><a href="#v:cast">cast</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dstT, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> srcT) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 srcT -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dstT</li><li class="src short"><a href="#v:concat">concat</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:constant">constant</a> :: <span class="keyword">forall</span> a. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">Shape</a> -&gt; [a] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a</li><li class="src short"><a href="#v:expandDims">expandDims</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:initializedVariable">initializedVariable</a> :: <span class="keyword">forall</span> a. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a)</li><li class="src short"><a href="#v:zeroInitializedVariable">zeroInitializedVariable</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a, <a href="../base-4.8.2.0/Prelude.html#t:Num">Num</a> a) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">Shape</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a)</li><li class="src short"><a href="#v:fill">fill</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:matMul">matMul</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))) t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:matTranspose">matTranspose</a> :: <span class="keyword">forall</span> a v. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a</li><li class="src short"><a href="#v:mul">mul</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))))))) t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:neg">neg</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))))))) t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:pack">pack</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:placeholder">placeholder</a> :: <span class="keyword">forall</span> a. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">Shape</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a)</li><li class="src short"><a href="#v:range">range</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tidx</li><li class="src short"><a href="#v:reducedShape">reducedShape</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t1, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t2) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t1 -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t2 -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></li><li class="src short"><a href="#v:relu">relu</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))))) t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:reluGrad">reluGrad</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))))) t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:reshape">reshape</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tshape, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) tshape) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tshape -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:restore">restore</a> :: <span class="keyword">forall</span> a. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:restoreFromName">restoreFromName</a> :: <span class="keyword">forall</span> a. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:save">save</a> :: <span class="keyword">forall</span> a v. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:scalar">scalar</a> :: <span class="keyword">forall</span> a. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; a -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a</li><li class="src short"><a href="#v:shape">shape</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></li><li class="src short"><a href="#v:sign">sign</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))))))) t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:size">size</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_type, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) out_type) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_type</li><li class="src short"><a href="#v:softmax">softmax</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))) t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:softmaxCrossEntropyWithLogits">softmaxCrossEntropyWithLogits</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))) t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</li><li class="src short"><a href="#v:sparseToDense">sparseToDense</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) tindices) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sub">sub</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))))))) t) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:sum">sum</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))))))) t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) tidx) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:topK">topK</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))))) t) =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>)</li><li class="src short"><a href="#v:transpose">transpose</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tperm, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) tperm) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tperm -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li><li class="src short"><a href="#v:truncatedNormal">truncatedNormal</a> :: <span class="keyword">forall</span> a v. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a)</li><li class="src short"><a href="#v:variable">variable</a> :: <span class="keyword">forall</span> a. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">Shape</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a)</li><li class="src short"><a href="#v:vector">vector</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; [a] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a</li><li class="src short"><a href="#v:zeros">zeros</a> :: <span class="keyword">forall</span> a. (<a href="../base-4.8.2.0/Prelude.html#t:Num">Num</a> a, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">Shape</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a</li><li class="src short"><a href="#v:zerosLike">zerosLike</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><a name="v:add" class="def">add</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))))))))))) t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns x + y element-wise.</p><ul><li>NOTE*: <code>Add</code> supports broadcasting. <code>AddN</code> does not. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:abs" class="def">abs</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))))) t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes the absolute value of a tensor.</p><p>Given a tensor <code>x</code>, this operation returns a tensor containing the absolute
+ value of each element in <code>x</code>. For example, if x is an input element and y is
+ an output element, this operation computes \(y = |x|\).</p></div></div><div class="top"><p class="src"><a name="v:addN" class="def">addN</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))))))) t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t]</td><td class="doc"><p><strong>inputs</strong>: Must all be the same size and shape.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>sum</strong></p></td></tr></table></div><div class="doc"><p>Add all input tensors element wise.</p></div></div><div class="top"><p class="src"><a name="v:argMax" class="def">argMax</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))))))) t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>dimension</strong>: int32, 0 &lt;= dimension &lt; rank(input).  Describes which dimension
+ of the input Tensor to reduce across. For vectors, use dimension = 0.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Returns the index with the largest value across dimensions of a tensor.</p></div></div><div class="top"><p class="src"><a name="v:assign" class="def">assign</a> :: <span class="keyword">forall</span> a v. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a) <a href="src/TensorFlow-Ops.html#assign" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:broadcastGradientArgs" class="def">broadcastGradientArgs</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>s0</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>s1</strong></p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><td class="doc"><p>(<strong>r0</strong>, <strong>r1</strong>)</p><ul><li><strong>r0</strong></li><li><strong>r1</strong></li></ul></td></tr></table></div><div class="doc"><p>Return the reduction indices for computing gradients of s0 op s1 with broadcast.</p><p>This is typically used by gradient computations for a broadcasting operation.</p></div></div><div class="top"><p class="src"><a name="v:cast" class="def">cast</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dstT, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> srcT)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 srcT</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dstT</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Cast x of type SrcT to y of DstT.</p></div></div><div class="top"><p class="src"><a name="v:concat" class="def">concat</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>concat_dim</strong>: 0-D.  The dimension along which to concatenate.  Must be in the
+ range [0, rank(values)).</p></td></tr><tr><td class="src">-&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t]</td><td class="doc"><p><strong>values</strong>: The <code>N</code> Tensors to concatenate. Their ranks and types must match,
+ and their sizes must match in all dimensions except <code>concat_dim</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: A <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> with the concatenation of values stacked along the
+ <code>concat_dim</code> dimension.  This tensor's shape matches that of <code>values</code> except
+ in <code>concat_dim</code> where it has the sum of the sizes.</p></td></tr></table></div><div class="doc"><p>Concatenates tensors along one dimension.</p></div></div><div class="top"><p class="src"><a name="v:constant" class="def">constant</a> :: <span class="keyword">forall</span> a. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">Shape</a> -&gt; [a] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a <a href="src/TensorFlow-Ops.html#constant" class="link">Source</a></p><div class="doc"><p>Create a constant tensor.</p><p>The values should be in row major order, e.g.,</p><p>element 0:   index (0, ..., 0)
+   element 1:   index (0, ..., 1)
+   ...</p></div></div><div class="top"><p class="src"><a name="v:expandDims" class="def">expandDims</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t <a href="src/TensorFlow-Ops.html#expandDims" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:initializedVariable" class="def">initializedVariable</a> :: <span class="keyword">forall</span> a. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a) <a href="src/TensorFlow-Ops.html#initializedVariable" class="link">Source</a></p><div class="doc"><p>Creates a variable initialized to the given value.
+ Initialization happens next time session runs.</p></div></div><div class="top"><p class="src"><a name="v:zeroInitializedVariable" class="def">zeroInitializedVariable</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a, <a href="../base-4.8.2.0/Prelude.html#t:Num">Num</a> a) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">Shape</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a) <a href="src/TensorFlow-Ops.html#zeroInitializedVariable" class="link">Source</a></p><div class="doc"><p>Creates a zero-initialized variable with the given shape.</p></div></div><div class="top"><p class="src"><a name="v:fill" class="def">fill</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><td class="doc"><p><strong>dims</strong>: 1-D. Represents the shape of the output tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>value</strong>: 0-D (scalar). Value to fill the returned tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Creates a tensor filled with a scalar value.</p><p>This operation creates a tensor of shape <code>dims</code> and fills it with <code><a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>.</p><p>For example:</p><p>```prettyprint
+ # Output tensor has shape [2, 3].
+ fill([2, 3], 9) ==&gt; [[9, 9, 9]
+                      [9, 9, 9]]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:matMul" class="def">matMul</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))) t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>a</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>b</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>product</strong></p></td></tr></table></div><div class="doc"><p>Multiply the matrix &quot;a&quot; by the matrix &quot;b&quot;.</p><p>The inputs must be two-dimensional matrices and the inner dimension of
+ &quot;a&quot; (after being transposed if transpose_a is true) must match the
+ outer dimension of &quot;b&quot; (after being transposed if transposed_b is
+ true).</p><ul><li>Note*: The default kernel implementation for MatMul on GPUs uses
+ cublas.</li></ul></div></div><div class="top"><p class="src"><a name="v:matTranspose" class="def">matTranspose</a> :: <span class="keyword">forall</span> a v. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a <a href="src/TensorFlow-Ops.html#matTranspose" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:mul" class="def">mul</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))))))) t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns x * y element-wise.</p><ul><li>NOTE*: <code>Mul</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:neg" class="def">neg</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))))))) t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Computes numerical negative value element-wise.</p><p>I.e., \(y = -x\).</p></div></div><div class="top"><p class="src"><a name="v:pack" class="def">pack</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t]</td><td class="doc"><p><strong>values</strong>: Must be of same shape and type.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The packed tensor.</p></td></tr></table></div><div class="doc"><p>Packs a list of <code>N</code> rank-<code>R</code> tensors into one rank-`(R+1)` tensor.</p><p>Packs the <code>N</code> tensors in <code>values</code> into a tensor with rank one higher than each
+ tensor in <code>values</code>, by packing them along the <code>axis</code> dimension.
+ Given a list of tensors of shape `(A, B, C)`;</p><p>if `axis == 0` then the <code>output</code> tensor will have the shape `(N, A, B, C)`.
+ if `axis == 1` then the <code>output</code> tensor will have the shape `(A, N, B, C)`.
+ Etc.</p><p>For example:</p><p>```prettyprint
+ # <code>x</code> is [1, 4]
+ # <code>y</code> is [2, 5]
+ # <code>z</code> is [3, 6]
+ pack([x, y, z]) =&gt; [[1, 4], [2, 5], [3, 6]]  # Pack along first dim.
+ pack([x, y, z], axis=1) =&gt; [[1, 2, 3], [4, 5, 6]]
+ ```</p><p>This is the opposite of <code><a href="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:unpack">unpack</a></code>.</p></div></div><div class="top"><p class="src"><a name="v:placeholder" class="def">placeholder</a> :: <span class="keyword">forall</span> a. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">Shape</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a) <a href="src/TensorFlow-Ops.html#placeholder" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:range" class="def">range</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tidx</td><td class="doc"><p><strong>start</strong>: 0-D (scalar). First entry in the sequence.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>limit</strong>: 0-D (scalar). Upper limit of sequence, exclusive.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tidx</td><td class="doc"><p><strong>delta</strong>: 0-D (scalar). Optional. Default is 1. Number that increments <code>start</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tidx</td><td class="doc"><p><strong>output</strong>: 1-D.</p></td></tr></table></div><div class="doc"><p>Creates a sequence of integers.</p><p>This operation creates a sequence of integers that begins at <code>start</code> and
+ extends by increments of <code>delta</code> up to but not including <code>limit</code>.</p><p>For example:</p><p>```
+ # <code>start</code> is 3
+ # <code>limit</code> is 18
+ # <code>delta</code> is 3
+ tf.range(start, limit, delta) ==&gt; [3, 6, 9, 12, 15]
+ ```</p></div></div><div class="top"><p class="src"><a name="v:reducedShape" class="def">reducedShape</a> :: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t1, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> `[<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>, <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]` t2) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t1 -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t2 -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> <a href="src/TensorFlow-Ops.html#reducedShape" class="link">Source</a></p><div class="doc"><p>Helper function for reduction ops (translation of math_ops.reduced_shape).</p></div></div><div class="top"><p class="src"><a name="v:relu" class="def">relu</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))))) t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>features</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>activations</strong></p></td></tr></table></div><div class="doc"><p>Computes rectified linear: `max(features, 0)`.</p></div></div><div class="top"><p class="src"><a name="v:reluGrad" class="def">reluGrad</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))))) t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>gradients</strong>: The backpropagated gradients to the corresponding Relu operation.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>features</strong>: The features passed as input to the corresponding Relu operation, OR
+ the outputs of that operation (both work equivalently).</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>backprops</strong>: `gradients * (features &gt; 0)`.</p></td></tr></table></div><div class="doc"><p>Computes rectified linear gradients for a Relu operation.</p></div></div><div class="top"><p class="src"><a name="v:reshape" class="def">reshape</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tshape, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) tshape)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>tensor</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tshape</td><td class="doc"><p><strong>shape</strong>: Defines the shape of the output tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Reshapes a tensor.</p><p>Given <code>tensor</code>, this operation returns a tensor that has the same values
+ as <code>tensor</code> with shape <code><a href="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shape">shape</a></code>.</p><p>If one component of <code><a href="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shape">shape</a></code> is the special value -1, the size of that dimension
+ is computed so that the total size remains constant.  In particular, a <code><a href="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shape">shape</a></code>
+ of `[-1]` flattens into 1-D.  At most one component of <code><a href="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shape">shape</a></code> can be -1.</p><p>If <code><a href="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shape">shape</a></code> is 1-D or higher, then the operation returns a tensor with shape
+ <code><a href="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shape">shape</a></code> filled with the values of <code>tensor</code>. In this case, the number of elements
+ implied by <code><a href="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shape">shape</a></code> must be the same as the number of elements in <code>tensor</code>.</p><p>For example:</p><p>```prettyprint
+ # tensor <code>t</code> is [1, 2, 3, 4, 5, 6, 7, 8, 9]
+ # tensor <code>t</code> has shape [9]
+ reshape(t, [3, 3]) ==&gt; [[1, 2, 3],
+                         [4, 5, 6],
+                         [7, 8, 9]]</p><p># tensor <code>t</code> is [[[1, 1], [2, 2]],
+ #                [[3, 3], [4, 4]]]
+ # tensor <code>t</code> has shape [2, 2, 2]
+ reshape(t, [2, 4]) ==&gt; [[1, 1, 2, 2],
+                         [3, 3, 4, 4]]</p><p># tensor <code>t</code> is [[[1, 1, 1],
+ #                 [2, 2, 2]],
+ #                [[3, 3, 3],
+ #                 [4, 4, 4]],
+ #                [[5, 5, 5],
+ #                 [6, 6, 6]]]
+ # tensor <code>t</code> has shape [3, 2, 3]
+ # pass '[-1]' to flatten <code>t</code>
+ reshape(t, [-1]) ==&gt; [1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6, 6, 6]</p><p># -1 can also be used to infer the shape</p><p># -1 is inferred to be 9:
+ reshape(t, [2, -1]) ==&gt; [[1, 1, 1, 2, 2, 2, 3, 3, 3],
+                          [4, 4, 4, 5, 5, 5, 6, 6, 6]]
+ # -1 is inferred to be 2:
+ reshape(t, [-1, 9]) ==&gt; [[1, 1, 1, 2, 2, 2, 3, 3, 3],
+                          [4, 4, 4, 5, 5, 5, 6, 6, 6]]
+ # -1 is inferred to be 3:
+ reshape(t, [ 2, -1, 3]) ==&gt; [[[1, 1, 1],
+                               [2, 2, 2],
+                               [3, 3, 3]],
+                              [[4, 4, 4],
+                               [5, 5, 5],
+                               [6, 6, 6]]]</p><p># tensor <code>t</code> is [7]
+ # shape `[]` reshapes to a scalar
+ reshape(t, []) ==&gt; 7
+ ```</p></div></div><div class="top"><p class="src"><a name="v:restore" class="def">restore</a> <a href="src/TensorFlow-Ops.html#restore" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p>File path.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a</td><td class="doc"><p>Tensor to restore.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Restore a tensor's value from a checkpoint file.</p></div></div><div class="top"><p class="src"><a name="v:restoreFromName" class="def">restoreFromName</a> <a href="src/TensorFlow-Ops.html#restoreFromName" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p>File path.</p></td></tr><tr><td class="src">-&gt; <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p>Tensor name override.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a</td><td class="doc"><p>Tensor to restore.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Restore a tensor's value from a checkpoint file.</p><p>This version allows restoring from a checkpoint file that uses a different
+ tensor name than the variable.</p></div></div><div class="top"><p class="src"><a name="v:save" class="def">save</a> <a href="src/TensorFlow-Ops.html#save" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p>File path.</p></td></tr><tr><td class="src">-&gt; [<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a]</td><td class="doc"><p>Tensors to save.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div><div class="top"><p class="src"><a name="v:scalar" class="def">scalar</a> :: <span class="keyword">forall</span> a. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; a -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a <a href="src/TensorFlow-Ops.html#scalar" class="link">Source</a></p><div class="doc"><p>Create a constant scalar.</p></div></div><div class="top"><p class="src"><a name="v:shape" class="def">shape</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> <a href="src/TensorFlow-Ops.html#shape" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:sign" class="def">sign</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))))))) t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Returns an element-wise indication of the sign of a number.</p><p>`y = sign(x) = -1` if `x <a href="0`;">0 if `x == 0`; 1 if `x</a> 0`.</p><p>For complex numbers, `y = sign(x) = x / |x|` if `x != 0`, otherwise `y = 0`.</p></div></div><div class="top"><p class="src"><a name="v:size" class="def">size</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_type, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) out_type)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> out_type</td><td class="doc"><p><strong>output</strong></p></td></tr></table></div><div class="doc"><p>Returns the size of a tensor.</p><p>This operation returns an integer representing the number of elements in
+ <code>input</code>.</p><p>For example:</p><p>```prettyprint
+ # <code>t</code> is [[[1, 1,, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]]
+ size(t) ==&gt; 12
+ ```</p></div></div><div class="top"><p class="src"><a name="v:softmax" class="def">softmax</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))) t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>logits</strong>: 2-D with shape `[batch_size, num_classes]`.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>softmax</strong>: Same shape as <code>logits</code>.</p></td></tr></table></div><div class="doc"><p>Computes softmax activations.</p><p>For each batch <code>i</code> and class <code>j</code> we have</p><p>softmax[i, j] = exp(logits[i, j]) / sum_j(exp(logits[i, j]))</p></div></div><div class="top"><p class="src"><a name="v:softmaxCrossEntropyWithLogits" class="def">softmaxCrossEntropyWithLogits</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))) t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>features</strong>: batch_size x num_classes matrix</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>labels</strong>: batch_size x num_classes matrix
+ The caller must ensure that each batch of labels represents a valid
+ probability distribution.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><td class="doc"><p>(<strong>loss</strong>, <strong>backprop</strong>)</p><ul><li><strong>loss</strong>: Per example loss (batch_size vector).</li><li><strong>backprop</strong>: backpropagated gradients (batch_size x num_classes matrix).</li></ul></td></tr></table></div><div class="doc"><p>Computes softmax cross entropy cost and gradients to backpropagate.</p><p>Inputs are the logits, not probabilities.</p></div></div><div class="top"><p class="src"><a name="v:sparseToDense" class="def">sparseToDense</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) tindices)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tindices</td><td class="doc"><p><strong>sparse_indices</strong>: 0-D, 1-D, or 2-D.  `sparse_indices[i]` contains the complete
+ index where `sparse_values[i]` will be placed.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices</td><td class="doc"><p><strong>output_shape</strong>: 1-D.  Shape of the dense output tensor.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><td class="doc"><p><strong>sparse_values</strong>: 1-D.  Values corresponding to each row of <code>sparse_indices</code>,
+ or a scalar value to be used for all sparse indices.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><td class="doc"><p><strong>default_value</strong>: Scalar value to set for indices not specified in
+ <code>sparse_indices</code>.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>dense</strong>: Dense output tensor of shape <code>output_shape</code>.</p></td></tr></table></div><div class="doc"><p>Converts a sparse representation into a dense tensor.</p><p>Builds an array <code>dense</code> with shape <code>output_shape</code> such that</p><p>```prettyprint
+ # If sparse_indices is scalar
+ dense[i] = (i == sparse_indices ? sparse_values : default_value)</p><p># If sparse_indices is a vector, then for each i
+ dense[sparse_indices[i]] = sparse_values[i]</p><p># If sparse_indices is an n by d matrix, then for each i in [0, n)
+ dense[sparse_indices[i][0], ..., sparse_indices[i][d-1]] = sparse_values[i]
+ ```</p><p>All other values in <code>dense</code> are set to <code>default_value</code>.  If <code>sparse_values</code> is a
+ scalar, all sparse indices are set to this single value.</p><p>Indices should be sorted in lexicographic order, and indices must not
+ contain any repeats. If <code>validate_indices</code> is true, these properties
+ are checked during execution.</p></div></div><div class="top"><p class="src"><a name="v:sub" class="def">sub</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))))))) t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><td class="doc"><p><strong>y</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>z</strong></p></td></tr></table></div><div class="doc"><p>Returns x - y element-wise.</p><ul><li>NOTE*: <code>Sub</code> supports broadcasting. More about broadcasting
+ <a href="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><div class="top"><p class="src"><a name="v:sum" class="def">sum</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<a href="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))))))) t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) tidx)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: The tensor to reduce.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><td class="doc"><p><strong>reduction_indices</strong>: The dimensions to reduce.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>output</strong>: The reduced tensor.</p></td></tr></table></div><div class="doc"><p>Computes the sum of elements across dimensions of a tensor.</p><p>Reduces <code>input</code> along the dimensions given in <code>reduction_indices</code>. Unless
+ <code>keep_dims</code> is true, the rank of the tensor is reduced by 1 for each entry in
+ <code>reduction_indices</code>. If <code>keep_dims</code> is true, the reduced dimensions are
+ retained with length 1.</p></div></div><div class="top"><p class="src"><a name="v:topK" class="def">topK</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <a href="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))))) t)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p><strong>k</strong>: Number of top elements to look for along the last dimension (along each
+ row for matrices).</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>input</strong>: 1-D or higher with last dimension at least <code>k</code>.</p></td></tr><tr><td class="src">-&gt; (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>)</td><td class="doc"><p>(<strong>values</strong>, <strong>indices</strong>)</p><ul><li><strong>values</strong>: The <code>k</code> largest elements along each last dimensional slice.</li><li><strong>indices</strong>: The indices of <code>values</code> within the last dimension of <code>input</code>.</li></ul></td></tr></table></div><div class="doc"><p>Finds values and indices of the <code>k</code> largest elements for the last dimension.</p><p>If the input is a vector (rank-1), finds the <code>k</code> largest entries in the vector
+ and outputs their values and indices as vectors.  Thus `values[j]` is the
+ <code>j</code>-th largest entry in <code>input</code>, and its index is `indices[j]`.</p><p>For matrices (resp. higher rank input), computes the top <code>k</code> entries in each
+ row (resp. vector along the last dimension).  Thus,</p><p>values.shape = indices.shape = input.shape[:-1] + [k]</p><p>If two elements are equal, the lower-index element appears first.</p><p>If <code>k</code> varies dynamically, use <code>TopKV2</code> below.</p></div></div><div class="top"><p class="src"><a name="v:transpose" class="def">transpose</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tperm, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) tperm)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tperm</td><td class="doc"><p><strong>perm</strong></p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong></p></td></tr></table></div><div class="doc"><p>Shuffle dimensions of x according to a permutation.</p><p>The output <code>y</code> has the same rank as <code>x</code>. The shapes of <code>x</code> and <code>y</code> satisfy:
+   `y.shape[i] == x.shape[perm[i]] for i in [0, 1, ..., rank(x) - 1]`</p></div></div><div class="top"><p class="src"><a name="v:truncatedNormal" class="def">truncatedNormal</a> <a href="src/TensorFlow-Ops.html#truncatedNormal" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p>Shape.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a)</td><td class="doc empty">&nbsp;</td></tr></table></div></div><div class="top"><p class="src"><a name="v:variable" class="def">variable</a> :: <span class="keyword">forall</span> a. <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">Shape</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a) <a href="src/TensorFlow-Ops.html#variable" class="link">Source</a></p><div class="doc"><p>Create a new, uninitialized stateful Tensor of the given shape.</p></div></div><div class="top"><p class="src"><a name="v:vector" class="def">vector</a> :: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =&gt; [a] -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a <a href="src/TensorFlow-Ops.html#vector" class="link">Source</a></p><div class="doc"><p>Create a constant vector.</p></div></div><div class="top"><p class="src"><a name="v:zeros" class="def">zeros</a> :: <span class="keyword">forall</span> a. (<a href="../base-4.8.2.0/Prelude.html#t:Num">Num</a> a, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a) =&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">Shape</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a <a href="src/TensorFlow-Ops.html#zeros" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:zerosLike" class="def">zerosLike</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><td class="doc"><p><strong>x</strong>: a tensor of type T.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><td class="doc"><p><strong>y</strong>: a tensor of the same shape and type as x but filled with zeros.</p></td></tr></table></div><div class="doc"><p>Returns a tensor of zeros with the same shape and type as x.</p></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/doc-index.html b/docs/haddock/tensorflow-ops-0.1.0.0/doc-index.html
new file mode 100644
index 0000000..e9f2531
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/doc-index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-ops-0.1.0.0: Friendly layer around TensorFlow bindings. (Index)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-ops-0.1.0.0: Friendly layer around TensorFlow bindings.</p></div><div id="content"><div id="index"><p class="caption">Index</p><table><tr><td class="src">abs</td><td class="module"><a href="TensorFlow-Ops.html#v:abs">TensorFlow.Ops</a></td></tr><tr><td class="src">add</td><td class="module"><a href="TensorFlow-Ops.html#v:add">TensorFlow.Ops</a></td></tr><tr><td class="src">addN</td><td class="module"><a href="TensorFlow-Ops.html#v:addN">TensorFlow.Ops</a></td></tr><tr><td class="src">argMax</td><td class="module"><a href="TensorFlow-Ops.html#v:argMax">TensorFlow.Ops</a></td></tr><tr><td class="src">assign</td><td class="module"><a href="TensorFlow-Ops.html#v:assign">TensorFlow.Ops</a></td></tr><tr><td class="src">broadcastGradientArgs</td><td class="module"><a href="TensorFlow-Ops.html#v:broadcastGradientArgs">TensorFlow.Ops</a></td></tr><tr><td class="src">cast</td><td class="module"><a href="TensorFlow-Ops.html#v:cast">TensorFlow.Ops</a></td></tr><tr><td class="src">concat</td><td class="module"><a href="TensorFlow-Ops.html#v:concat">TensorFlow.Ops</a></td></tr><tr><td class="src">constant</td><td class="module"><a href="TensorFlow-Ops.html#v:constant">TensorFlow.Ops</a></td></tr><tr><td class="src">embeddingLookup</td><td class="module"><a href="TensorFlow-EmbeddingOps.html#v:embeddingLookup">TensorFlow.EmbeddingOps</a></td></tr><tr><td class="src">expandDims</td><td class="module"><a href="TensorFlow-Ops.html#v:expandDims">TensorFlow.Ops</a></td></tr><tr><td class="src">fill</td><td class="module"><a href="TensorFlow-Ops.html#v:fill">TensorFlow.Ops</a></td></tr><tr><td class="src">gradients</td><td class="module"><a href="TensorFlow-Gradient.html#v:gradients">TensorFlow.Gradient</a></td></tr><tr><td class="src">initializedVariable</td><td class="module"><a href="TensorFlow-Ops.html#v:initializedVariable">TensorFlow.Ops</a></td></tr><tr><td class="src">matMul</td><td class="module"><a href="TensorFlow-Ops.html#v:matMul">TensorFlow.Ops</a></td></tr><tr><td class="src">matTranspose</td><td class="module"><a href="TensorFlow-Ops.html#v:matTranspose">TensorFlow.Ops</a></td></tr><tr><td class="src">mul</td><td class="module"><a href="TensorFlow-Ops.html#v:mul">TensorFlow.Ops</a></td></tr><tr><td class="src">neg</td><td class="module"><a href="TensorFlow-Ops.html#v:neg">TensorFlow.Ops</a></td></tr><tr><td class="src">pack</td><td class="module"><a href="TensorFlow-Ops.html#v:pack">TensorFlow.Ops</a></td></tr><tr><td class="src">placeholder</td><td class="module"><a href="TensorFlow-Ops.html#v:placeholder">TensorFlow.Ops</a></td></tr><tr><td class="src">range</td><td class="module"><a href="TensorFlow-Ops.html#v:range">TensorFlow.Ops</a></td></tr><tr><td class="src">reducedShape</td><td class="module"><a href="TensorFlow-Ops.html#v:reducedShape">TensorFlow.Ops</a></td></tr><tr><td class="src">relu</td><td class="module"><a href="TensorFlow-Ops.html#v:relu">TensorFlow.Ops</a></td></tr><tr><td class="src">reluGrad</td><td class="module"><a href="TensorFlow-Ops.html#v:reluGrad">TensorFlow.Ops</a></td></tr><tr><td class="src">reshape</td><td class="module"><a href="TensorFlow-Ops.html#v:reshape">TensorFlow.Ops</a></td></tr><tr><td class="src">restore</td><td class="module"><a href="TensorFlow-Ops.html#v:restore">TensorFlow.Ops</a></td></tr><tr><td class="src">restoreFromName</td><td class="module"><a href="TensorFlow-Ops.html#v:restoreFromName">TensorFlow.Ops</a></td></tr><tr><td class="src">save</td><td class="module"><a href="TensorFlow-Ops.html#v:save">TensorFlow.Ops</a></td></tr><tr><td class="src">scalar</td><td class="module"><a href="TensorFlow-Ops.html#v:scalar">TensorFlow.Ops</a></td></tr><tr><td class="src">shape</td><td class="module"><a href="TensorFlow-Ops.html#v:shape">TensorFlow.Ops</a></td></tr><tr><td class="src">sign</td><td class="module"><a href="TensorFlow-Ops.html#v:sign">TensorFlow.Ops</a></td></tr><tr><td class="src">size</td><td class="module"><a href="TensorFlow-Ops.html#v:size">TensorFlow.Ops</a></td></tr><tr><td class="src">softmax</td><td class="module"><a href="TensorFlow-Ops.html#v:softmax">TensorFlow.Ops</a></td></tr><tr><td class="src">softmaxCrossEntropyWithLogits</td><td class="module"><a href="TensorFlow-Ops.html#v:softmaxCrossEntropyWithLogits">TensorFlow.Ops</a></td></tr><tr><td class="src">sparseToDense</td><td class="module"><a href="TensorFlow-Ops.html#v:sparseToDense">TensorFlow.Ops</a></td></tr><tr><td class="src">sub</td><td class="module"><a href="TensorFlow-Ops.html#v:sub">TensorFlow.Ops</a></td></tr><tr><td class="src">sum</td><td class="module"><a href="TensorFlow-Ops.html#v:sum">TensorFlow.Ops</a></td></tr><tr><td class="src">topK</td><td class="module"><a href="TensorFlow-Ops.html#v:topK">TensorFlow.Ops</a></td></tr><tr><td class="src">transpose</td><td class="module"><a href="TensorFlow-Ops.html#v:transpose">TensorFlow.Ops</a></td></tr><tr><td class="src">truncatedNormal</td><td class="module"><a href="TensorFlow-Ops.html#v:truncatedNormal">TensorFlow.Ops</a></td></tr><tr><td class="src">variable</td><td class="module"><a href="TensorFlow-Ops.html#v:variable">TensorFlow.Ops</a></td></tr><tr><td class="src">vector</td><td class="module"><a href="TensorFlow-Ops.html#v:vector">TensorFlow.Ops</a></td></tr><tr><td class="src">zeroInitializedVariable</td><td class="module"><a href="TensorFlow-Ops.html#v:zeroInitializedVariable">TensorFlow.Ops</a></td></tr><tr><td class="src">zeros</td><td class="module"><a href="TensorFlow-Ops.html#v:zeros">TensorFlow.Ops</a></td></tr><tr><td class="src">zerosLike</td><td class="module"><a href="TensorFlow-Ops.html#v:zerosLike">TensorFlow.Ops</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/frames.html b/docs/haddock/tensorflow-ops-0.1.0.0/frames.html
new file mode 100644
index 0000000..1b4e38d
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/frames.html
@@ -0,0 +1,30 @@
+<!DOCTYPE html 
+     PUBLIC "-//W3C//DTD XHTML 1.0 Frameset//EN"
+     "http://www.w3.org/TR/xhtml1/DTD/xhtml1-frameset.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<title></title>
+<script src="haddock-util.js" type="text/javascript"></script>
+<script type="text/javascript"><!--
+/*
+
+  The synopsis frame needs to be updated using javascript, so we hide
+  it by default and only show it if javascript is enabled.
+
+  TODO: provide some means to disable it.
+*/
+function load() {
+  var d = document.getElementById("inner-fs");
+  d.rows = "50%,50%";
+  postReframe();
+}
+--></script>
+</head>
+<frameset id="outer-fs" cols="25%,75%" onload="load()">
+  <frameset id="inner-fs" rows="100%,0%">
+    <frame src="index-frames.html" name="modules" />
+    <frame src="" name="synopsis" />
+  </frameset>
+  <frame src="index.html" name="main" />
+</frameset>
+</html>
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/haddock-util.js b/docs/haddock/tensorflow-ops-0.1.0.0/haddock-util.js
new file mode 100644
index 0000000..9a6fccf
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/haddock-util.js
@@ -0,0 +1,344 @@
+// Haddock JavaScript utilities
+
+var rspace = /\s\s+/g,
+	  rtrim = /^\s+|\s+$/g;
+
+function spaced(s) { return (" " + s + " ").replace(rspace, " "); }
+function trim(s)   { return s.replace(rtrim, ""); }
+
+function hasClass(elem, value) {
+  var className = spaced(elem.className || "");
+  return className.indexOf( " " + value + " " ) >= 0;
+}
+
+function addClass(elem, value) {
+  var className = spaced(elem.className || "");
+  if ( className.indexOf( " " + value + " " ) < 0 ) {
+    elem.className = trim(className + " " + value);
+  }
+}
+
+function removeClass(elem, value) {
+  var className = spaced(elem.className || "");
+  className = className.replace(" " + value + " ", " ");
+  elem.className = trim(className);
+}
+
+function toggleClass(elem, valueOn, valueOff, bool) {
+  if (bool == null) { bool = ! hasClass(elem, valueOn); }
+  if (bool) {
+    removeClass(elem, valueOff);
+    addClass(elem, valueOn);
+  }
+  else {
+    removeClass(elem, valueOn);
+    addClass(elem, valueOff);
+  }
+  return bool;
+}
+
+
+function makeClassToggle(valueOn, valueOff)
+{
+  return function(elem, bool) {
+    return toggleClass(elem, valueOn, valueOff, bool);
+  }
+}
+
+toggleShow = makeClassToggle("show", "hide");
+toggleCollapser = makeClassToggle("collapser", "expander");
+
+function toggleSection(id)
+{
+  var b = toggleShow(document.getElementById("section." + id));
+  toggleCollapser(document.getElementById("control." + id), b);
+  rememberCollapsed(id, b);
+  return b;
+}
+
+var collapsed = {};
+function rememberCollapsed(id, b)
+{
+  if(b)
+    delete collapsed[id]
+  else
+    collapsed[id] = null;
+
+  var sections = [];
+  for(var i in collapsed)
+  {
+    if(collapsed.hasOwnProperty(i))
+      sections.push(i);
+  }
+  // cookie specific to this page; don't use setCookie which sets path=/
+  document.cookie = "collapsed=" + escape(sections.join('+'));
+}
+
+function restoreCollapsed()
+{
+  var cookie = getCookie("collapsed");
+  if(!cookie)
+    return;
+
+  var ids = cookie.split('+');
+  for(var i in ids)
+  {
+    if(document.getElementById("section." + ids[i]))
+      toggleSection(ids[i]);
+  }
+}
+
+function setCookie(name, value) {
+  document.cookie = name + "=" + escape(value) + ";path=/;";
+}
+
+function clearCookie(name) {
+  document.cookie = name + "=;path=/;expires=Thu, 01-Jan-1970 00:00:01 GMT;";
+}
+
+function getCookie(name) {
+  var nameEQ = name + "=";
+  var ca = document.cookie.split(';');
+  for(var i=0;i < ca.length;i++) {
+    var c = ca[i];
+    while (c.charAt(0)==' ') c = c.substring(1,c.length);
+    if (c.indexOf(nameEQ) == 0) {
+      return unescape(c.substring(nameEQ.length,c.length));
+    }
+  }
+  return null;
+}
+
+
+
+var max_results = 75; // 50 is not enough to search for map in the base libraries
+var shown_range = null;
+var last_search = null;
+
+function quick_search()
+{
+    perform_search(false);
+}
+
+function full_search()
+{
+    perform_search(true);
+}
+
+
+function perform_search(full)
+{
+    var text = document.getElementById("searchbox").value.toLowerCase();
+    if (text == last_search && !full) return;
+    last_search = text;
+    
+    var table = document.getElementById("indexlist");
+    var status = document.getElementById("searchmsg");
+    var children = table.firstChild.childNodes;
+    
+    // first figure out the first node with the prefix
+    var first = bisect(-1);
+    var last = (first == -1 ? -1 : bisect(1));
+
+    if (first == -1)
+    {
+        table.className = "";
+        status.innerHTML = "No results found, displaying all";
+    }
+    else if (first == 0 && last == children.length - 1)
+    {
+        table.className = "";
+        status.innerHTML = "";
+    }
+    else if (last - first >= max_results && !full)
+    {
+        table.className = "";
+        status.innerHTML = "More than " + max_results + ", press Search to display";
+    }
+    else
+    {
+        // decide what you need to clear/show
+        if (shown_range)
+            setclass(shown_range[0], shown_range[1], "indexrow");
+        setclass(first, last, "indexshow");
+        shown_range = [first, last];
+        table.className = "indexsearch";
+        status.innerHTML = "";
+    }
+
+    
+    function setclass(first, last, status)
+    {
+        for (var i = first; i <= last; i++)
+        {
+            children[i].className = status;
+        }
+    }
+    
+    
+    // do a binary search, treating 0 as ...
+    // return either -1 (no 0's found) or location of most far match
+    function bisect(dir)
+    {
+        var first = 0, finish = children.length - 1;
+        var mid, success = false;
+
+        while (finish - first > 3)
+        {
+            mid = Math.floor((finish + first) / 2);
+
+            var i = checkitem(mid);
+            if (i == 0) i = dir;
+            if (i == -1)
+                finish = mid;
+            else
+                first = mid;
+        }
+        var a = (dir == 1 ? first : finish);
+        var b = (dir == 1 ? finish : first);
+        for (var i = b; i != a - dir; i -= dir)
+        {
+            if (checkitem(i) == 0) return i;
+        }
+        return -1;
+    }    
+    
+    
+    // from an index, decide what the result is
+    // 0 = match, -1 is lower, 1 is higher
+    function checkitem(i)
+    {
+        var s = getitem(i).toLowerCase().substr(0, text.length);
+        if (s == text) return 0;
+        else return (s > text ? -1 : 1);
+    }
+    
+    
+    // from an index, get its string
+    // this abstracts over alternates
+    function getitem(i)
+    {
+        for ( ; i >= 0; i--)
+        {
+            var s = children[i].firstChild.firstChild.data;
+            if (s.indexOf(' ') == -1)
+                return s;
+        }
+        return ""; // should never be reached
+    }
+}
+
+function setSynopsis(filename) {
+    if (parent.window.synopsis) {
+        if (parent.window.synopsis.location.replace) {
+            // In Firefox this avoids adding the change to the history.
+            parent.window.synopsis.location.replace(filename);
+        } else {
+            parent.window.synopsis.location = filename;
+        }
+    }
+}
+
+function addMenuItem(html) {
+  var menu = document.getElementById("page-menu");
+  if (menu) {
+    var btn = menu.firstChild.cloneNode(false);
+    btn.innerHTML = html;
+    menu.appendChild(btn);
+  }
+}
+
+function adjustForFrames() {
+  var bodyCls;
+  
+  if (parent.location.href == window.location.href) {
+    // not in frames, so add Frames button
+    addMenuItem("<a href='#' onclick='reframe();return true;'>Frames</a>");
+    bodyCls = "no-frame";
+  }
+  else {
+    bodyCls = "in-frame";
+  }
+  addClass(document.body, bodyCls);
+}
+
+function reframe() {
+  setCookie("haddock-reframe", document.URL);
+  window.location = "frames.html";
+}
+
+function postReframe() {
+  var s = getCookie("haddock-reframe");
+  if (s) {
+    parent.window.main.location = s;
+    clearCookie("haddock-reframe");
+  }
+}
+
+function styles() {
+  var i, a, es = document.getElementsByTagName("link"), rs = [];
+  for (i = 0; a = es[i]; i++) {
+    if(a.rel.indexOf("style") != -1 && a.title) {
+      rs.push(a);
+    }
+  }
+  return rs;
+}
+
+function addStyleMenu() {
+  var as = styles();
+  var i, a, btns = "";
+  for(i=0; a = as[i]; i++) {
+    btns += "<li><a href='#' onclick=\"setActiveStyleSheet('"
+      + a.title + "'); return false;\">"
+      + a.title + "</a></li>"
+  }
+  if (as.length > 1) {
+    var h = "<div id='style-menu-holder'>"
+      + "<a href='#' onclick='styleMenu(); return false;'>Style &#9662;</a>"
+      + "<ul id='style-menu' class='hide'>" + btns + "</ul>"
+      + "</div>";
+    addMenuItem(h);
+  }
+}
+
+function setActiveStyleSheet(title) {
+  var as = styles();
+  var i, a, found;
+  for(i=0; a = as[i]; i++) {
+    a.disabled = true;
+          // need to do this always, some browsers are edge triggered
+    if(a.title == title) {
+      found = a;
+    }
+  }
+  if (found) {
+    found.disabled = false;
+    setCookie("haddock-style", title);
+  }
+  else {
+    as[0].disabled = false;
+    clearCookie("haddock-style");
+  }
+  styleMenu(false);
+}
+
+function resetStyle() {
+  var s = getCookie("haddock-style");
+  if (s) setActiveStyleSheet(s);
+}
+
+
+function styleMenu(show) {
+  var m = document.getElementById('style-menu');
+  if (m) toggleShow(m, show);
+}
+
+
+function pageLoad() {
+  addStyleMenu();
+  adjustForFrames();
+  resetStyle();
+  restoreCollapsed();
+}
+
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/hslogo-16.png b/docs/haddock/tensorflow-ops-0.1.0.0/hslogo-16.png
new file mode 100644
index 0000000000000000000000000000000000000000..0ff8579fbd897417b0d6dad6e920f8882138a7c0
GIT binary patch
literal 1684
zcmV;F25b3=P)<h;3K|Lk000e1NJLTq000*N000mO1^@s6{V(Jg0008+X+uL$Nkc;*
zP;zf(X>4Tx0C)j~RL^S@K@|QrZmG~B2wH0nvUrdpNm;9CMbtL^5n^i$+aI<?kYqDS
z)^0ZI?k2URh~(nYD&9ryp$Gqf9(wR(FG_`4MJV2@C}N=(E%D86Otz)!z|Ma2-Z$@k
zZ+71R4RX>n^?(HA4aZWV5ov6ELTdbo0FI&wK{O>*+w4vx20?>!`FrQsdJlnHR>OPy
zcd~b_n$otK2Za4V;76L-DzNVtaSB-y0*E}{p()372;bw_^6ZZ}PI-92wGS&j#91PI
zKs7DSe@(bk%_Y-7gGe}(^>I=@oY#w#*Bu9GZf3^F5WP>3rn}7Ut74&?PWBFvy`A)a
zPP5)V!Xd&78LdA?xQ(9mjMYElVd13a#D+Z_7&Y|xU=_C-srWU*6kiZcC!$nw*)9$7
zn6CX+@=AhmkT}X@VSsa5NKe;HZuq)~1$`#h6R+ZTR#D-3j}vF!)ZOnz+5)dI4jl{{
z44Mr{P!L4~VVJN`K!!XTF*LGrKO?IK8<Tr7btG!LbYeuYL3=jbJ-1P$-8}v%B5{;M
zwFr{@LH;VQ$xr2Z`O93e*jD$Ht(%&<^58qg<(at}9@o>z<8w`3e3jI8lUGNUta*C8
zn(P`s>{pjD=7Kek#B;Fw@hxAK%$F&Q6vg9J^Xf~4by_hu-=A!MJ3Znq&n~srbFGPs
zH&&aMXZ>nO`|hf|ljc?VPhR!${AbO?W8x_>CU%PFA&Hm8F7cAsOREdwU~R_;ot1_u
z(ruCYB-LPGn!NQdT|ZlRy+(fw^-+`=%+gee_kY4FWHg<*4sZI8+sFJD<oAl_pC|$^
zY~aY5x@}W&?+~G7rEYVs0vEs0eekw!YomTR`~+A$s}`+NHJ>270UUORdLHO0nA4V)
z%{fwsET5CQ>B?eK%uw4yQc~9?*JVo<vzb+5>2}ze(;aRcp*ceL#HUJSllrgm5wQKR
zQu+C;QrUh^8rFfA`ftFz{YAidi-`aL010qNS#tmY4c7nw4c7reD4Tcy00T@(L_t(I
z5sj2vNEA^R$7gqDc6T=2^@fUA2(c`MltuL5<|<Z1YDGjBEr^N`3VJpQ$`C4&9(*Z6
z=|*B$y;-1^F6vXRAd7+(;)77Qps2euv)AeWmj>KW>RWz$&YbU@|M|{$E*8Tu-Ux!w
z1Y*Dr&Ubfr&v-nZaaB{3ilRumrjPmk{sZvQEWlW+{o~IH|8)=s6c#X9S5s5d%<Agu
zgM$153Ttk++ZPA~ZqgqN1|^pNV-|?9-zo48n9+Yiy$*su5czgp*VFj6!1h&DRZR>J
z4@)&QH5|xQY-)^L1n<EM!01(Q6kM8}ojndgRD~2KVIa@>0pTRu0Lx9`08YTjTwn^6
z0;b1+aQ@)n;Em$q;=7BBi)v0zj&o^g>0Whp^_^5IbxIUP8C@y9;R?*Ouu}rmfxbU=
zwtWVNke-m!=`7bYEhWpcI5#)9qp`8E0lr6IQ)ARL3Ui}Af@grj8aN1=r>Cb+prlzO
zNfJs*N_tUm2ZL<f%*@E?H`Ak_J`Q#Qri<;K{{H@Bg@uKWS5{WE+}vCx91geD*4DNm
zJ?D^~dGz7lWtbWH`o<0O#|{R6h9M*GMpjnV?Y_RgatOb~1le#ikPKTNIn%<;=>%5*
zPmL2??da$TR904gL(VDAQ<BE%k@5ZA>-Fv_Dk}Pdw*4T(%*f4MKLRg=4ekMjhe2mW
zMFsBwg%ftWT}0kxRaIk1k7qJ8*#cKB;Ft{i`zVIs-Nqge;!!Ld7#O&Qqu7e0sJmP)
z$MW*>L$vSB&dxp@iA3U9fo)-7!Czlr{|o7<LYh`Fd@X2-b0fIa*Wdy$({Zt@t7`&L
zhoaGF!l5{1P1Qm;<GaSj#=Oa7atf1JZ8n=F9*@uC?b(X48O9kPoH~h;!o2=WM?ZLl
z$Lj;4K~)fAdW@^N%nzT>Hv{1oyg3xsu%gn@(b1>$;SM-ZaQ`HV=V0s;lr%d8bd;xY
zGw<i+<$17ay36gdrqk(Lxc1|{y}dSeM5OIV(FjhBg5#OM<!Z*8(@<7crlVw+B-8R1
zl{3J6w9fYQ^vojauaRgZ=~SE&CK!R3pCe!v3>Nvm3=Iu=tyXIgtJnf@A(2S@M140N
ew{UA~tMxaJq;$xaSSi*30000<MNUMnLSTa0p)&^n

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/index-frames.html b/docs/haddock/tensorflow-ops-0.1.0.0/index-frames.html
new file mode 100644
index 0000000..e1ed028
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/index-frames.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-ops-0.1.0.0: Friendly layer around TensorFlow bindings.</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-list"><p class="caption">Modules</p><ul><li class="module"><a href="TensorFlow-EmbeddingOps.html" target="main">TensorFlow.EmbeddingOps</a></li><li class="module"><a href="TensorFlow-Gradient.html" target="main">TensorFlow.Gradient</a></li><li class="module"><a href="TensorFlow-Ops.html" target="main">TensorFlow.Ops</a></li></ul></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/index.html b/docs/haddock/tensorflow-ops-0.1.0.0/index.html
new file mode 100644
index 0000000..ef676e3
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-ops-0.1.0.0: Friendly layer around TensorFlow bindings.</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-ops-0.1.0.0: Friendly layer around TensorFlow bindings.</p></div><div id="content"><div id="description"><h1>tensorflow-ops-0.1.0.0: Friendly layer around TensorFlow bindings.</h1><div class="doc"><p>Please see README.md</p></div></div><div id="module-list"><p class="caption">Modules</p><ul><li><span id="control.n.1" class="module collapser" onclick="toggleSection('n.1')">TensorFlow</span><ul id="section.n.1" class="show"><li><span class="module"><a href="TensorFlow-EmbeddingOps.html">TensorFlow.EmbeddingOps</a></span></li><li><span class="module"><a href="TensorFlow-Gradient.html">TensorFlow.Gradient</a></span></li><li><span class="module"><a href="TensorFlow-Ops.html">TensorFlow.Ops</a></span></li></ul></li></ul></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/mini_TensorFlow-EmbeddingOps.html b/docs/haddock/tensorflow-ops-0.1.0.0/mini_TensorFlow-EmbeddingOps.html
new file mode 100644
index 0000000..6bb59df
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/mini_TensorFlow-EmbeddingOps.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.EmbeddingOps</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.EmbeddingOps</p></div><div id="interface"><div class="top"><p class="src"><a href="TensorFlow-EmbeddingOps.html#v:embeddingLookup" target="main">embeddingLookup</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/mini_TensorFlow-Gradient.html b/docs/haddock/tensorflow-ops-0.1.0.0/mini_TensorFlow-Gradient.html
new file mode 100644
index 0000000..d309aff
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/mini_TensorFlow-Gradient.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Gradient</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.Gradient</p></div><div id="interface"><div class="top"><p class="src"><a href="TensorFlow-Gradient.html#v:gradients" target="main">gradients</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/mini_TensorFlow-Ops.html b/docs/haddock/tensorflow-ops-0.1.0.0/mini_TensorFlow-Ops.html
new file mode 100644
index 0000000..bc6c10a
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/mini_TensorFlow-Ops.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Ops</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.Ops</p></div><div id="interface"><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:add" target="main">add</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:abs" target="main">abs</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:addN" target="main">addN</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:argMax" target="main">argMax</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:assign" target="main">assign</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:broadcastGradientArgs" target="main">broadcastGradientArgs</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:cast" target="main">cast</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:concat" target="main">concat</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:constant" target="main">constant</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:expandDims" target="main">expandDims</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:initializedVariable" target="main">initializedVariable</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:zeroInitializedVariable" target="main">zeroInitializedVariable</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:fill" target="main">fill</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:matMul" target="main">matMul</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:matTranspose" target="main">matTranspose</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:mul" target="main">mul</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:neg" target="main">neg</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:pack" target="main">pack</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:placeholder" target="main">placeholder</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:range" target="main">range</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:reducedShape" target="main">reducedShape</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:relu" target="main">relu</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:reluGrad" target="main">reluGrad</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:reshape" target="main">reshape</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:restore" target="main">restore</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:restoreFromName" target="main">restoreFromName</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:save" target="main">save</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:scalar" target="main">scalar</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:shape" target="main">shape</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:sign" target="main">sign</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:size" target="main">size</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:softmax" target="main">softmax</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:softmaxCrossEntropyWithLogits" target="main">softmaxCrossEntropyWithLogits</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:sparseToDense" target="main">sparseToDense</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:sub" target="main">sub</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:sum" target="main">sum</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:topK" target="main">topK</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:transpose" target="main">transpose</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:truncatedNormal" target="main">truncatedNormal</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:variable" target="main">variable</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:vector" target="main">vector</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:zeros" target="main">zeros</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Ops.html#v:zerosLike" target="main">zerosLike</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/minus.gif b/docs/haddock/tensorflow-ops-0.1.0.0/minus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..1deac2fe1a42e35b994f1b855488f392c50f6a89
GIT binary patch
literal 56
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS53HJ0a;6C;`j@u!^I(6E|
Jh4LZ{)&P#)4$%Mr

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/ocean.css b/docs/haddock/tensorflow-ops-0.1.0.0/ocean.css
new file mode 100644
index 0000000..1110b40
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/ocean.css
@@ -0,0 +1,600 @@
+/* @group Fundamentals */
+
+* { margin: 0; padding: 0 }
+
+/* Is this portable? */
+html {
+  background-color: white;
+  width: 100%;
+  height: 100%;
+}
+
+body {
+  background: white;
+  color: black;
+  text-align: left;
+  min-height: 100%;
+  position: relative;
+}
+
+p {
+  margin: 0.8em 0;
+}
+
+ul, ol {
+  margin: 0.8em 0 0.8em 2em;
+}
+
+dl {
+  margin: 0.8em 0;
+}
+
+dt {
+  font-weight: bold;
+}
+dd {
+  margin-left: 2em;
+}
+
+a { text-decoration: none; }
+a[href]:link { color: rgb(196,69,29); }
+a[href]:visited { color: rgb(171,105,84); }
+a[href]:hover { text-decoration:underline; }
+
+/* @end */
+
+/* @group Fonts & Sizes */
+
+/* Basic technique & IE workarounds from YUI 3
+   For reasons, see:
+      http://yui.yahooapis.com/3.1.1/build/cssfonts/fonts.css
+ */
+
+body {
+	font:13px/1.4 sans-serif;
+	*font-size:small; /* for IE */
+	*font:x-small; /* for IE in quirks mode */
+}
+
+h1 { font-size: 146.5%; /* 19pt */ }
+h2 { font-size: 131%;   /* 17pt */ }
+h3 { font-size: 116%;   /* 15pt */ }
+h4 { font-size: 100%;   /* 13pt */ }
+h5 { font-size: 100%;   /* 13pt */ }
+
+select, input, button, textarea {
+	font:99% sans-serif;
+}
+
+table {
+	font-size:inherit;
+	font:100%;
+}
+
+pre, code, kbd, samp, tt, .src {
+	font-family:monospace;
+	*font-size:108%;
+	line-height: 124%;
+}
+
+.links, .link {
+  font-size: 85%; /* 11pt */
+}
+
+#module-header .caption {
+  font-size: 182%; /* 24pt */
+}
+
+.info  {
+  font-size: 85%; /* 11pt */
+}
+
+#table-of-contents, #synopsis  {
+  /* font-size: 85%; /* 11pt */
+}
+
+
+/* @end */
+
+/* @group Common */
+
+.caption, h1, h2, h3, h4, h5, h6 {
+  font-weight: bold;
+  color: rgb(78,98,114);
+  margin: 0.8em 0 0.4em;
+}
+
+* + h1, * + h2, * + h3, * + h4, * + h5, * + h6 {
+  margin-top: 2em;
+}
+
+h1 + h2, h2 + h3, h3 + h4, h4 + h5, h5 + h6 {
+  margin-top: inherit;
+}
+
+ul.links {
+  list-style: none;
+  text-align: left;
+  float: right;
+  display: inline-table;
+  margin: 0 0 0 1em;
+}
+
+ul.links li {
+  display: inline;
+  border-left: 1px solid #d5d5d5;
+  white-space: nowrap;
+  padding: 0;
+}
+
+ul.links li a {
+  padding: 0.2em 0.5em;
+}
+
+.hide { display: none; }
+.show { display: inherit; }
+.clear { clear: both; }
+
+.collapser {
+  background-image: url(minus.gif);
+  background-repeat: no-repeat;
+}
+.expander {
+  background-image: url(plus.gif);
+  background-repeat: no-repeat;
+}
+p.caption.collapser,
+p.caption.expander {
+  background-position: 0 0.4em;
+}
+.collapser, .expander {
+  padding-left: 14px;
+  margin-left: -14px;
+  cursor: pointer;
+}
+
+pre {
+  padding: 0.25em;
+  margin: 0.8em 0;
+  background: rgb(229,237,244);
+  overflow: auto;
+  border-bottom: 0.25em solid white;
+  /* white border adds some space below the box to compensate
+     for visual extra space that paragraphs have between baseline
+     and the bounding box */
+}
+
+.src {
+  background: #f0f0f0;
+  padding: 0.2em 0.5em;
+}
+
+.keyword { font-weight: normal; }
+.def { font-weight: bold; }
+
+
+/* @end */
+
+/* @group Page Structure */
+
+#content {
+  margin: 0 auto;
+  padding: 0 2em 6em;
+}
+
+#package-header {
+  background: rgb(41,56,69);
+  border-top: 5px solid rgb(78,98,114);
+  color: #ddd;
+  padding: 0.2em;
+  position: relative;
+  text-align: left;
+}
+
+#package-header .caption {
+  background: url(hslogo-16.png) no-repeat 0em;
+  color: white;
+  margin: 0 2em;
+  font-weight: normal;
+  font-style: normal;
+  padding-left: 2em;
+}
+
+#package-header a:link, #package-header a:visited { color: white; }
+#package-header a:hover { background: rgb(78,98,114); }
+
+#module-header .caption {
+  color: rgb(78,98,114);
+  font-weight: bold;
+  border-bottom: 1px solid #ddd;
+}
+
+table.info {
+  float: right;
+  padding: 0.5em 1em;
+  border: 1px solid #ddd;
+  color: rgb(78,98,114);
+  background-color: #fff;
+  max-width: 40%;
+  border-spacing: 0;
+  position: relative;
+  top: -0.5em;
+  margin: 0 0 0 2em;
+}
+
+.info th {
+	padding: 0 1em 0 0;
+}
+
+div#style-menu-holder {
+  position: relative;
+  z-index: 2;
+  display: inline;
+}
+
+#style-menu {
+  position: absolute;
+  z-index: 1;
+  overflow: visible;
+  background: #374c5e;
+  margin: 0;
+  text-align: center;
+  right: 0;
+  padding: 0;
+  top: 1.25em;
+}
+
+#style-menu li {
+	display: list-item;
+	border-style: none;
+	margin: 0;
+	padding: 0;
+	color: #000;
+	list-style-type: none;
+}
+
+#style-menu li + li {
+	border-top: 1px solid #919191;
+}
+
+#style-menu a {
+  width: 6em;
+  padding: 3px;
+  display: block;
+}
+
+#footer {
+  background: #ddd;
+  border-top: 1px solid #aaa;
+  padding: 0.5em 0;
+  color: #666;
+  text-align: center;
+  position: absolute;
+  bottom: 0;
+  width: 100%;
+  height: 3em;
+}
+
+/* @end */
+
+/* @group Front Matter */
+
+#table-of-contents {
+  float: right;
+  clear: right;
+  background: #faf9dc;
+  border: 1px solid #d8d7ad;
+  padding: 0.5em 1em;
+  max-width: 20em;
+  margin: 0.5em 0 1em 1em;
+}
+
+#table-of-contents .caption {
+  text-align: center;
+  margin: 0;
+}
+
+#table-of-contents ul {
+  list-style: none;
+  margin: 0;
+}
+
+#table-of-contents ul ul {
+  margin-left: 2em;
+}
+
+#description .caption {
+  display: none;
+}
+
+#synopsis {
+  display: none;
+}
+
+.no-frame #synopsis {
+  display: block;
+  position: fixed;
+  right: 0;
+  height: 80%;
+  top: 10%;
+  padding: 0;
+  max-width: 75%;
+}
+
+#synopsis .caption {
+  float: left;
+  width: 29px;
+  color: rgba(255,255,255,0);
+  height: 110px;
+  margin: 0;
+  font-size: 1px;
+  padding: 0;
+}
+
+#synopsis p.caption.collapser {
+  background: url(synopsis.png) no-repeat -64px -8px;
+}
+
+#synopsis p.caption.expander {
+  background: url(synopsis.png) no-repeat 0px -8px;
+}
+
+#synopsis ul {
+  height: 100%;
+  overflow: auto;
+  padding: 0.5em;
+  margin: 0;
+}
+
+#synopsis ul ul {
+  overflow: hidden;
+}
+
+#synopsis ul,
+#synopsis ul li.src {
+  background-color: #faf9dc;
+  white-space: nowrap;
+  list-style: none;
+  margin-left: 0;
+}
+
+/* @end */
+
+/* @group Main Content */
+
+#interface div.top { margin: 2em 0; }
+#interface h1 + div.top,
+#interface h2 + div.top,
+#interface h3 + div.top,
+#interface h4 + div.top,
+#interface h5 + div.top {
+ 	margin-top: 1em;
+}
+#interface p.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface td.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface span.fixity {
+  color: #919191;
+  border-left: 1px solid #919191;
+  padding: 0.2em 0.5em 0.2em 0.5em;
+  margin: 0 -1em 0 1em;
+}
+
+#interface span.rightedge {
+  border-left: 1px solid #919191;
+  padding: 0.2em 0 0.2em 0;
+  margin: 0 0 0 1em;
+}
+
+#interface table { border-spacing: 2px; }
+#interface td {
+  vertical-align: top;
+  padding-left: 0.5em;
+}
+#interface td.src {
+  white-space: nowrap;
+}
+#interface td.doc p {
+  margin: 0;
+}
+#interface td.doc p + p {
+  margin-top: 0.8em;
+}
+
+.clearfix:after {
+  clear: both;
+  content: " ";
+  display: block;
+  height: 0;
+  visibility: hidden;
+}
+
+.subs dl {
+  margin: 0;
+}
+
+.subs dt {
+  float: left;
+  clear: left;
+  display: block;
+  margin: 1px 0;
+}
+
+.subs dd {
+  float: right;
+  width: 90%;
+  display: block;
+  padding-left: 0.5em;
+  margin-bottom: 0.5em;
+}
+
+.subs dd.empty {
+  display: none;
+}
+
+.subs dd p {
+  margin: 0;
+}
+
+/* Render short-style data instances */
+.inst ul {
+  height: 100%;
+  padding: 0.5em;
+  margin: 0;
+}
+
+.inst, .inst li {
+  list-style: none;
+  margin-left: 1em;
+}
+
+/* Workaround for bug in Firefox (issue #384) */
+.inst-left {
+  float: left;
+}
+
+.top p.src {
+  border-top: 1px solid #ccc;
+}
+
+.subs, .doc {
+  /* use this selector for one level of indent */
+  padding-left: 2em;
+}
+
+.warning {
+  color: red;
+}
+
+.arguments {
+  margin-top: -0.4em;
+}
+.arguments .caption {
+  display: none;
+}
+
+.fields { padding-left: 1em; }
+
+.fields .caption { display: none; }
+
+.fields p { margin: 0 0; }
+
+/* this seems bulky to me
+.methods, .constructors {
+  background: #f8f8f8;
+  border: 1px solid #eee;
+}
+*/
+
+/* @end */
+
+/* @group Auxillary Pages */
+
+
+.extension-list {
+    list-style-type: none;
+    margin-left: 0;
+}
+
+#mini {
+  margin: 0 auto;
+  padding: 0 1em 1em;
+}
+
+#mini > * {
+  font-size: 93%; /* 12pt */
+}
+
+#mini #module-list .caption,
+#mini #module-header .caption {
+  font-size: 125%; /* 15pt */
+}
+
+#mini #interface h1,
+#mini #interface h2,
+#mini #interface h3,
+#mini #interface h4 {
+  font-size: 109%; /* 13pt */
+  margin: 1em 0 0;
+}
+
+#mini #interface .top,
+#mini #interface .src {
+  margin: 0;
+}
+
+#mini #module-list ul {
+  list-style: none;
+  margin: 0;
+}
+
+#alphabet ul {
+	list-style: none;
+	padding: 0;
+	margin: 0.5em 0 0;
+	text-align: center;
+}
+
+#alphabet li {
+	display: inline;
+	margin: 0 0.25em;
+}
+
+#alphabet a {
+	font-weight: bold;
+}
+
+#index .caption,
+#module-list .caption { font-size: 131%; /* 17pt */ }
+
+#index table {
+  margin-left: 2em;
+}
+
+#index .src {
+  font-weight: bold;
+}
+#index .alt {
+  font-size: 77%; /* 10pt */
+  font-style: italic;
+  padding-left: 2em;
+}
+
+#index td + td {
+  padding-left: 1em;
+}
+
+#module-list ul {
+  list-style: none;
+  margin: 0 0 0 2em;
+}
+
+#module-list li {
+  clear: right;
+}
+
+#module-list span.collapser,
+#module-list span.expander {
+  background-position: 0 0.3em;
+}
+
+#module-list .package {
+  float: right;
+}
+
+/* @end */
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/plus.gif b/docs/haddock/tensorflow-ops-0.1.0.0/plus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..2d15c14173d23f664b955cd24f51c82f5f09d91d
GIT binary patch
literal 59
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS5iS+a@;NG@rPSdH*>gbBX
M^XE!9f*2UA0nx1yDgXcg

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/src/TensorFlow-EmbeddingOps.html b/docs/haddock/tensorflow-ops-0.1.0.0/src/TensorFlow-EmbeddingOps.html
new file mode 100644
index 0000000..e97c085
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/src/TensorFlow-EmbeddingOps.html
@@ -0,0 +1,87 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/EmbeddingOps.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE ConstraintKinds #-}</span>
+<a name="line-16"></a><span class='hs-comment'>{-# LANGUAGE DataKinds #-}</span>
+<a name="line-17"></a><span class='hs-comment'>{-# LANGUAGE NoMonomorphismRestriction #-}</span>
+<a name="line-18"></a><span class='hs-comment'>{-# LANGUAGE OverloadedStrings #-}</span>
+<a name="line-19"></a><span class='hs-comment'>{-# LANGUAGE RankNTypes #-}</span>
+<a name="line-20"></a>
+<a name="line-21"></a><span class='hs-comment'>-- | Parallel lookups on the list of tensors.</span>
+<a name="line-22"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>EmbeddingOps</span> <span class='hs-keyword'>where</span>
+<a name="line-23"></a>
+<a name="line-24"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Monad</span> <span class='hs-layout'>(</span><span class='hs-varid'>zipWithM</span><span class='hs-layout'>)</span>
+<a name="line-25"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span> <span class='hs-layout'>(</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-26"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>List</span> <span class='hs-layout'>(</span><span class='hs-varid'>genericLength</span><span class='hs-layout'>)</span>
+<a name="line-27"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>Build</span><span class='hs-layout'>,</span> <span class='hs-varid'>colocateWith</span><span class='hs-layout'>,</span> <span class='hs-varid'>render</span><span class='hs-layout'>)</span>
+<a name="line-28"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Ops</span> <span class='hs-conid'>()</span>  <span class='hs-comment'>-- Num instance for Tensor</span>
+<a name="line-29"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span><span class='hs-layout'>,</span> <span class='hs-conid'>Value</span><span class='hs-layout'>)</span>
+<a name="line-30"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span> <span class='hs-layout'>(</span><span class='hs-conid'>OneOf</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span><span class='hs-layout'>)</span>
+<a name="line-31"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>GenOps</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>CoreOps</span>
+<a name="line-32"></a>
+<a name="line-33"></a><a name="embeddingLookup"></a><span class='hs-comment'>-- | Looks up `ids` in a list of embedding tensors.</span>
+<a name="line-34"></a><span class='hs-comment'>--</span>
+<a name="line-35"></a><span class='hs-comment'>-- This function is used to perform parallel lookups on the list of</span>
+<a name="line-36"></a><span class='hs-comment'>-- tensors in `params`.  It is a generalization of `TF.gather`, where</span>
+<a name="line-37"></a><span class='hs-comment'>-- `params` is interpreted as a partition of a larger embedding</span>
+<a name="line-38"></a><span class='hs-comment'>-- tensor.</span>
+<a name="line-39"></a><span class='hs-comment'>--</span>
+<a name="line-40"></a><span class='hs-comment'>-- The partition_strategy is "mod", we assign each id to partition</span>
+<a name="line-41"></a><span class='hs-comment'>-- `p = id % len(params)`. For instance,</span>
+<a name="line-42"></a><span class='hs-comment'>-- 13 ids are split across 5 partitions as:</span>
+<a name="line-43"></a><span class='hs-comment'>-- `[[0, 5, 10], [1, 6, 11], [2, 7, 12], [3, 8], [4, 9]]`</span>
+<a name="line-44"></a><span class='hs-comment'>--</span>
+<a name="line-45"></a><span class='hs-comment'>-- The results of the lookup are concatenated into a dense</span>
+<a name="line-46"></a><span class='hs-comment'>-- tensor. The returned tensor has shape `shape(ids) + shape(params)[1:]`.</span>
+<a name="line-47"></a><span class='hs-definition'>embeddingLookup</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varid'>b</span> <span class='hs-varid'>v</span> <span class='hs-varop'>.</span>
+<a name="line-48"></a>                   <span class='hs-layout'>(</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span>
+<a name="line-49"></a>                   <span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span> <span class='hs-conid'>Int32</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>b</span>
+<a name="line-50"></a>                   <span class='hs-layout'>,</span> <span class='hs-conid'>Num</span> <span class='hs-varid'>b</span>
+<a name="line-51"></a>                   <span class='hs-layout'>)</span>
+<a name="line-52"></a>                <span class='hs-keyglyph'>=&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span>
+<a name="line-53"></a>                <span class='hs-comment'>-- ^ A list of tensors which can be concatenated along</span>
+<a name="line-54"></a>                <span class='hs-comment'>-- dimension 0. Each `Tensor` must be appropriately</span>
+<a name="line-55"></a>                <span class='hs-comment'>-- sized for `mod` partition strategy.</span>
+<a name="line-56"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>b</span>
+<a name="line-57"></a>                <span class='hs-comment'>-- ^ A `Tensor` with type `int32` or `int64`</span>
+<a name="line-58"></a>                <span class='hs-comment'>-- containing the ids to be looked up in `params`.</span>
+<a name="line-59"></a>                <span class='hs-comment'>-- The ids are required to be flat on entry and have</span>
+<a name="line-60"></a>                <span class='hs-comment'>-- fewer than 2^31 entries.</span>
+<a name="line-61"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-62"></a>                <span class='hs-comment'>-- ^ A dense tensor with shape `shape(ids) + shape(params)[1:]`.</span>
+<a name="line-63"></a><span class='hs-definition'>embeddingLookup</span> <span class='hs-varid'>params</span> <span class='hs-varid'>ids</span> <span class='hs-keyglyph'>=</span>
+<a name="line-64"></a>    <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>dynamicStitch</span> <span class='hs-varid'>pindices</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>partitionedResult</span>
+<a name="line-65"></a>  <span class='hs-keyword'>where</span> <span class='hs-varid'>np</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>genericLength</span> <span class='hs-varid'>params</span>
+<a name="line-66"></a>        <span class='hs-varid'>pAssignments</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>cast</span> <span class='hs-layout'>(</span><span class='hs-varid'>ids</span> <span class='hs-varop'>`</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>mod</span><span class='hs-varop'>`</span> <span class='hs-varid'>np</span><span class='hs-layout'>)</span>
+<a name="line-67"></a>        <span class='hs-varid'>newIds</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>ids</span> <span class='hs-varop'>`</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>div</span><span class='hs-varop'>`</span> <span class='hs-varid'>np</span>
+<a name="line-68"></a>        <span class='hs-varid'>originalIndices</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>range</span> <span class='hs-num'>0</span> <span class='hs-layout'>(</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>size</span> <span class='hs-varid'>ids</span><span class='hs-layout'>)</span> <span class='hs-num'>1</span>
+<a name="line-69"></a>        <span class='hs-comment'>-- Partition list of ids based on assignments into np separate lists</span>
+<a name="line-70"></a>        <span class='hs-varid'>gatherIds</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>dynamicPartition</span> <span class='hs-varid'>np</span> <span class='hs-varid'>newIds</span> <span class='hs-varid'>pAssignments</span>
+<a name="line-71"></a>        <span class='hs-comment'>-- Similarly, partition the original indices.</span>
+<a name="line-72"></a>        <span class='hs-varid'>pindices</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>dynamicPartition</span> <span class='hs-varid'>np</span> <span class='hs-varid'>originalIndices</span> <span class='hs-varid'>pAssignments</span>
+<a name="line-73"></a>        <span class='hs-comment'>-- Do np separate lookups, finding embeddings for plist[p] in params[p]</span>
+<a name="line-74"></a>        <span class='hs-varid'>partitionedResult</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>zipWithM</span>
+<a name="line-75"></a>                            <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>p</span> <span class='hs-varid'>g</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>colocateWith</span> <span class='hs-varid'>p</span> <span class='hs-varop'>$</span> <span class='hs-varid'>render</span> <span class='hs-varop'>$</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>gather</span> <span class='hs-varid'>p</span> <span class='hs-varid'>g</span><span class='hs-layout'>)</span>
+<a name="line-76"></a>                            <span class='hs-varid'>params</span> <span class='hs-varid'>gatherIds</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/src/TensorFlow-Gradient.html b/docs/haddock/tensorflow-ops-0.1.0.0/src/TensorFlow-Gradient.html
new file mode 100644
index 0000000..9f35a43
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/src/TensorFlow-Gradient.html
@@ -0,0 +1,708 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/Gradient.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE ConstraintKinds #-}</span>
+<a name="line-16"></a><span class='hs-comment'>{-# LANGUAGE DataKinds #-}</span>
+<a name="line-17"></a><span class='hs-comment'>{-# LANGUAGE FlexibleContexts #-}</span>
+<a name="line-18"></a><span class='hs-comment'>{-# LANGUAGE OverloadedStrings #-}</span>
+<a name="line-19"></a><span class='hs-comment'>{-# LANGUAGE RankNTypes #-}</span>
+<a name="line-20"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables #-}</span>
+<a name="line-21"></a><span class='hs-comment'>{-# LANGUAGE TypeFamilies #-}</span>
+<a name="line-22"></a><span class='hs-comment'>{-# LANGUAGE ViewPatterns #-}</span>
+<a name="line-23"></a>
+<a name="line-24"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Gradient</span>
+<a name="line-25"></a>    <span class='hs-layout'>(</span> <span class='hs-varid'>gradients</span>
+<a name="line-26"></a>    <span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-27"></a>
+<a name="line-28"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Monad</span> <span class='hs-layout'>(</span><span class='hs-varid'>forM</span><span class='hs-layout'>,</span> <span class='hs-varid'>zipWithM</span><span class='hs-layout'>)</span>
+<a name="line-29"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Control</span><span class='hs-varop'>.</span><span class='hs-conid'>Monad</span><span class='hs-varop'>.</span><span class='hs-conid'>State</span><span class='hs-varop'>.</span><span class='hs-conid'>Strict</span> <span class='hs-layout'>(</span><span class='hs-conid'>State</span><span class='hs-layout'>,</span> <span class='hs-varid'>evalState</span><span class='hs-layout'>,</span> <span class='hs-varid'>gets</span><span class='hs-layout'>,</span> <span class='hs-varid'>modify</span><span class='hs-layout'>)</span>
+<a name="line-30"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-layout'>(</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>
+<a name="line-31"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-layout'>(</span><span class='hs-conid'>Complex</span><span class='hs-layout'>)</span>
+<a name="line-32"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-layout'>(</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-33"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span> <span class='hs-layout'>(</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-34"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>List</span> <span class='hs-layout'>(</span><span class='hs-varid'>foldl'</span><span class='hs-layout'>,</span> <span class='hs-varid'>sortBy</span><span class='hs-layout'>)</span>
+<a name="line-35"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Strict</span> <span class='hs-layout'>(</span><span class='hs-conid'>Map</span><span class='hs-layout'>)</span>
+<a name="line-36"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromMaybe</span><span class='hs-layout'>,</span> <span class='hs-varid'>maybeToList</span><span class='hs-layout'>,</span> <span class='hs-varid'>mapMaybe</span><span class='hs-layout'>)</span>
+<a name="line-37"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Ord</span> <span class='hs-layout'>(</span><span class='hs-varid'>comparing</span><span class='hs-layout'>)</span>
+<a name="line-38"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>TextFormat</span> <span class='hs-layout'>(</span><span class='hs-varid'>showMessage</span><span class='hs-layout'>)</span>
+<a name="line-39"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Set</span> <span class='hs-layout'>(</span><span class='hs-conid'>Set</span><span class='hs-layout'>)</span>
+<a name="line-40"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-layout'>(</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-41"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Tuple</span> <span class='hs-layout'>(</span><span class='hs-varid'>swap</span><span class='hs-layout'>)</span>
+<a name="line-42"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span> <span class='hs-layout'>(</span><span class='hs-conid'>Lens'</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>&amp;</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>^.</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>.~</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>%~</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-43"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>State</span><span class='hs-varop'>.</span><span class='hs-conid'>Strict</span> <span class='hs-layout'>(</span><span class='hs-varid'>uses</span><span class='hs-layout'>)</span>
+<a name="line-44"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Stock</span> <span class='hs-layout'>(</span><span class='hs-varid'>at</span><span class='hs-layout'>,</span> <span class='hs-varid'>intAt</span><span class='hs-layout'>)</span>
+<a name="line-45"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span> <span class='hs-layout'>(</span><span class='hs-varid'>lens</span><span class='hs-layout'>,</span> <span class='hs-varid'>iso</span><span class='hs-layout'>)</span>
+<a name="line-46"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Prelude</span> <span class='hs-varid'>hiding</span> <span class='hs-layout'>(</span><span class='hs-varid'>sum</span><span class='hs-layout'>)</span>
+<a name="line-47"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Printf</span> <span class='hs-layout'>(</span><span class='hs-varid'>printf</span><span class='hs-layout'>)</span>
+<a name="line-48"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Graph</span><span class='hs-varop'>.</span><span class='hs-conid'>Inductive</span><span class='hs-varop'>.</span><span class='hs-conid'>Basic</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>FGL</span>
+<a name="line-49"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Graph</span><span class='hs-varop'>.</span><span class='hs-conid'>Inductive</span><span class='hs-varop'>.</span><span class='hs-conid'>Graph</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>FGL</span>
+<a name="line-50"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Graph</span><span class='hs-varop'>.</span><span class='hs-conid'>Inductive</span><span class='hs-varop'>.</span><span class='hs-conid'>PatriciaTree</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>FGL</span>
+<a name="line-51"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Graph</span><span class='hs-varop'>.</span><span class='hs-conid'>Inductive</span><span class='hs-varop'>.</span><span class='hs-conid'>Query</span><span class='hs-varop'>.</span><span class='hs-conid'>DFS</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>FGL</span>
+<a name="line-52"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>IntMap</span><span class='hs-varop'>.</span><span class='hs-conid'>Strict</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>IntMap</span>
+<a name="line-53"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Strict</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Map</span>
+<a name="line-54"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Set</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Set</span>
+<a name="line-55"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Text</span>
+<a name="line-56"></a>
+<a name="line-57"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>GenOps</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>CoreOps</span>
+<a name="line-58"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Build</span>
+<a name="line-59"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>Build</span>
+<a name="line-60"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>render</span>
+<a name="line-61"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>renderNodeName</span>
+<a name="line-62"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>renderedNodeDefs</span>
+<a name="line-63"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>opDef</span>
+<a name="line-64"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>opAttr</span>
+<a name="line-65"></a>    <span class='hs-layout'>)</span>
+<a name="line-66"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>BuildOp</span>
+<a name="line-67"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Ops</span>
+<a name="line-68"></a>    <span class='hs-layout'>(</span> <span class='hs-varid'>addN</span>
+<a name="line-69"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>broadcastGradientArgs</span>
+<a name="line-70"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>expandDims</span>
+<a name="line-71"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>fill</span>
+<a name="line-72"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>matMul</span>
+<a name="line-73"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>reducedShape</span>
+<a name="line-74"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>reluGrad</span>
+<a name="line-75"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>reshape</span>
+<a name="line-76"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>scalar</span>
+<a name="line-77"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>shape</span>
+<a name="line-78"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>softmaxCrossEntropyWithLogits</span>
+<a name="line-79"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>sum</span>
+<a name="line-80"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>vector</span>
+<a name="line-81"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>zerosLike</span>
+<a name="line-82"></a>    <span class='hs-layout'>)</span>
+<a name="line-83"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Output</span>
+<a name="line-84"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>NodeName</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-85"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Op</span> <span class='hs-layout'>(</span><span class='hs-conid'>Rendered</span><span class='hs-layout'>)</span>
+<a name="line-86"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Output</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-87"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>OutputIx</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-88"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>outputIndex</span>
+<a name="line-89"></a>    <span class='hs-layout'>)</span>
+<a name="line-90"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span>
+<a name="line-91"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>Tensor</span><span class='hs-layout'>(</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span>
+<a name="line-92"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>TensorKind</span> <span class='hs-layout'>(</span><span class='hs-conid'>ValueKind</span><span class='hs-layout'>)</span>
+<a name="line-93"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Value</span>
+<a name="line-94"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>tensorOutput</span>
+<a name="line-95"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>tensorAttr</span>
+<a name="line-96"></a>    <span class='hs-layout'>)</span>
+<a name="line-97"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span> <span class='hs-layout'>(</span><span class='hs-conid'>OneOf</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span><span class='hs-layout'>,</span> <span class='hs-varid'>attrLens</span><span class='hs-layout'>)</span>
+<a name="line-98"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span>
+<a name="line-99"></a>    <span class='hs-layout'>(</span><span class='hs-conid'>NodeDef</span><span class='hs-layout'>,</span> <span class='hs-varid'>attr</span><span class='hs-layout'>,</span> <span class='hs-varid'>input</span><span class='hs-layout'>,</span> <span class='hs-varid'>op</span><span class='hs-layout'>,</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span>
+<a name="line-100"></a>
+<a name="line-101"></a><a name="GradientCompatible"></a><span class='hs-keyword'>type</span> <span class='hs-conid'>GradientCompatible</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=</span>
+<a name="line-102"></a>    <span class='hs-comment'>-- TODO(fmayle): MaxPoolGrad doesn't support Double for some reason.</span>
+<a name="line-103"></a>    <span class='hs-layout'>(</span><span class='hs-conid'>Num</span> <span class='hs-varid'>a</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span> <span class='hs-conid'>Float</span><span class='hs-layout'>,</span> <span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>,</span> <span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span> <span class='hs-keyglyph'>]</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-104"></a>
+<a name="line-105"></a><span class='hs-comment'>-- TODO(fmayle): Support control flow.</span>
+<a name="line-106"></a><span class='hs-comment'>-- TODO(fmayle): Support gate_gradients-like option to avoid race conditions.</span>
+<a name="line-107"></a><span class='hs-comment'>-- TODO(fmayle): Do we need to consider control inputs? See _PendingCount in</span>
+<a name="line-108"></a><span class='hs-comment'>-- tensorflow/python/ops/gradients.py.</span>
+<a name="line-109"></a><span class='hs-comment'>-- TODO(fmayle): Maybe store the gradient functions and numOutputs on the OpDef.</span>
+<a name="line-110"></a>
+<a name="line-111"></a>
+<a name="line-112"></a><a name="gradients"></a><span class='hs-comment'>-- | Gradient of @y@ w.r.t. each element of @xs@.</span>
+<a name="line-113"></a><span class='hs-definition'>gradients</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span> <span class='hs-conid'>Num</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-114"></a>                                <span class='hs-comment'>-- TODO(gnezdo): remove indirect constraint.</span>
+<a name="line-115"></a>                               <span class='hs-comment'>-- It's a wart inherited from Num instance.</span>
+<a name="line-116"></a>                              <span class='hs-layout'>,</span> <span class='hs-varid'>v1</span> <span class='hs-keyglyph'>~</span> <span class='hs-conid'>Value</span>
+<a name="line-117"></a>                              <span class='hs-layout'>,</span> <span class='hs-conid'>GradientCompatible</span> <span class='hs-varid'>a</span>
+<a name="line-118"></a>                              <span class='hs-layout'>)</span>
+<a name="line-119"></a>          <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>a</span>  <span class='hs-comment'>-- ^ The output of the graph.</span>
+<a name="line-120"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span>  <span class='hs-comment'>-- ^ Tensors for which gradients are computed.</span>
+<a name="line-121"></a>          <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span>
+<a name="line-122"></a><span class='hs-definition'>gradients</span> <span class='hs-varid'>y</span> <span class='hs-varid'>xs</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-123"></a>    <span class='hs-comment'>-- The gradients are computed using "reverse accumulation", similarly to</span>
+<a name="line-124"></a>    <span class='hs-comment'>-- what is described here:</span>
+<a name="line-125"></a>    <span class='hs-comment'>-- https://en.wikipedia.org/wiki/Automatic_differentiation#The_chain_rule.2C_forward_and_reverse_accumulation</span>
+<a name="line-126"></a>    <span class='hs-comment'>--</span>
+<a name="line-127"></a>    <span class='hs-comment'>-- The code is summarised as follows:</span>
+<a name="line-128"></a>    <span class='hs-comment'>--</span>
+<a name="line-129"></a>    <span class='hs-comment'>-- 1. Create an fgl graph of the relevant nodes (ops) and edges (tensors).</span>
+<a name="line-130"></a>    <span class='hs-comment'>-- 2. Initialize the gradient of y to 1 (∂y/∂y = 1) and the rest of tensor's</span>
+<a name="line-131"></a>    <span class='hs-comment'>--    gradients to nothing.</span>
+<a name="line-132"></a>    <span class='hs-comment'>-- 3. Process the nodes in reverse topological order (i.e. each node comes</span>
+<a name="line-133"></a>    <span class='hs-comment'>--    after all of its outputs so that the output gradients for a node have</span>
+<a name="line-134"></a>    <span class='hs-comment'>--    been completely calculated before it is processed):</span>
+<a name="line-135"></a>    <span class='hs-comment'>--      a. Record the gradient for each of the node's output tensors (∂y/∂w</span>
+<a name="line-136"></a>    <span class='hs-comment'>--         for each output tensor w).</span>
+<a name="line-137"></a>    <span class='hs-comment'>--      b. Calculate the gradient of y w.r.t. each of the node's input</span>
+<a name="line-138"></a>    <span class='hs-comment'>--         tensors using the gradients of the node's output tensors.</span>
+<a name="line-139"></a>    <span class='hs-comment'>--</span>
+<a name="line-140"></a>    <span class='hs-comment'>--         Written differently, for each output tensor w and input tensor v:</span>
+<a name="line-141"></a>    <span class='hs-comment'>--           ∂y/∂w = ...            (calculated in previous steps)</span>
+<a name="line-142"></a>    <span class='hs-comment'>--           ∂w/∂v = ...            (op specific)</span>
+<a name="line-143"></a>    <span class='hs-comment'>--           ∂y/∂v = ∂y/∂w * ∂w/∂v  (technically, if tensor v is an input</span>
+<a name="line-144"></a>    <span class='hs-comment'>--                                   to multiple nodes, then this is only</span>
+<a name="line-145"></a>    <span class='hs-comment'>--                                   part of ∂y/∂v)</span>
+<a name="line-146"></a>    <span class='hs-comment'>--</span>
+<a name="line-147"></a>    <span class='hs-comment'>-- 4. Lookup the recorded gradient for each x in xs.</span>
+<a name="line-148"></a>
+<a name="line-149"></a>    <span class='hs-varid'>yName</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>renderNodeName</span> <span class='hs-varid'>y</span>
+<a name="line-150"></a>    <span class='hs-comment'>-- TODO(fmayle): Move this into Build.hs and call it unsafeNodeDefFromName?</span>
+<a name="line-151"></a>    <span class='hs-varid'>nodeDefLookup</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>NodeName</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>NodeDef</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>uses</span> <span class='hs-varid'>renderedNodeDefs</span> <span class='hs-varop'>$</span>
+<a name="line-152"></a>        <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>f</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>fromMaybe</span> <span class='hs-layout'>(</span><span class='hs-varid'>error</span> <span class='hs-varop'>$</span> <span class='hs-str'>"no NodeDef found for "</span> <span class='hs-varop'>++</span> <span class='hs-varid'>show</span> <span class='hs-varid'>x</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>f</span> <span class='hs-varid'>x</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-153"></a>        <span class='hs-varop'>.</span> <span class='hs-varid'>flip</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>lookup</span>
+<a name="line-154"></a>    <span class='hs-keyword'>let</span> <span class='hs-layout'>(</span><span class='hs-varid'>gr</span><span class='hs-layout'>,</span> <span class='hs-varid'>nodeMap</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>createGraph</span> <span class='hs-varid'>yName</span> <span class='hs-varid'>nodeDefLookup</span>
+<a name="line-155"></a>    <span class='hs-comment'>-- Set gradient of y to one.</span>
+<a name="line-156"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>initPending</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-conid'>Node</span> <span class='hs-layout'>(</span><span class='hs-conid'>PendingGradients</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-157"></a>        <span class='hs-varid'>initPending</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>empty</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>at</span> <span class='hs-layout'>(</span><span class='hs-varid'>nodeMap</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.!</span> <span class='hs-varid'>yName</span><span class='hs-layout'>)</span>
+<a name="line-158"></a>                                <span class='hs-varop'>.</span> <span class='hs-varid'>nonEmpty</span>
+<a name="line-159"></a>                                <span class='hs-varop'>.</span> <span class='hs-varid'>outputIxAt</span> <span class='hs-layout'>(</span><span class='hs-varid'>y</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>tensorOutput</span> <span class='hs-varop'>.</span> <span class='hs-varid'>outputIndex</span><span class='hs-layout'>)</span>
+<a name="line-160"></a>                                <span class='hs-varop'>.</span> <span class='hs-varid'>nonEmpty</span>
+<a name="line-161"></a>                                <span class='hs-varop'>.~</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>fill</span> <span class='hs-layout'>(</span><span class='hs-varid'>shape</span> <span class='hs-varid'>y</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>scalar</span> <span class='hs-num'>1</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span>
+<a name="line-162"></a>    <span class='hs-comment'>-- Calculate the gradients of y w.r.t. each node in the graph.</span>
+<a name="line-163"></a>    <span class='hs-varid'>gradientMap</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>graphGrads</span> <span class='hs-varid'>gr</span> <span class='hs-varid'>initPending</span>
+<a name="line-164"></a>    <span class='hs-comment'>-- Lookup the gradients for each x.</span>
+<a name="line-165"></a>    <span class='hs-varid'>forM</span> <span class='hs-varid'>xs</span> <span class='hs-varop'>$</span> <span class='hs-keyglyph'>\</span><span class='hs-varid'>x</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyword'>do</span>
+<a name="line-166"></a>        <span class='hs-varid'>xName</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>renderNodeName</span> <span class='hs-varid'>x</span>
+<a name="line-167"></a>        <span class='hs-varid'>render</span> <span class='hs-varop'>$</span> <span class='hs-varid'>fromMaybe</span> <span class='hs-layout'>(</span><span class='hs-varid'>zerosLike</span> <span class='hs-varid'>x</span><span class='hs-layout'>)</span> <span class='hs-varop'>$</span> <span class='hs-keyword'>do</span>
+<a name="line-168"></a>            <span class='hs-varid'>n</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>nodeMap</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>at</span> <span class='hs-varid'>xName</span>
+<a name="line-169"></a>            <span class='hs-keyword'>let</span> <span class='hs-varid'>i</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>tensorOutput</span> <span class='hs-varop'>.</span> <span class='hs-varid'>outputIndex</span>
+<a name="line-170"></a>            <span class='hs-varid'>gradientMap</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>at</span> <span class='hs-varid'>n</span> <span class='hs-varop'>.</span> <span class='hs-varid'>nonEmpty</span> <span class='hs-varop'>.</span> <span class='hs-varid'>outputIxAt</span> <span class='hs-varid'>i</span>
+<a name="line-171"></a>
+<a name="line-172"></a><a name="outputIxAt"></a><span class='hs-definition'>outputIxAt</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OutputIx</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Lens'</span> <span class='hs-layout'>(</span><span class='hs-conid'>IntMap</span><span class='hs-varop'>.</span><span class='hs-conid'>IntMap</span> <span class='hs-varid'>v</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-conid'>Maybe</span> <span class='hs-varid'>v</span><span class='hs-layout'>)</span>
+<a name="line-173"></a><span class='hs-definition'>outputIxAt</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>intAt</span> <span class='hs-varop'>.</span> <span class='hs-varid'>unOutputIx</span>
+<a name="line-174"></a>
+<a name="line-175"></a><a name="PendingGradients"></a><span class='hs-comment'>-- | Incomplete gradients of a node's outputs.</span>
+<a name="line-176"></a><a name="PendingGradients"></a><span class='hs-comment'>--</span>
+<a name="line-177"></a><a name="PendingGradients"></a><span class='hs-comment'>-- The lists represent partial sums. The key is an OutputIx sans newtype.</span>
+<a name="line-178"></a><a name="PendingGradients"></a><span class='hs-keyword'>type</span> <span class='hs-conid'>PendingGradients</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>IntMap</span><span class='hs-varop'>.</span><span class='hs-conid'>IntMap</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span>
+<a name="line-179"></a>
+<a name="line-180"></a><a name="Gradients"></a><span class='hs-comment'>-- | Gradients of a node's outputs. The key is an OutputIx sans newtype.</span>
+<a name="line-181"></a><a name="Gradients"></a><span class='hs-keyword'>type</span> <span class='hs-conid'>Gradients</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>IntMap</span><span class='hs-varop'>.</span><span class='hs-conid'>IntMap</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-182"></a>
+<a name="line-183"></a><a name="Graph"></a><span class='hs-comment'>-- | Graph of TensorFlow operations.</span>
+<a name="line-184"></a><a name="Graph"></a><span class='hs-keyword'>type</span> <span class='hs-conid'>Graph</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-conid'>Gr</span> <span class='hs-conid'>NodeDef</span> <span class='hs-conid'>EdgeLabel</span>
+<a name="line-185"></a>
+<a name="line-186"></a><a name="EdgeLabel"></a><span class='hs-comment'>-- | Data associated with an edge.</span>
+<a name="line-187"></a><a name="EdgeLabel"></a><span class='hs-comment'>--</span>
+<a name="line-188"></a><a name="EdgeLabel"></a><span class='hs-comment'>-- Pair of</span>
+<a name="line-189"></a><a name="EdgeLabel"></a><span class='hs-comment'>--   1. Output index of a tensor from the source node.</span>
+<a name="line-190"></a><a name="EdgeLabel"></a><span class='hs-comment'>--   2. Input index that the tensor connects to on the destination node.</span>
+<a name="line-191"></a><a name="EdgeLabel"></a><span class='hs-keyword'>type</span> <span class='hs-conid'>EdgeLabel</span> <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>OutputIx</span><span class='hs-layout'>,</span> <span class='hs-conid'>OutputIx</span><span class='hs-layout'>)</span>
+<a name="line-192"></a>
+<a name="line-193"></a>
+<a name="line-194"></a><a name="GradientsState"></a><span class='hs-comment'>-- | State used for calculating gradients.</span>
+<a name="line-195"></a><a name="GradientsState"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>GradientsState</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>GradientsState</span>
+<a name="line-196"></a>                      <span class='hs-layout'>{</span> <span class='hs-sel'>_gradientsPending</span> <span class='hs-keyglyph'>::</span> <span class='hs-varop'>!</span><span class='hs-layout'>(</span><span class='hs-conid'>Map</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-conid'>Node</span> <span class='hs-layout'>(</span><span class='hs-conid'>PendingGradients</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-197"></a>                      <span class='hs-layout'>,</span> <span class='hs-sel'>_gradientsResult</span>  <span class='hs-keyglyph'>::</span> <span class='hs-varop'>!</span><span class='hs-layout'>(</span><span class='hs-conid'>Map</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-conid'>Node</span> <span class='hs-layout'>(</span><span class='hs-conid'>Gradients</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-198"></a>                      <span class='hs-layout'>}</span>
+<a name="line-199"></a>
+<a name="line-200"></a><a name="gradientsPending"></a><span class='hs-definition'>gradientsPending</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-layout'>(</span><span class='hs-conid'>GradientsState</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-conid'>Map</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-conid'>Node</span> <span class='hs-layout'>(</span><span class='hs-conid'>PendingGradients</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-201"></a><span class='hs-definition'>gradientsPending</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_gradientsPending</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span> <span class='hs-layout'>{</span> <span class='hs-sel'>_gradientsPending</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y</span> <span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-202"></a>
+<a name="line-203"></a><a name="gradientsResult"></a><span class='hs-definition'>gradientsResult</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Lens'</span> <span class='hs-layout'>(</span><span class='hs-conid'>GradientsState</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-conid'>Map</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-conid'>Node</span> <span class='hs-layout'>(</span><span class='hs-conid'>Gradients</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-204"></a><span class='hs-definition'>gradientsResult</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lens</span> <span class='hs-sel'>_gradientsResult</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span> <span class='hs-layout'>{</span> <span class='hs-sel'>_gradientsResult</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y</span> <span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-205"></a>
+<a name="line-206"></a>
+<a name="line-207"></a><a name="safeIndex"></a><span class='hs-comment'>-- TODO(fmayle): Use something like Data.List.Safe.</span>
+<a name="line-208"></a><span class='hs-comment'>-- | Safe version of (!!).</span>
+<a name="line-209"></a><span class='hs-definition'>safeIndex</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Int</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Maybe</span> <span class='hs-varid'>a</span>
+<a name="line-210"></a><span class='hs-keyword'>_</span>      <span class='hs-varop'>`safeIndex`</span> <span class='hs-varid'>n</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>n</span> <span class='hs-varop'>&lt;</span> <span class='hs-num'>0</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Nothing</span>
+<a name="line-211"></a><span class='hs-conid'>[]</span>     <span class='hs-varop'>`safeIndex`</span> <span class='hs-keyword'>_</span>         <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Nothing</span>
+<a name="line-212"></a><a name="safeIndex"></a><span class='hs-layout'>(</span><span class='hs-varid'>x</span><span class='hs-conop'>:</span><span class='hs-keyword'>_</span><span class='hs-layout'>)</span>  <span class='hs-varop'>`safeIndex`</span> <span class='hs-num'>0</span>         <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Just</span> <span class='hs-varid'>x</span>
+<a name="line-213"></a><a name="safeIndex"></a><span class='hs-layout'>(</span><span class='hs-keyword'>_</span><span class='hs-conop'>:</span><span class='hs-varid'>xs</span><span class='hs-layout'>)</span> <span class='hs-varop'>`safeIndex`</span> <span class='hs-varid'>n</span>         <span class='hs-keyglyph'>=</span> <span class='hs-varid'>xs</span> <span class='hs-varop'>`safeIndex`</span> <span class='hs-layout'>(</span><span class='hs-varid'>n</span><span class='hs-comment'>-</span><span class='hs-num'>1</span><span class='hs-layout'>)</span>
+<a name="line-214"></a>
+<a name="line-215"></a><a name="anon"></a><span class='hs-comment'>-- Copy of <a href="http://hackage.haskell.org/package/lens-3.9.0.2/docs/Control-Lens-Iso.html#v%3anon">http://hackage.haskell.org/package/lens-3.9.0.2/docs/Control-Lens-Iso.html#v%3anon</a></span>
+<a name="line-216"></a><span class='hs-definition'>anon</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Bool</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Lens'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Maybe</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-varid'>a</span>
+<a name="line-217"></a><span class='hs-definition'>anon</span> <span class='hs-varid'>a</span> <span class='hs-varid'>p</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>iso</span> <span class='hs-layout'>(</span><span class='hs-varid'>fromMaybe</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-varid'>go</span> <span class='hs-keyword'>where</span>
+<a name="line-218"></a>  <span class='hs-varid'>go</span> <span class='hs-varid'>b</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>p</span> <span class='hs-varid'>b</span>       <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Nothing</span>
+<a name="line-219"></a>       <span class='hs-keyglyph'>|</span> <span class='hs-varid'>otherwise</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Just</span> <span class='hs-varid'>b</span>
+<a name="line-220"></a>
+<a name="line-221"></a><a name="non"></a><span class='hs-definition'>non</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Eq</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Lens'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Maybe</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-varid'>a</span>
+<a name="line-222"></a><span class='hs-definition'>non</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>anon</span> <span class='hs-varid'>a</span> <span class='hs-layout'>(</span><span class='hs-varid'>a</span><span class='hs-varop'>==</span><span class='hs-layout'>)</span>
+<a name="line-223"></a>
+<a name="line-224"></a><a name="nonEmpty"></a><span class='hs-comment'>-- | Lens that defaults Nothing to mempty.</span>
+<a name="line-225"></a><span class='hs-definition'>nonEmpty</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>Monoid</span> <span class='hs-layout'>(</span><span class='hs-varid'>t</span> <span class='hs-varid'>v</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-conid'>Foldable</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Lens'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Maybe</span> <span class='hs-layout'>(</span><span class='hs-varid'>t</span> <span class='hs-varid'>v</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>t</span> <span class='hs-varid'>v</span><span class='hs-layout'>)</span>
+<a name="line-226"></a><span class='hs-definition'>nonEmpty</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>anon</span> <span class='hs-varid'>mempty</span> <span class='hs-varid'>null</span>
+<a name="line-227"></a>
+<a name="line-228"></a><a name="graphGrads"></a><span class='hs-comment'>-- | Calculate the gradients for every node in a graph.</span>
+<a name="line-229"></a><span class='hs-definition'>graphGrads</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span><span class='hs-varop'>.</span> <span class='hs-conid'>GradientCompatible</span> <span class='hs-varid'>a</span>
+<a name="line-230"></a>           <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Graph</span>
+<a name="line-231"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Map</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-conid'>Node</span> <span class='hs-layout'>(</span><span class='hs-conid'>PendingGradients</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-232"></a>           <span class='hs-comment'>-- ^ Initial gradients (usually just 1 for the node of interest).</span>
+<a name="line-233"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>Map</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-conid'>Node</span> <span class='hs-layout'>(</span><span class='hs-conid'>Gradients</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-234"></a><span class='hs-definition'>graphGrads</span> <span class='hs-varid'>gr</span> <span class='hs-varid'>initPending</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>pure</span> <span class='hs-layout'>(</span><span class='hs-varid'>foldl'</span> <span class='hs-varid'>go</span> <span class='hs-varid'>initState</span> <span class='hs-varid'>nodeOrder</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>gradientsResult</span><span class='hs-layout'>)</span>
+<a name="line-235"></a>  <span class='hs-keyword'>where</span>
+<a name="line-236"></a>    <span class='hs-varid'>initState</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>GradientsState</span> <span class='hs-varid'>initPending</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>empty</span>
+<a name="line-237"></a>    <span class='hs-comment'>-- Reverse topological sort.</span>
+<a name="line-238"></a>    <span class='hs-comment'>-- TODO(fmayle): Filter out nodes that are not successors of any x in xs to</span>
+<a name="line-239"></a>    <span class='hs-comment'>-- avoid calculating gradients that won't be used.</span>
+<a name="line-240"></a>    <span class='hs-varid'>nodeOrder</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-varid'>topsort</span> <span class='hs-varop'>$</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-varid'>grev</span> <span class='hs-varid'>gr</span>
+<a name="line-241"></a>    <span class='hs-varid'>go</span> <span class='hs-varid'>state</span> <span class='hs-varid'>node</span> <span class='hs-keyglyph'>=</span>
+<a name="line-242"></a>        <span class='hs-comment'>-- Aggregate the accumulated gradients for this node.</span>
+<a name="line-243"></a>        <span class='hs-keyword'>let</span> <span class='hs-varid'>outputGrads</span> <span class='hs-keyglyph'>=</span>
+<a name="line-244"></a>                <span class='hs-varid'>sumPendingGradient</span> <span class='hs-layout'>(</span><span class='hs-varid'>state</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>gradientsPending</span> <span class='hs-varop'>.</span> <span class='hs-varid'>at</span> <span class='hs-varid'>node</span> <span class='hs-varop'>.</span> <span class='hs-varid'>nonEmpty</span><span class='hs-layout'>)</span>
+<a name="line-245"></a>        <span class='hs-keyword'>in</span> <span class='hs-keyword'>if</span> <span class='hs-varid'>null</span> <span class='hs-varid'>outputGrads</span>
+<a name="line-246"></a>           <span class='hs-keyword'>then</span> <span class='hs-varid'>state</span>
+<a name="line-247"></a>           <span class='hs-keyword'>else</span>
+<a name="line-248"></a>              <span class='hs-comment'>-- Calculate the gradients for each of the node's inputs.</span>
+<a name="line-249"></a>              <span class='hs-keyword'>let</span> <span class='hs-varid'>nextState</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>state</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>gradientsResult</span> <span class='hs-varop'>%~</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>insert</span> <span class='hs-varid'>node</span> <span class='hs-varid'>outputGrads</span>
+<a name="line-250"></a>                  <span class='hs-varid'>ctx</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-varid'>context</span> <span class='hs-varid'>gr</span> <span class='hs-varid'>node</span>
+<a name="line-251"></a>              <span class='hs-keyword'>in</span> <span class='hs-varid'>updatePendingGradients</span>
+<a name="line-252"></a>                 <span class='hs-varid'>ctx</span>
+<a name="line-253"></a>                 <span class='hs-layout'>(</span><span class='hs-varid'>calculateInputGrads</span> <span class='hs-varid'>ctx</span> <span class='hs-varid'>outputGrads</span> <span class='hs-varid'>gr</span><span class='hs-layout'>)</span>
+<a name="line-254"></a>                 <span class='hs-varid'>nextState</span>
+<a name="line-255"></a>
+<a name="line-256"></a><a name="sumPendingGradient"></a><span class='hs-comment'>-- | Reduce accumulated gradients for each output to one Tensor.</span>
+<a name="line-257"></a><span class='hs-definition'>sumPendingGradient</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>GradientCompatible</span> <span class='hs-varid'>a</span>
+<a name="line-258"></a>                   <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>PendingGradients</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Gradients</span> <span class='hs-varid'>a</span>
+<a name="line-259"></a><span class='hs-definition'>sumPendingGradient</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>IntMap</span><span class='hs-varop'>.</span><span class='hs-varid'>mapMaybe</span> <span class='hs-varid'>f</span>
+<a name="line-260"></a>  <span class='hs-keyword'>where</span>
+<a name="line-261"></a>    <span class='hs-varid'>f</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Nothing</span>
+<a name="line-262"></a>    <span class='hs-varid'>f</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>x</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Just</span> <span class='hs-varid'>x</span>
+<a name="line-263"></a>    <span class='hs-varid'>f</span> <span class='hs-varid'>xs</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Just</span> <span class='hs-layout'>(</span><span class='hs-varid'>addN</span> <span class='hs-varid'>xs</span><span class='hs-layout'>)</span>
+<a name="line-264"></a>
+<a name="line-265"></a>
+<a name="line-266"></a><a name="calculateInputGrads"></a><span class='hs-comment'>-- | Calculate the gradients of a node's input tensors.</span>
+<a name="line-267"></a><span class='hs-comment'>--</span>
+<a name="line-268"></a><span class='hs-comment'>-- This is mostly just a wrapper around opGrad.</span>
+<a name="line-269"></a><span class='hs-definition'>calculateInputGrads</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span><span class='hs-varop'>.</span> <span class='hs-conid'>GradientCompatible</span> <span class='hs-varid'>a</span>
+<a name="line-270"></a>                    <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-conid'>Context</span> <span class='hs-conid'>NodeDef</span> <span class='hs-conid'>EdgeLabel</span>
+<a name="line-271"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Gradients</span> <span class='hs-varid'>a</span>  <span class='hs-comment'>-- ^ Output gradients of the node.</span>
+<a name="line-272"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Graph</span>
+<a name="line-273"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Maybe</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span>
+<a name="line-274"></a><span class='hs-definition'>calculateInputGrads</span> <span class='hs-layout'>(</span><span class='hs-varid'>inputEdges</span><span class='hs-layout'>,</span> <span class='hs-keyword'>_</span><span class='hs-layout'>,</span> <span class='hs-varid'>nodeDef</span><span class='hs-layout'>,</span> <span class='hs-keyword'>_</span><span class='hs-layout'>)</span> <span class='hs-varid'>outputGrads</span> <span class='hs-varid'>gr</span> <span class='hs-keyglyph'>=</span>
+<a name="line-275"></a>    <span class='hs-varid'>opGrad</span> <span class='hs-layout'>(</span><span class='hs-varid'>nodeDef</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>op</span><span class='hs-layout'>)</span> <span class='hs-varid'>nodeDef</span> <span class='hs-varid'>inputTensors</span> <span class='hs-varid'>fullOutGrads</span>
+<a name="line-276"></a>  <span class='hs-keyword'>where</span>
+<a name="line-277"></a>    <span class='hs-varid'>fullOutGrads</span> <span class='hs-keyglyph'>=</span>
+<a name="line-278"></a>        <span class='hs-varid'>fullOutputGrads</span> <span class='hs-layout'>(</span><span class='hs-varid'>numOutputs</span> <span class='hs-varid'>nodeDef</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-conid'>Rendered</span> <span class='hs-varid'>nodeDef</span><span class='hs-layout'>)</span> <span class='hs-varid'>outputGrads</span>
+<a name="line-279"></a>    <span class='hs-comment'>-- Create a tensor from an edge (technically an Output, but it seems less</span>
+<a name="line-280"></a>    <span class='hs-comment'>-- confusing to refer to it as a tensor here).</span>
+<a name="line-281"></a>    <span class='hs-varid'>edgeToTensor</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>EdgeLabel</span><span class='hs-layout'>,</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-conid'>Node</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Output</span>
+<a name="line-282"></a>    <span class='hs-varid'>edgeToTensor</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varid'>i</span><span class='hs-layout'>,</span> <span class='hs-keyword'>_</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-varid'>n</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span>
+<a name="line-283"></a>        <span class='hs-keyword'>case</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-varid'>lab</span> <span class='hs-varid'>gr</span> <span class='hs-varid'>n</span> <span class='hs-keyword'>of</span>
+<a name="line-284"></a>            <span class='hs-conid'>Just</span> <span class='hs-varid'>edgeNodeDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Output</span> <span class='hs-varid'>i</span> <span class='hs-layout'>(</span><span class='hs-conid'>Rendered</span> <span class='hs-varid'>edgeNodeDef</span><span class='hs-layout'>)</span>
+<a name="line-285"></a>            <span class='hs-conid'>Nothing</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>error</span> <span class='hs-varop'>$</span> <span class='hs-str'>"calculateInputGrads: missing input node for "</span>
+<a name="line-286"></a>                               <span class='hs-varop'>++</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>unpack</span> <span class='hs-layout'>(</span><span class='hs-varid'>nodeDef</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span>
+<a name="line-287"></a>    <span class='hs-comment'>-- Input tensors, sorted by input index.</span>
+<a name="line-288"></a>    <span class='hs-varid'>inputTensors</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>map</span> <span class='hs-varid'>edgeToTensor</span> <span class='hs-varop'>$</span> <span class='hs-varid'>sortBy</span> <span class='hs-layout'>(</span><span class='hs-varid'>comparing</span> <span class='hs-layout'>(</span><span class='hs-varid'>snd</span> <span class='hs-varop'>.</span> <span class='hs-varid'>fst</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-varid'>inputEdges</span>
+<a name="line-289"></a>
+<a name="line-290"></a><a name="fullOutputGrads"></a><span class='hs-comment'>-- | Convert a Map of gradients to a list, with zeros for missing outputs.</span>
+<a name="line-291"></a><span class='hs-definition'>fullOutputGrads</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span><span class='hs-layout'>,</span> <span class='hs-conid'>Num</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-292"></a>                <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>OutputIx</span>  <span class='hs-comment'>-- ^ Number of outputs.</span>
+<a name="line-293"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Op</span>
+<a name="line-294"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Gradients</span> <span class='hs-varid'>a</span>
+<a name="line-295"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span>
+<a name="line-296"></a><span class='hs-definition'>fullOutputGrads</span> <span class='hs-varid'>n</span> <span class='hs-varid'>o</span> <span class='hs-varid'>gs</span> <span class='hs-keyglyph'>=</span>
+<a name="line-297"></a>    <span class='hs-varid'>map</span> <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span><span class='hs-varid'>i</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>fromMaybe</span> <span class='hs-layout'>(</span><span class='hs-varid'>zero</span> <span class='hs-varid'>i</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>gs</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>outputIxAt</span> <span class='hs-varid'>i</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>[</span><span class='hs-num'>0</span><span class='hs-keyglyph'>..</span><span class='hs-varid'>n</span><span class='hs-comment'>-</span><span class='hs-num'>1</span><span class='hs-keyglyph'>]</span>
+<a name="line-298"></a>  <span class='hs-keyword'>where</span>
+<a name="line-299"></a>    <span class='hs-comment'>-- A tensor of zeros with the same shape as the i'th output.</span>
+<a name="line-300"></a>    <span class='hs-varid'>zero</span> <span class='hs-varid'>i</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>zerosLike</span> <span class='hs-varop'>$</span> <span class='hs-varid'>toT</span> <span class='hs-layout'>(</span><span class='hs-conid'>Output</span> <span class='hs-varid'>i</span> <span class='hs-varid'>o</span><span class='hs-layout'>)</span>
+<a name="line-301"></a>
+<a name="line-302"></a>
+<a name="line-303"></a><a name="updatePendingGradients"></a><span class='hs-comment'>-- | Update the pending gradients of a node's inputs.</span>
+<a name="line-304"></a><span class='hs-definition'>updatePendingGradients</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span><span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span><span class='hs-layout'>,</span> <span class='hs-conid'>Num</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-305"></a>                       <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-conid'>Context</span> <span class='hs-conid'>NodeDef</span> <span class='hs-conid'>EdgeLabel</span>
+<a name="line-306"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Maybe</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span>
+<a name="line-307"></a>                       <span class='hs-comment'>-- ^ Gradient of each input tensor.</span>
+<a name="line-308"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>GradientsState</span> <span class='hs-varid'>a</span>
+<a name="line-309"></a>                       <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>GradientsState</span> <span class='hs-varid'>a</span>
+<a name="line-310"></a><span class='hs-definition'>updatePendingGradients</span> <span class='hs-layout'>(</span><span class='hs-varid'>inputEdges</span><span class='hs-layout'>,</span> <span class='hs-keyword'>_</span><span class='hs-layout'>,</span> <span class='hs-varid'>nodeDef</span><span class='hs-layout'>,</span> <span class='hs-keyword'>_</span><span class='hs-layout'>)</span> <span class='hs-varid'>inputGrads</span> <span class='hs-varid'>initState</span> <span class='hs-keyglyph'>=</span>
+<a name="line-311"></a>    <span class='hs-varid'>foldl'</span> <span class='hs-varid'>go</span> <span class='hs-varid'>initState</span> <span class='hs-varid'>inputEdges</span>
+<a name="line-312"></a>  <span class='hs-keyword'>where</span>
+<a name="line-313"></a>    <span class='hs-varid'>go</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>GradientsState</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>EdgeLabel</span><span class='hs-layout'>,</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-conid'>Node</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>GradientsState</span> <span class='hs-varid'>a</span>
+<a name="line-314"></a>    <span class='hs-varid'>go</span> <span class='hs-varid'>state</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varid'>outIndex</span><span class='hs-layout'>,</span> <span class='hs-conid'>OutputIx</span> <span class='hs-varid'>inIndex</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-varid'>node</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span>
+<a name="line-315"></a>        <span class='hs-keyword'>case</span> <span class='hs-varid'>maybeGradient</span> <span class='hs-keyword'>of</span>
+<a name="line-316"></a>            <span class='hs-conid'>Nothing</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>state</span>
+<a name="line-317"></a>            <span class='hs-conid'>Just</span> <span class='hs-varid'>g</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-318"></a>                <span class='hs-comment'>-- Add to the list of pending gradients for this tensor.</span>
+<a name="line-319"></a>                <span class='hs-varid'>state</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>gradientsPending</span>
+<a name="line-320"></a>                      <span class='hs-varop'>.</span> <span class='hs-varid'>at</span> <span class='hs-varid'>node</span>
+<a name="line-321"></a>                      <span class='hs-varop'>.</span> <span class='hs-varid'>nonEmpty</span>
+<a name="line-322"></a>                      <span class='hs-varop'>.</span> <span class='hs-varid'>outputIxAt</span> <span class='hs-varid'>outIndex</span>
+<a name="line-323"></a>                      <span class='hs-varop'>.</span> <span class='hs-varid'>nonEmpty</span>
+<a name="line-324"></a>                      <span class='hs-varop'>%~</span> <span class='hs-layout'>(</span><span class='hs-varid'>g</span><span class='hs-conop'>:</span><span class='hs-layout'>)</span>
+<a name="line-325"></a>      <span class='hs-keyword'>where</span>
+<a name="line-326"></a>        <span class='hs-varid'>badSizeErr</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-varop'>$</span> <span class='hs-varid'>printf</span> <span class='hs-str'>"updatePendingGradients: bad input index \
+<a name="line-327"></a>                                    \%d for inputGrads of length %d in %s"</span>
+<a name="line-328"></a>                                    <span class='hs-varid'>inIndex</span> <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>inputGrads</span><span class='hs-layout'>)</span>
+<a name="line-329"></a>                                    <span class='hs-layout'>(</span><span class='hs-varid'>show</span> <span class='hs-layout'>(</span><span class='hs-varid'>nodeDef</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-330"></a>        <span class='hs-varid'>maybeGradient</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>fromMaybe</span> <span class='hs-varid'>badSizeErr</span> <span class='hs-layout'>(</span><span class='hs-varid'>safeIndex</span> <span class='hs-varid'>inputGrads</span> <span class='hs-varid'>inIndex</span><span class='hs-layout'>)</span>
+<a name="line-331"></a>
+<a name="line-332"></a>
+<a name="line-333"></a><a name="createGraph"></a><span class='hs-comment'>-- | Create a graph that includes a node and its transitive dependencies.</span>
+<a name="line-334"></a><span class='hs-definition'>createGraph</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>NodeName</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>NodeName</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>NodeDef</span><span class='hs-layout'>)</span>
+<a name="line-335"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-layout'>(</span><span class='hs-conid'>Graph</span><span class='hs-layout'>,</span> <span class='hs-conid'>Map</span> <span class='hs-conid'>NodeName</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-conid'>Node</span><span class='hs-layout'>)</span>
+<a name="line-336"></a><span class='hs-definition'>createGraph</span> <span class='hs-varid'>nodeName</span> <span class='hs-varid'>nodeDefLookup</span> <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-varid'>nmap</span> <span class='hs-varid'>nodeDefLookup</span> <span class='hs-varid'>graph</span><span class='hs-layout'>,</span> <span class='hs-varid'>nodeMap</span><span class='hs-layout'>)</span>
+<a name="line-337"></a>  <span class='hs-keyword'>where</span>
+<a name="line-338"></a>    <span class='hs-comment'>-- Parse a tensor name.</span>
+<a name="line-339"></a>    <span class='hs-varid'>parseTensorName</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Maybe</span> <span class='hs-layout'>(</span><span class='hs-conid'>NodeName</span><span class='hs-layout'>,</span> <span class='hs-conid'>OutputIx</span><span class='hs-layout'>)</span>
+<a name="line-340"></a>    <span class='hs-varid'>parseTensorName</span> <span class='hs-varid'>n</span>
+<a name="line-341"></a>        <span class='hs-keyglyph'>|</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>null</span> <span class='hs-varid'>n</span>        <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-str'>"parseTensorName: empty name"</span>
+<a name="line-342"></a>        <span class='hs-keyglyph'>|</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>head</span> <span class='hs-varid'>n</span> <span class='hs-varop'>==</span> <span class='hs-chr'>'^'</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Nothing</span>  <span class='hs-comment'>-- Control edge</span>
+<a name="line-343"></a>        <span class='hs-keyglyph'>|</span> <span class='hs-varid'>otherwise</span>          <span class='hs-keyglyph'>=</span>
+<a name="line-344"></a>            <span class='hs-keyword'>let</span> <span class='hs-layout'>(</span><span class='hs-varid'>nm</span><span class='hs-layout'>,</span> <span class='hs-varid'>indexStr</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>breakOn</span> <span class='hs-str'>":"</span> <span class='hs-varid'>n</span>
+<a name="line-345"></a>                <span class='hs-varid'>index</span> <span class='hs-keyglyph'>|</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>null</span> <span class='hs-varid'>indexStr</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>0</span>
+<a name="line-346"></a>                      <span class='hs-keyglyph'>|</span> <span class='hs-varid'>otherwise</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>read</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>unpack</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-varid'>tail</span> <span class='hs-varid'>indexStr</span>
+<a name="line-347"></a>            <span class='hs-keyword'>in</span> <span class='hs-conid'>Just</span> <span class='hs-layout'>(</span><span class='hs-conid'>NodeName</span> <span class='hs-varid'>nm</span><span class='hs-layout'>,</span> <span class='hs-conid'>OutputIx</span> <span class='hs-varid'>index</span><span class='hs-layout'>)</span>
+<a name="line-348"></a>
+<a name="line-349"></a>    <span class='hs-comment'>-- Build a map from node name to outward edges.</span>
+<a name="line-350"></a>    <span class='hs-comment'>--</span>
+<a name="line-351"></a>    <span class='hs-comment'>-- The state is the set of visited nodes.</span>
+<a name="line-352"></a>    <span class='hs-varid'>collect</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Maybe</span> <span class='hs-layout'>(</span><span class='hs-conid'>NodeName</span><span class='hs-layout'>,</span> <span class='hs-conid'>OutputIx</span><span class='hs-layout'>,</span> <span class='hs-conid'>OutputIx</span><span class='hs-layout'>)</span>
+<a name="line-353"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>NodeName</span>
+<a name="line-354"></a>            <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>State</span> <span class='hs-layout'>(</span><span class='hs-conid'>Set</span> <span class='hs-conid'>NodeName</span><span class='hs-layout'>)</span>
+<a name="line-355"></a>                     <span class='hs-layout'>(</span><span class='hs-conid'>Map</span> <span class='hs-conid'>NodeName</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>NodeName</span><span class='hs-layout'>,</span> <span class='hs-conid'>OutputIx</span><span class='hs-layout'>,</span> <span class='hs-conid'>OutputIx</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-356"></a>    <span class='hs-varid'>collect</span> <span class='hs-varid'>outgoingEdge</span> <span class='hs-varid'>nm</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-357"></a>        <span class='hs-keyword'>let</span> <span class='hs-varid'>nextLookup</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>singleton</span> <span class='hs-varid'>nm</span> <span class='hs-layout'>(</span><span class='hs-varid'>maybeToList</span> <span class='hs-varid'>outgoingEdge</span><span class='hs-layout'>)</span>
+<a name="line-358"></a>        <span class='hs-varid'>seen</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>gets</span> <span class='hs-layout'>(</span><span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>member</span> <span class='hs-varid'>nm</span><span class='hs-layout'>)</span>
+<a name="line-359"></a>        <span class='hs-varid'>modify</span> <span class='hs-layout'>(</span><span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>insert</span> <span class='hs-varid'>nm</span><span class='hs-layout'>)</span>
+<a name="line-360"></a>        <span class='hs-keyword'>if</span> <span class='hs-varid'>seen</span>
+<a name="line-361"></a>            <span class='hs-keyword'>then</span> <span class='hs-varid'>pure</span> <span class='hs-varid'>nextLookup</span>
+<a name="line-362"></a>            <span class='hs-keyword'>else</span> <span class='hs-keyword'>do</span>
+<a name="line-363"></a>                <span class='hs-keyword'>let</span> <span class='hs-varid'>inputs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>nodeDefLookup</span> <span class='hs-varid'>nm</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>input</span>
+<a name="line-364"></a>                    <span class='hs-varid'>recurse</span> <span class='hs-varid'>inIndex</span> <span class='hs-layout'>(</span><span class='hs-varid'>parentName</span><span class='hs-layout'>,</span> <span class='hs-varid'>outIndex</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span>
+<a name="line-365"></a>                        <span class='hs-varid'>collect</span> <span class='hs-layout'>(</span><span class='hs-conid'>Just</span> <span class='hs-layout'>(</span><span class='hs-varid'>nm</span><span class='hs-layout'>,</span> <span class='hs-varid'>outIndex</span><span class='hs-layout'>,</span> <span class='hs-varid'>inIndex</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-varid'>parentName</span>
+<a name="line-366"></a>                <span class='hs-varid'>subEdgeLookups</span> <span class='hs-keyglyph'>&lt;-</span>
+<a name="line-367"></a>                    <span class='hs-varid'>zipWithM</span> <span class='hs-varid'>recurse</span> <span class='hs-keyglyph'>[</span><span class='hs-num'>0</span><span class='hs-keyglyph'>..</span><span class='hs-keyglyph'>]</span> <span class='hs-varop'>$</span> <span class='hs-varid'>mapMaybe</span> <span class='hs-varid'>parseTensorName</span> <span class='hs-varid'>inputs</span>
+<a name="line-368"></a>                <span class='hs-varid'>pure</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>unionsWith</span> <span class='hs-layout'>(</span><span class='hs-varop'>++</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>nextLookup</span><span class='hs-conop'>:</span><span class='hs-varid'>subEdgeLookups</span><span class='hs-layout'>)</span>
+<a name="line-369"></a>
+<a name="line-370"></a>    <span class='hs-varid'>edgeLookup</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>evalState</span> <span class='hs-layout'>(</span><span class='hs-varid'>collect</span> <span class='hs-conid'>Nothing</span> <span class='hs-varid'>nodeName</span><span class='hs-layout'>)</span> <span class='hs-conid'>Set</span><span class='hs-varop'>.</span><span class='hs-varid'>empty</span>
+<a name="line-371"></a>    <span class='hs-comment'>-- Associate an ID with each node name.</span>
+<a name="line-372"></a>    <span class='hs-varid'>nodeMap</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span> <span class='hs-varop'>$</span> <span class='hs-varid'>zip</span> <span class='hs-layout'>(</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>keys</span> <span class='hs-varid'>edgeLookup</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>[</span><span class='hs-num'>0</span><span class='hs-keyglyph'>..</span><span class='hs-keyglyph'>]</span>
+<a name="line-373"></a>    <span class='hs-comment'>-- Create the graph.</span>
+<a name="line-374"></a>    <span class='hs-varid'>graph</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>FGL</span><span class='hs-varop'>.</span><span class='hs-varid'>mkGraph</span> <span class='hs-layout'>(</span><span class='hs-varid'>swap</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>toList</span> <span class='hs-varid'>nodeMap</span><span class='hs-layout'>)</span>
+<a name="line-375"></a>                        <span class='hs-keyglyph'>[</span> <span class='hs-layout'>(</span><span class='hs-varid'>nodeMap</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.!</span> <span class='hs-varid'>n</span><span class='hs-layout'>,</span> <span class='hs-varid'>nodeMap</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.!</span> <span class='hs-varid'>m</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varid'>i</span><span class='hs-layout'>,</span> <span class='hs-varid'>j</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-376"></a>                        <span class='hs-keyglyph'>|</span> <span class='hs-layout'>(</span><span class='hs-varid'>n</span><span class='hs-layout'>,</span> <span class='hs-varid'>edges</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>toList</span> <span class='hs-varid'>edgeLookup</span>
+<a name="line-377"></a>                        <span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varid'>m</span><span class='hs-layout'>,</span> <span class='hs-varid'>i</span><span class='hs-layout'>,</span> <span class='hs-varid'>j</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>edges</span>
+<a name="line-378"></a>                        <span class='hs-keyglyph'>]</span>
+<a name="line-379"></a>
+<a name="line-380"></a><a name="GradientFunc"></a><span class='hs-comment'>-- | Function to compute the gradient of y w.r.t. each input.</span>
+<a name="line-381"></a><a name="GradientFunc"></a><span class='hs-comment'>--</span>
+<a name="line-382"></a><a name="GradientFunc"></a><span class='hs-comment'>-- Let y be an arbitrary tensor</span>
+<a name="line-383"></a><a name="GradientFunc"></a><span class='hs-comment'>-- and [w_0, ..., w_n] be the output tensors of a node</span>
+<a name="line-384"></a><a name="GradientFunc"></a><span class='hs-comment'>-- and [v_0, ..., v_n] be the input tensors of the same node.</span>
+<a name="line-385"></a><a name="GradientFunc"></a><span class='hs-comment'>--</span>
+<a name="line-386"></a><a name="GradientFunc"></a><span class='hs-comment'>-- Given [∂y/∂w_0, ..., ∂y/∂w_n] and [v_0, ..., v_n], a GradientFunc computes</span>
+<a name="line-387"></a><a name="GradientFunc"></a><span class='hs-comment'>-- [∂y/∂v_0, ..., ∂y/∂v_n] for a particular op type.</span>
+<a name="line-388"></a><a name="GradientFunc"></a><span class='hs-comment'>--</span>
+<a name="line-389"></a><a name="GradientFunc"></a><span class='hs-comment'>-- A Nothing gradient is equivalent to zero (but allows for short circuiting</span>
+<a name="line-390"></a><a name="GradientFunc"></a><span class='hs-comment'>-- computation when all the gradients for something are Nothing).</span>
+<a name="line-391"></a><a name="GradientFunc"></a><span class='hs-keyword'>type</span> <span class='hs-conid'>GradientFunc</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NodeDef</span>
+<a name="line-392"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Output</span><span class='hs-keyglyph'>]</span>
+<a name="line-393"></a>                    <span class='hs-comment'>-- ^ Input tensors.</span>
+<a name="line-394"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span>
+<a name="line-395"></a>                    <span class='hs-comment'>-- ^ Gradient of y w.r.t. each output tensor.</span>
+<a name="line-396"></a>                    <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Maybe</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span>
+<a name="line-397"></a>                    <span class='hs-comment'>-- ^ Gradient of y w.r.t. each input tensor.</span>
+<a name="line-398"></a>
+<a name="line-399"></a>
+<a name="line-400"></a><a name="toT"></a><span class='hs-comment'>-- TODO(fmayle): Assert the type is correct.</span>
+<a name="line-401"></a><span class='hs-comment'>-- | Create a Tensor from an Output.</span>
+<a name="line-402"></a><span class='hs-definition'>toT</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Output</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span>
+<a name="line-403"></a><span class='hs-definition'>toT</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>ValueKind</span>
+<a name="line-404"></a>
+<a name="line-405"></a><a name="opGrad"></a><span class='hs-comment'>-- | The gradient function for an op type.</span>
+<a name="line-406"></a><span class='hs-comment'>--</span>
+<a name="line-407"></a><span class='hs-comment'>-- These implementations should match their python counterparts in:</span>
+<a name="line-408"></a><span class='hs-comment'>-- third_party/tensorflow/python/ops/*_grad.py</span>
+<a name="line-409"></a><span class='hs-definition'>opGrad</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varop'>.</span> <span class='hs-conid'>GradientCompatible</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Text</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>GradientFunc</span> <span class='hs-varid'>a</span>
+<a name="line-410"></a>
+<a name="line-411"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Abs"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>dz</span> <span class='hs-varop'>*</span> <span class='hs-varid'>signum</span> <span class='hs-varid'>x</span><span class='hs-keyglyph'>]</span>
+<a name="line-412"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Neg"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-keyword'>_</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-comment'>-</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span>
+<a name="line-413"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Relu"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>reluGrad</span> <span class='hs-varid'>dz</span> <span class='hs-varid'>x</span><span class='hs-keyglyph'>]</span>
+<a name="line-414"></a>
+<a name="line-415"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Square"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-416"></a>    <span class='hs-comment'>-- TODO(fmayle): Handle complex numbers.</span>
+<a name="line-417"></a>    <span class='hs-comment'>-- TODO(fmayle): The python code makes dz a control dependency of the 2*x</span>
+<a name="line-418"></a>    <span class='hs-comment'>-- (for performance reasons?). Will need to put these functions in the Build</span>
+<a name="line-419"></a>    <span class='hs-comment'>-- monad to replicate that.</span>
+<a name="line-420"></a>    <span class='hs-keyglyph'>[</span><span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>dz</span> <span class='hs-varop'>*</span> <span class='hs-layout'>(</span><span class='hs-num'>2</span> <span class='hs-varop'>*</span> <span class='hs-varid'>x</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span>
+<a name="line-421"></a>
+<a name="line-422"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Gather"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>indices</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-423"></a>    <span class='hs-comment'>-- TODO(fmayle): The python version uses a better performance implementation</span>
+<a name="line-424"></a>    <span class='hs-comment'>-- when the shape is known without having to run the graph.</span>
+<a name="line-425"></a>    <span class='hs-comment'>-- TODO(fmayle): We shouldn't convert the result to a dense tensor. Sparse</span>
+<a name="line-426"></a>    <span class='hs-comment'>-- tensor support will require some thinking.</span>
+<a name="line-427"></a>    <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>unsortedSegmentSum</span> <span class='hs-varid'>values</span> <span class='hs-varid'>indices'</span> <span class='hs-varid'>numRows</span>
+<a name="line-428"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Nothing</span>
+<a name="line-429"></a>    <span class='hs-keyglyph'>]</span>
+<a name="line-430"></a>  <span class='hs-keyword'>where</span>
+<a name="line-431"></a>    <span class='hs-comment'>-- TODO(gnezdo): Use colocateWith but it requires Build monad.</span>
+<a name="line-432"></a>    <span class='hs-varid'>denseShape</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>shape</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-433"></a>    <span class='hs-varid'>numRows</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>slice</span> <span class='hs-varid'>denseShape</span> <span class='hs-num'>0</span> <span class='hs-layout'>(</span><span class='hs-num'>1</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-434"></a>    <span class='hs-varid'>valuesShape</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>concat</span> <span class='hs-num'>0</span> <span class='hs-keyglyph'>[</span>
+<a name="line-435"></a>                                 <span class='hs-varid'>allDimensions</span>
+<a name="line-436"></a>                               <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>slice</span> <span class='hs-varid'>denseShape</span> <span class='hs-num'>1</span> <span class='hs-layout'>(</span><span class='hs-comment'>-</span><span class='hs-num'>1</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-437"></a>                               <span class='hs-keyglyph'>]</span>
+<a name="line-438"></a>    <span class='hs-varid'>values</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>reshape</span> <span class='hs-varid'>dz</span> <span class='hs-varid'>valuesShape</span>
+<a name="line-439"></a>    <span class='hs-comment'>-- TODO(fmayle): This could be either Int32 or Int64.</span>
+<a name="line-440"></a>    <span class='hs-varid'>indices'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>reshape</span> <span class='hs-varid'>indices</span> <span class='hs-varid'>allDimensions</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Int32</span>
+<a name="line-441"></a>
+<a name="line-442"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Max"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>indices</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-443"></a>    <span class='hs-keyglyph'>[</span><span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>indicators</span> <span class='hs-varop'>`</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>div</span><span class='hs-varop'>`</span> <span class='hs-varid'>numSelected</span> <span class='hs-varop'>*</span> <span class='hs-varid'>dz'</span><span class='hs-layout'>,</span> <span class='hs-conid'>Nothing</span><span class='hs-keyglyph'>]</span>
+<a name="line-444"></a>  <span class='hs-keyword'>where</span>
+<a name="line-445"></a>    <span class='hs-varid'>sx</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>shape</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-446"></a>    <span class='hs-varid'>outputShapeKeptDims</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>reducedShape</span> <span class='hs-varid'>sx</span> <span class='hs-layout'>(</span><span class='hs-varid'>indices</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-447"></a>    <span class='hs-varid'>x'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>reshape</span> <span class='hs-varid'>x</span> <span class='hs-varid'>outputShapeKeptDims</span>
+<a name="line-448"></a>    <span class='hs-varid'>dz'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>reshape</span> <span class='hs-varid'>dz</span> <span class='hs-varid'>outputShapeKeptDims</span>
+<a name="line-449"></a>    <span class='hs-varid'>indicators</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>cast</span> <span class='hs-varop'>$</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>equal</span> <span class='hs-varid'>x'</span> <span class='hs-varid'>x</span>
+<a name="line-450"></a>    <span class='hs-varid'>numSelected</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>reshape</span> <span class='hs-layout'>(</span><span class='hs-varid'>sum</span> <span class='hs-varid'>indicators</span> <span class='hs-varid'>indices</span><span class='hs-layout'>)</span> <span class='hs-varid'>outputShapeKeptDims</span>
+<a name="line-451"></a>
+<a name="line-452"></a><span class='hs-comment'>-- Min and Max have identical gradient implementations.</span>
+<a name="line-453"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Min"</span> <span class='hs-varid'>u</span> <span class='hs-varid'>v</span> <span class='hs-varid'>w</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>opGrad</span> <span class='hs-str'>"Max"</span> <span class='hs-varid'>u</span> <span class='hs-varid'>v</span> <span class='hs-varid'>w</span>
+<a name="line-454"></a>
+<a name="line-455"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Sum"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>indices</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-456"></a>    <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>tile</span> <span class='hs-varid'>grad</span> <span class='hs-varid'>tileScaling</span><span class='hs-layout'>,</span> <span class='hs-conid'>Nothing</span> <span class='hs-keyglyph'>]</span>
+<a name="line-457"></a>  <span class='hs-keyword'>where</span>
+<a name="line-458"></a>    <span class='hs-comment'>-- TODO(gnezdo): Implement the fast-path from math_grad._SumGrad.</span>
+<a name="line-459"></a>    <span class='hs-varid'>sx</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>shape</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-460"></a>    <span class='hs-varid'>outputShapeKeptDims</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>reducedShape</span> <span class='hs-varid'>sx</span> <span class='hs-layout'>(</span><span class='hs-varid'>indices</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-461"></a>    <span class='hs-varid'>tileScaling</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>safeShapeDiv</span> <span class='hs-varid'>sx</span> <span class='hs-varid'>outputShapeKeptDims</span>
+<a name="line-462"></a>    <span class='hs-varid'>grad</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>reshape</span> <span class='hs-varid'>dz</span> <span class='hs-varid'>outputShapeKeptDims</span>
+<a name="line-463"></a>
+<a name="line-464"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Mean"</span> <span class='hs-varid'>u</span> <span class='hs-varid'>v</span><span class='hs-keyglyph'>@</span><span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-keyword'>_</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>w</span> <span class='hs-keyglyph'>=</span>
+<a name="line-465"></a>    <span class='hs-keyglyph'>[</span><span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>dz</span> <span class='hs-varop'>`</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>div</span><span class='hs-varop'>`</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>cast</span> <span class='hs-varid'>factor</span><span class='hs-layout'>,</span> <span class='hs-conid'>Nothing</span><span class='hs-keyglyph'>]</span>
+<a name="line-466"></a>  <span class='hs-keyword'>where</span>
+<a name="line-467"></a>    <span class='hs-keyglyph'>[</span><span class='hs-conid'>Just</span> <span class='hs-varid'>dz</span><span class='hs-layout'>,</span> <span class='hs-conid'>Nothing</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>opGrad</span> <span class='hs-str'>"Sum"</span> <span class='hs-varid'>u</span> <span class='hs-varid'>v</span> <span class='hs-varid'>w</span>
+<a name="line-468"></a>    <span class='hs-varid'>inputShape</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>shape</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-469"></a>    <span class='hs-varid'>outputShape</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>shape</span> <span class='hs-layout'>(</span><span class='hs-varid'>dz</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-470"></a>    <span class='hs-comment'>-- TODO(fmayle): Add fast path when shape is known.</span>
+<a name="line-471"></a>    <span class='hs-varid'>inputSize</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>prod</span> <span class='hs-varid'>inputShape</span> <span class='hs-varop'>$</span> <span class='hs-varid'>rangeOfRank</span> <span class='hs-varid'>inputShape</span>
+<a name="line-472"></a>    <span class='hs-varid'>outputSize</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>prod</span> <span class='hs-varid'>outputShape</span> <span class='hs-varop'>$</span> <span class='hs-varid'>rangeOfRank</span> <span class='hs-varid'>outputShape</span>
+<a name="line-473"></a>    <span class='hs-varid'>factor</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>safeShapeDiv</span> <span class='hs-varid'>inputSize</span> <span class='hs-varid'>outputSize</span>
+<a name="line-474"></a>
+<a name="line-475"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Add"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>y</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-476"></a>    <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>reshape</span> <span class='hs-layout'>(</span><span class='hs-varid'>sum</span> <span class='hs-varid'>dz</span> <span class='hs-varid'>rx</span><span class='hs-layout'>)</span> <span class='hs-varid'>sx</span>
+<a name="line-477"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>reshape</span> <span class='hs-layout'>(</span><span class='hs-varid'>sum</span> <span class='hs-varid'>dz</span> <span class='hs-varid'>ry</span><span class='hs-layout'>)</span> <span class='hs-varid'>sy</span> <span class='hs-keyglyph'>]</span>
+<a name="line-478"></a>  <span class='hs-keyword'>where</span>
+<a name="line-479"></a>    <span class='hs-varid'>sx</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>shape</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-480"></a>    <span class='hs-varid'>sy</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>shape</span> <span class='hs-layout'>(</span><span class='hs-varid'>y</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-481"></a>    <span class='hs-layout'>(</span><span class='hs-varid'>rx</span><span class='hs-layout'>,</span> <span class='hs-varid'>ry</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>broadcastGradientArgs</span> <span class='hs-varid'>sx</span> <span class='hs-varid'>sy</span>
+<a name="line-482"></a>
+<a name="line-483"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Sub"</span> <span class='hs-varid'>u</span> <span class='hs-varid'>v</span> <span class='hs-varid'>w</span> <span class='hs-keyglyph'>=</span>
+<a name="line-484"></a>    <span class='hs-keyglyph'>[</span><span class='hs-conid'>Just</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-conid'>Just</span> <span class='hs-layout'>(</span><span class='hs-comment'>-</span><span class='hs-varid'>y</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span>
+<a name="line-485"></a>  <span class='hs-keyword'>where</span>
+<a name="line-486"></a>    <span class='hs-keyglyph'>[</span><span class='hs-conid'>Just</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-conid'>Just</span> <span class='hs-varid'>y</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>opGrad</span> <span class='hs-str'>"Add"</span> <span class='hs-varid'>u</span> <span class='hs-varid'>v</span> <span class='hs-varid'>w</span>
+<a name="line-487"></a>
+<a name="line-488"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"SoftmaxCrossEntropyWithLogits"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>y</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-layout'>,</span> <span class='hs-keyword'>_</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-489"></a>    <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>expandDims</span> <span class='hs-varid'>dz</span> <span class='hs-layout'>(</span><span class='hs-comment'>-</span><span class='hs-num'>1</span><span class='hs-layout'>)</span> <span class='hs-varop'>*</span> <span class='hs-varid'>snd</span> <span class='hs-layout'>(</span><span class='hs-varid'>softmaxCrossEntropyWithLogits</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span><span class='hs-layout'>)</span>
+<a name="line-490"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Nothing</span> <span class='hs-keyglyph'>]</span>
+<a name="line-491"></a>
+<a name="line-492"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Mul"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>y</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-493"></a>    <span class='hs-comment'>-- TODO(fmayle): Handle complex numbers.</span>
+<a name="line-494"></a>    <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>reshape</span> <span class='hs-layout'>(</span><span class='hs-varid'>sum</span> <span class='hs-layout'>(</span><span class='hs-varid'>dz</span> <span class='hs-varop'>*</span> <span class='hs-varid'>y</span><span class='hs-layout'>)</span> <span class='hs-varid'>rx</span><span class='hs-layout'>)</span> <span class='hs-varid'>sx</span>
+<a name="line-495"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>reshape</span> <span class='hs-layout'>(</span><span class='hs-varid'>sum</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span> <span class='hs-varop'>*</span> <span class='hs-varid'>dz</span><span class='hs-layout'>)</span> <span class='hs-varid'>ry</span><span class='hs-layout'>)</span> <span class='hs-varid'>sy</span> <span class='hs-keyglyph'>]</span>
+<a name="line-496"></a>  <span class='hs-keyword'>where</span>
+<a name="line-497"></a>    <span class='hs-varid'>sx</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>shape</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-498"></a>    <span class='hs-varid'>sy</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>shape</span> <span class='hs-layout'>(</span><span class='hs-varid'>y</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-499"></a>    <span class='hs-layout'>(</span><span class='hs-varid'>rx</span><span class='hs-layout'>,</span> <span class='hs-varid'>ry</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>broadcastGradientArgs</span> <span class='hs-varid'>sx</span> <span class='hs-varid'>sy</span>
+<a name="line-500"></a>
+<a name="line-501"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Div"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>y</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-502"></a>    <span class='hs-comment'>-- TODO(fmayle): Handle complex numbers.</span>
+<a name="line-503"></a>    <span class='hs-comment'>-- TODO(gnezdo): Provide Fractional instance and use '/' instead of div.</span>
+<a name="line-504"></a>    <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>reshape</span> <span class='hs-layout'>(</span><span class='hs-varid'>sum</span> <span class='hs-layout'>(</span><span class='hs-varid'>dz</span> <span class='hs-varop'>`</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>div</span><span class='hs-varop'>`</span> <span class='hs-varid'>y</span><span class='hs-layout'>)</span> <span class='hs-varid'>rx</span><span class='hs-layout'>)</span> <span class='hs-varid'>sx</span>
+<a name="line-505"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>reshape</span> <span class='hs-layout'>(</span><span class='hs-varid'>sum</span> <span class='hs-layout'>(</span><span class='hs-varid'>dz</span> <span class='hs-varop'>*</span> <span class='hs-layout'>(</span><span class='hs-varid'>negate</span> <span class='hs-varid'>x</span> <span class='hs-varop'>`</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>div</span><span class='hs-varop'>`</span> <span class='hs-layout'>(</span><span class='hs-varid'>y</span> <span class='hs-varop'>*</span> <span class='hs-varid'>y</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-varid'>ry</span><span class='hs-layout'>)</span> <span class='hs-varid'>sy</span>
+<a name="line-506"></a>    <span class='hs-keyglyph'>]</span>
+<a name="line-507"></a>  <span class='hs-keyword'>where</span>
+<a name="line-508"></a>    <span class='hs-varid'>sx</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>shape</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-509"></a>    <span class='hs-varid'>sy</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>shape</span> <span class='hs-layout'>(</span><span class='hs-varid'>y</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-510"></a>    <span class='hs-layout'>(</span><span class='hs-varid'>rx</span><span class='hs-layout'>,</span> <span class='hs-varid'>ry</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>broadcastGradientArgs</span> <span class='hs-varid'>sx</span> <span class='hs-varid'>sy</span>
+<a name="line-511"></a>
+<a name="line-512"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"MatMul"</span> <span class='hs-varid'>nodeDef</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>y</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-513"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>transposeA</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lookupAttr</span> <span class='hs-varid'>nodeDef</span> <span class='hs-str'>"transpose_a"</span>
+<a name="line-514"></a>        <span class='hs-varid'>transposeB</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lookupAttr</span> <span class='hs-varid'>nodeDef</span> <span class='hs-str'>"transpose_b"</span>
+<a name="line-515"></a>        <span class='hs-varid'>transAttrs</span> <span class='hs-varid'>a</span> <span class='hs-varid'>b</span> <span class='hs-keyglyph'>=</span>
+<a name="line-516"></a>            <span class='hs-layout'>(</span><span class='hs-varid'>tensorAttr</span> <span class='hs-str'>"transpose_a"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-varid'>tensorAttr</span> <span class='hs-str'>"transpose_b"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>b</span><span class='hs-layout'>)</span>
+<a name="line-517"></a>    <span class='hs-keyword'>in</span> <span class='hs-keyword'>case</span> <span class='hs-layout'>(</span><span class='hs-varid'>transposeA</span><span class='hs-layout'>,</span> <span class='hs-varid'>transposeB</span><span class='hs-layout'>)</span> <span class='hs-keyword'>of</span>
+<a name="line-518"></a>       <span class='hs-layout'>(</span><span class='hs-conid'>False</span><span class='hs-layout'>,</span> <span class='hs-conid'>False</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-519"></a>           <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-layout'>(</span><span class='hs-varid'>dz</span> <span class='hs-varop'>`matMul`</span> <span class='hs-varid'>y</span><span class='hs-layout'>)</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>transAttrs</span> <span class='hs-conid'>False</span> <span class='hs-conid'>True</span>
+<a name="line-520"></a>           <span class='hs-layout'>,</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span> <span class='hs-varop'>`matMul`</span> <span class='hs-varid'>dz</span><span class='hs-layout'>)</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>transAttrs</span> <span class='hs-conid'>True</span> <span class='hs-conid'>False</span> <span class='hs-keyglyph'>]</span>
+<a name="line-521"></a>       <span class='hs-layout'>(</span><span class='hs-conid'>False</span><span class='hs-layout'>,</span> <span class='hs-conid'>True</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-522"></a>           <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>dz</span> <span class='hs-varop'>`matMul`</span> <span class='hs-varid'>y</span>
+<a name="line-523"></a>           <span class='hs-layout'>,</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span> <span class='hs-varop'>`matMul`</span> <span class='hs-varid'>dz</span><span class='hs-layout'>)</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>transAttrs</span> <span class='hs-conid'>True</span> <span class='hs-conid'>False</span> <span class='hs-keyglyph'>]</span>
+<a name="line-524"></a>       <span class='hs-layout'>(</span><span class='hs-conid'>True</span><span class='hs-layout'>,</span> <span class='hs-conid'>False</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-525"></a>           <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-layout'>(</span><span class='hs-varid'>dz</span> <span class='hs-varop'>`matMul`</span> <span class='hs-varid'>y</span><span class='hs-layout'>)</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>transAttrs</span> <span class='hs-conid'>False</span> <span class='hs-conid'>True</span>
+<a name="line-526"></a>           <span class='hs-layout'>,</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>x</span> <span class='hs-varop'>`matMul`</span> <span class='hs-varid'>dz</span> <span class='hs-keyglyph'>]</span>
+<a name="line-527"></a>       <span class='hs-layout'>(</span><span class='hs-conid'>True</span><span class='hs-layout'>,</span> <span class='hs-conid'>True</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-528"></a>           <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-layout'>(</span><span class='hs-varid'>dz</span> <span class='hs-varop'>`matMul`</span> <span class='hs-varid'>y</span><span class='hs-layout'>)</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>transAttrs</span> <span class='hs-conid'>True</span> <span class='hs-conid'>True</span>
+<a name="line-529"></a>           <span class='hs-layout'>,</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span> <span class='hs-varop'>`matMul`</span> <span class='hs-varid'>dz</span><span class='hs-layout'>)</span> <span class='hs-varop'>&amp;</span> <span class='hs-varid'>transAttrs</span> <span class='hs-conid'>True</span> <span class='hs-conid'>True</span> <span class='hs-keyglyph'>]</span>
+<a name="line-530"></a>
+<a name="line-531"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Transpose"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-keyword'>_</span><span class='hs-layout'>,</span> <span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>p</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-532"></a>    <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>transpose</span> <span class='hs-varid'>dz</span>
+<a name="line-533"></a>            <span class='hs-layout'>(</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>invertPermutation</span> <span class='hs-varid'>p</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-534"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Nothing</span>
+<a name="line-535"></a>    <span class='hs-keyglyph'>]</span>
+<a name="line-536"></a>
+<a name="line-537"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Conv2D"</span> <span class='hs-varid'>nodeDef</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>y</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-538"></a>    <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>conv2DBackpropInput</span> <span class='hs-layout'>(</span><span class='hs-varid'>shape</span> <span class='hs-varid'>x</span><span class='hs-layout'>)</span> <span class='hs-varid'>y</span> <span class='hs-varid'>dz</span>
+<a name="line-539"></a>          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>tensorAttr</span> <span class='hs-str'>"strides"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>strides</span>
+<a name="line-540"></a>          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>tensorAttr</span> <span class='hs-str'>"padding"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>padding</span>
+<a name="line-541"></a>          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>tensorAttr</span> <span class='hs-str'>"use_cudnn_on_gpu"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>useCudnnOnGpu</span>
+<a name="line-542"></a>          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>tensorAttr</span> <span class='hs-str'>"data_format"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>dataFormat</span>
+<a name="line-543"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>conv2DBackpropFilter</span> <span class='hs-varid'>x</span> <span class='hs-layout'>(</span><span class='hs-varid'>shape</span> <span class='hs-varid'>y</span><span class='hs-layout'>)</span> <span class='hs-varid'>dz</span>
+<a name="line-544"></a>          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>tensorAttr</span> <span class='hs-str'>"strides"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>strides</span>
+<a name="line-545"></a>          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>tensorAttr</span> <span class='hs-str'>"padding"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>padding</span>
+<a name="line-546"></a>          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>tensorAttr</span> <span class='hs-str'>"use_cudnn_on_gpu"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>useCudnnOnGpu</span>
+<a name="line-547"></a>          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>tensorAttr</span> <span class='hs-str'>"data_format"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>dataFormat</span>
+<a name="line-548"></a>    <span class='hs-keyglyph'>]</span>
+<a name="line-549"></a>  <span class='hs-keyword'>where</span>
+<a name="line-550"></a>    <span class='hs-varid'>strides</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lookupAttr</span> <span class='hs-varid'>nodeDef</span> <span class='hs-str'>"strides"</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span>
+<a name="line-551"></a>    <span class='hs-varid'>padding</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lookupAttr</span> <span class='hs-varid'>nodeDef</span> <span class='hs-str'>"padding"</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>ByteString</span>
+<a name="line-552"></a>    <span class='hs-varid'>useCudnnOnGpu</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lookupAttr</span> <span class='hs-varid'>nodeDef</span> <span class='hs-str'>"use_cudnn_on_gpu"</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Bool</span>
+<a name="line-553"></a>    <span class='hs-varid'>dataFormat</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lookupAttr</span> <span class='hs-varid'>nodeDef</span> <span class='hs-str'>"data_format"</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>ByteString</span>
+<a name="line-554"></a>
+<a name="line-555"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"MaxPool"</span> <span class='hs-varid'>nodeDef</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-556"></a>    <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>maxPoolGrad</span> <span class='hs-varid'>x</span> <span class='hs-varid'>output</span> <span class='hs-varid'>dz</span>
+<a name="line-557"></a>          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>tensorAttr</span> <span class='hs-str'>"ksize"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>ksize</span>
+<a name="line-558"></a>          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>tensorAttr</span> <span class='hs-str'>"strides"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>strides</span>
+<a name="line-559"></a>          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>tensorAttr</span> <span class='hs-str'>"padding"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>padding</span>
+<a name="line-560"></a>          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>tensorAttr</span> <span class='hs-str'>"data_format"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>dataFormat</span>
+<a name="line-561"></a>    <span class='hs-keyglyph'>]</span>
+<a name="line-562"></a>  <span class='hs-keyword'>where</span>
+<a name="line-563"></a>    <span class='hs-varid'>output</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span>
+<a name="line-564"></a>    <span class='hs-varid'>output</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>toT</span> <span class='hs-varop'>$</span> <span class='hs-conid'>Output</span> <span class='hs-num'>0</span> <span class='hs-layout'>(</span><span class='hs-conid'>Rendered</span> <span class='hs-varid'>nodeDef</span><span class='hs-layout'>)</span>
+<a name="line-565"></a>    <span class='hs-varid'>ksize</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lookupAttr</span> <span class='hs-varid'>nodeDef</span> <span class='hs-str'>"ksize"</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span>
+<a name="line-566"></a>    <span class='hs-varid'>strides</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lookupAttr</span> <span class='hs-varid'>nodeDef</span> <span class='hs-str'>"strides"</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span>
+<a name="line-567"></a>    <span class='hs-varid'>padding</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lookupAttr</span> <span class='hs-varid'>nodeDef</span> <span class='hs-str'>"padding"</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>ByteString</span>
+<a name="line-568"></a>    <span class='hs-varid'>dataFormat</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lookupAttr</span> <span class='hs-varid'>nodeDef</span> <span class='hs-str'>"data_format"</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>ByteString</span>
+<a name="line-569"></a>
+<a name="line-570"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Reshape"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-keyword'>_</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-571"></a>    <span class='hs-keyglyph'>[</span><span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>reshape</span> <span class='hs-varid'>dz</span> <span class='hs-varop'>$</span> <span class='hs-varid'>shape</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-conid'>Nothing</span><span class='hs-keyglyph'>]</span>
+<a name="line-572"></a>
+<a name="line-573"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"OneHot"</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span> <span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span> <span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span> <span class='hs-conid'>Nothing</span><span class='hs-keyglyph'>]</span>
+<a name="line-574"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"TruncatedNormal"</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Nothing</span><span class='hs-keyglyph'>]</span>
+<a name="line-575"></a>
+<a name="line-576"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"RefIdentity"</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Just</span> <span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span>
+<a name="line-577"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Cast"</span> <span class='hs-varid'>nodeDef</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Just</span> <span class='hs-varid'>reverseCast</span><span class='hs-keyglyph'>]</span>
+<a name="line-578"></a>  <span class='hs-keyword'>where</span>
+<a name="line-579"></a>    <span class='hs-comment'>-- TODO(gnezdo): too permissive, python only allows float types as src_type.</span>
+<a name="line-580"></a>    <span class='hs-varid'>reverseCast</span> <span class='hs-keyglyph'>=</span>
+<a name="line-581"></a>        <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Cast"</span>
+<a name="line-582"></a>                 <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"DstT"</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>lookupAttr</span> <span class='hs-varid'>nodeDef</span> <span class='hs-str'>"SrcT"</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>
+<a name="line-583"></a>                 <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"SrcT"</span> <span class='hs-varop'>.~</span> <span class='hs-layout'>(</span><span class='hs-varid'>lookupAttr</span> <span class='hs-varid'>nodeDef</span> <span class='hs-str'>"DstT"</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-584"></a>        <span class='hs-varid'>dz</span>
+<a name="line-585"></a>
+<a name="line-586"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"DynamicStitch"</span> <span class='hs-varid'>nodeDef</span> <span class='hs-varid'>inputs</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-587"></a>    <span class='hs-varid'>replicate</span> <span class='hs-varid'>halfLen</span> <span class='hs-conid'>Nothing</span> <span class='hs-varop'>++</span> <span class='hs-varid'>valuesGrads</span>
+<a name="line-588"></a>  <span class='hs-keyword'>where</span>
+<a name="line-589"></a>    <span class='hs-varid'>halfLen</span> <span class='hs-keyglyph'>=</span>
+<a name="line-590"></a>        <span class='hs-keyword'>let</span> <span class='hs-varid'>len</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>length</span> <span class='hs-varid'>inputs</span>
+<a name="line-591"></a>            <span class='hs-varid'>half</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>len</span> <span class='hs-varop'>`div`</span> <span class='hs-num'>2</span>
+<a name="line-592"></a>        <span class='hs-keyword'>in</span> <span class='hs-keyword'>if</span> <span class='hs-num'>2</span> <span class='hs-varop'>*</span> <span class='hs-varid'>half</span> <span class='hs-varop'>==</span> <span class='hs-varid'>len</span>
+<a name="line-593"></a>           <span class='hs-keyword'>then</span> <span class='hs-varid'>half</span>
+<a name="line-594"></a>           <span class='hs-keyword'>else</span> <span class='hs-varid'>error</span> <span class='hs-layout'>(</span><span class='hs-str'>"Uneven input size "</span> <span class='hs-varop'>++</span> <span class='hs-varid'>show</span> <span class='hs-layout'>(</span><span class='hs-varid'>len</span><span class='hs-layout'>,</span> <span class='hs-varid'>showMessage</span> <span class='hs-varid'>nodeDef</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-595"></a>    <span class='hs-varid'>valuesGrads</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>gather</span> <span class='hs-varid'>dz</span> <span class='hs-layout'>(</span><span class='hs-varid'>toT</span> <span class='hs-varid'>idx</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-596"></a>                  <span class='hs-keyglyph'>|</span> <span class='hs-varid'>idx</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>take</span> <span class='hs-varid'>halfLen</span> <span class='hs-varid'>inputs</span>
+<a name="line-597"></a>                  <span class='hs-keyglyph'>]</span>
+<a name="line-598"></a>
+<a name="line-599"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"DynamicPartition"</span> <span class='hs-varid'>nodeDef</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>xs</span><span class='hs-layout'>,</span> <span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>indices</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>dz</span> <span class='hs-keyglyph'>=</span>
+<a name="line-600"></a>    <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varid'>reconstructed</span><span class='hs-layout'>,</span> <span class='hs-conid'>Nothing</span> <span class='hs-keyglyph'>]</span>
+<a name="line-601"></a>  <span class='hs-keyword'>where</span>
+<a name="line-602"></a>    <span class='hs-varid'>reconstructed</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>reshape</span> <span class='hs-varid'>stitched</span>
+<a name="line-603"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>shape</span> <span class='hs-layout'>(</span><span class='hs-varid'>xs</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-604"></a>    <span class='hs-varid'>stitched</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>dynamicStitch</span> <span class='hs-varid'>partitionedIndices</span> <span class='hs-varid'>dz</span>
+<a name="line-605"></a>    <span class='hs-varid'>partitionedIndices</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>dynamicPartition</span> <span class='hs-varid'>np</span> <span class='hs-varid'>originalIndices</span> <span class='hs-varid'>indices</span>
+<a name="line-606"></a>    <span class='hs-varid'>np</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>lookupAttr</span> <span class='hs-varid'>nodeDef</span> <span class='hs-str'>"num_partitions"</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int64</span>
+<a name="line-607"></a>    <span class='hs-varid'>originalIndices</span> <span class='hs-keyglyph'>=</span>
+<a name="line-608"></a>        <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>reshape</span> <span class='hs-layout'>(</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>range</span> <span class='hs-num'>0</span> <span class='hs-layout'>(</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>size</span> <span class='hs-varid'>indices</span><span class='hs-layout'>)</span> <span class='hs-num'>1</span><span class='hs-layout'>)</span> <span class='hs-varid'>prefixShape</span>
+<a name="line-609"></a>    <span class='hs-varid'>prefixShape</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>shapeInt32</span> <span class='hs-varid'>indices</span>
+<a name="line-610"></a>    <span class='hs-varid'>shapeInt32</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>shape</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Int32</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Int32</span>
+<a name="line-611"></a>
+<a name="line-612"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Select"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>c</span><span class='hs-layout'>,</span> <span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-keyword'>_</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-613"></a>    <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Nothing</span>
+<a name="line-614"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>select</span> <span class='hs-varid'>c</span> <span class='hs-varid'>dz</span> <span class='hs-varid'>zeros</span>
+<a name="line-615"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>select</span> <span class='hs-varid'>c</span> <span class='hs-varid'>zeros</span> <span class='hs-varid'>dz</span>
+<a name="line-616"></a>    <span class='hs-keyglyph'>]</span>
+<a name="line-617"></a>  <span class='hs-keyword'>where</span> <span class='hs-varid'>zeros</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>zerosLike</span> <span class='hs-varid'>x</span>
+<a name="line-618"></a>
+<a name="line-619"></a><span class='hs-comment'>-- TODO(gnezdo): Unlike Python, no control dependency on dz.</span>
+<a name="line-620"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Log"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>dz</span> <span class='hs-varop'>*</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>inv</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>]</span>
+<a name="line-621"></a><span class='hs-comment'>-- TODO(gnezdo): Reuse the output instead of doing another exp,</span>
+<a name="line-622"></a><span class='hs-comment'>-- though, it is probably CSE'd away anyway.</span>
+<a name="line-623"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Exp"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-varid'>dz</span> <span class='hs-varop'>*</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>exp</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>]</span>
+<a name="line-624"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"SparseSegmentSum"</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x</span><span class='hs-layout'>,</span> <span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>y</span><span class='hs-layout'>,</span> <span class='hs-varid'>toT</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>t</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>dz</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>=</span>
+<a name="line-625"></a>    <span class='hs-keyglyph'>[</span> <span class='hs-conid'>Just</span> <span class='hs-varop'>$</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>unsortedSegmentSum</span>
+<a name="line-626"></a>             <span class='hs-layout'>(</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>gather</span> <span class='hs-varid'>dz</span> <span class='hs-layout'>(</span><span class='hs-varid'>t</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Int32</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-627"></a>             <span class='hs-layout'>(</span><span class='hs-varid'>y</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Int32</span><span class='hs-layout'>)</span> <span class='hs-varid'>inputRows</span>
+<a name="line-628"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Nothing</span>
+<a name="line-629"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>Nothing</span>
+<a name="line-630"></a>    <span class='hs-keyglyph'>]</span>
+<a name="line-631"></a>  <span class='hs-keyword'>where</span> <span class='hs-varid'>inputRows</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>slice</span> <span class='hs-layout'>(</span><span class='hs-varid'>shape</span> <span class='hs-layout'>(</span><span class='hs-varid'>x</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>scalar</span> <span class='hs-layout'>(</span><span class='hs-num'>0</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int32</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>scalar</span> <span class='hs-num'>1</span><span class='hs-layout'>)</span>
+<a name="line-632"></a>
+<a name="line-633"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"LabelClasses"</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span> <span class='hs-conid'>Nothing</span><span class='hs-keyglyph'>]</span>
+<a name="line-634"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"LabelWeights"</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Nothing</span><span class='hs-keyglyph'>]</span>
+<a name="line-635"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Size"</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Nothing</span><span class='hs-keyglyph'>]</span>
+<a name="line-636"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"ZerosLike"</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Nothing</span><span class='hs-keyglyph'>]</span>
+<a name="line-637"></a>
+<a name="line-638"></a><span class='hs-comment'>-- TODO(fmayle): These can go away if we properly prune the graph.</span>
+<a name="line-639"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Const"</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span> <span class='hs-conid'>Nothing</span><span class='hs-keyglyph'>]</span>
+<a name="line-640"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Placeholder"</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span>
+<a name="line-641"></a><span class='hs-definition'>opGrad</span> <span class='hs-str'>"Variable"</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span>
+<a name="line-642"></a>
+<a name="line-643"></a><span class='hs-definition'>opGrad</span> <span class='hs-varid'>n</span> <span class='hs-varid'>nodeDef</span> <span class='hs-varid'>ins</span> <span class='hs-varid'>grads</span> <span class='hs-keyglyph'>=</span>
+<a name="line-644"></a>    <span class='hs-varid'>error</span> <span class='hs-varop'>$</span> <span class='hs-str'>"no gradient implemented for "</span> <span class='hs-varop'>++</span>
+<a name="line-645"></a>            <span class='hs-varid'>show</span> <span class='hs-layout'>(</span><span class='hs-varid'>n</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>ins</span><span class='hs-layout'>,</span> <span class='hs-varid'>length</span> <span class='hs-varid'>grads</span><span class='hs-layout'>,</span> <span class='hs-varid'>showMessage</span> <span class='hs-varid'>nodeDef</span><span class='hs-layout'>,</span> <span class='hs-varid'>ins</span><span class='hs-layout'>)</span>
+<a name="line-646"></a>
+<a name="line-647"></a><a name="numOutputs"></a><span class='hs-comment'>-- | The number of outputs for an op type.</span>
+<a name="line-648"></a><span class='hs-definition'>numOutputs</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>NodeDef</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>OutputIx</span>
+<a name="line-649"></a><span class='hs-definition'>numOutputs</span> <span class='hs-varid'>o</span> <span class='hs-keyglyph'>=</span>
+<a name="line-650"></a>    <span class='hs-keyword'>case</span> <span class='hs-varid'>o</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>op</span> <span class='hs-keyword'>of</span>
+<a name="line-651"></a>        <span class='hs-str'>"Abs"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-652"></a>        <span class='hs-str'>"Add"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-653"></a>        <span class='hs-str'>"Cast"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-654"></a>        <span class='hs-str'>"Const"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-655"></a>        <span class='hs-str'>"Conv2D"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-656"></a>        <span class='hs-str'>"Div"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-657"></a>        <span class='hs-str'>"DynamicStitch"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-658"></a>        <span class='hs-str'>"DynamicPartition"</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-659"></a>            <span class='hs-varid'>fromIntegral</span> <span class='hs-layout'>(</span><span class='hs-varid'>lookupAttr</span> <span class='hs-varid'>o</span> <span class='hs-str'>"num_partitions"</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-660"></a>        <span class='hs-str'>"Exp"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-661"></a>        <span class='hs-str'>"Gather"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-662"></a>        <span class='hs-str'>"LabelClasses"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-663"></a>        <span class='hs-str'>"LabelWeights"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-664"></a>        <span class='hs-str'>"Log"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-665"></a>        <span class='hs-str'>"MatMul"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-666"></a>        <span class='hs-str'>"Max"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-667"></a>        <span class='hs-str'>"MaxPool"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-668"></a>        <span class='hs-str'>"Mean"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-669"></a>        <span class='hs-str'>"Min"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-670"></a>        <span class='hs-str'>"Mul"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-671"></a>        <span class='hs-str'>"Neg"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-672"></a>        <span class='hs-str'>"Placeholder"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-673"></a>        <span class='hs-str'>"OneHot"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-674"></a>        <span class='hs-str'>"RefIdentity"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-675"></a>        <span class='hs-str'>"Relu"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-676"></a>        <span class='hs-str'>"Reshape"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-677"></a>        <span class='hs-str'>"Select"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-678"></a>        <span class='hs-str'>"Size"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-679"></a>        <span class='hs-str'>"SoftmaxCrossEntropyWithLogits"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>2</span>
+<a name="line-680"></a>        <span class='hs-str'>"Square"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-681"></a>        <span class='hs-str'>"SparseSegmentSum"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-682"></a>        <span class='hs-str'>"Sub"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-683"></a>        <span class='hs-str'>"Sum"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-684"></a>        <span class='hs-str'>"Transpose"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-685"></a>        <span class='hs-str'>"TruncatedNormal"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-686"></a>        <span class='hs-str'>"Variable"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-687"></a>        <span class='hs-str'>"ZerosLike"</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-num'>1</span>
+<a name="line-688"></a>        <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>error</span> <span class='hs-varop'>$</span> <span class='hs-str'>"numOuputs not implemented for "</span> <span class='hs-varop'>++</span> <span class='hs-varid'>show</span> <span class='hs-layout'>(</span><span class='hs-varid'>o</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>op</span><span class='hs-layout'>)</span>
+<a name="line-689"></a>
+<a name="line-690"></a><a name="safeShapeDiv"></a><span class='hs-comment'>-- Divides `x / y` assuming `x, y &gt;= 0`, treating `0 / 0 = 0`</span>
+<a name="line-691"></a><span class='hs-definition'>safeShapeDiv</span> <span class='hs-varid'>x</span> <span class='hs-varid'>y</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>x</span> <span class='hs-varop'>`</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>div</span><span class='hs-varop'>`</span> <span class='hs-layout'>(</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>maximum</span> <span class='hs-varid'>y</span> <span class='hs-num'>1</span><span class='hs-layout'>)</span>
+<a name="line-692"></a>
+<a name="line-693"></a><a name="allDimensions"></a><span class='hs-definition'>allDimensions</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>vector</span> <span class='hs-keyglyph'>[</span><span class='hs-comment'>-</span><span class='hs-num'>1</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int32</span><span class='hs-keyglyph'>]</span>
+<a name="line-694"></a>
+<a name="line-695"></a><a name="rangeOfRank"></a><span class='hs-definition'>rangeOfRank</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>range</span> <span class='hs-num'>0</span> <span class='hs-layout'>(</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>rank</span> <span class='hs-varid'>x</span><span class='hs-layout'>)</span> <span class='hs-num'>1</span>
+<a name="line-696"></a>
+<a name="line-697"></a><a name="lookupAttr"></a><span class='hs-definition'>lookupAttr</span> <span class='hs-varid'>nodeDef</span> <span class='hs-varid'>attrName</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>nodeDef</span> <span class='hs-varop'>^.</span> <span class='hs-varid'>attr</span> <span class='hs-varop'>.</span> <span class='hs-varid'>at</span> <span class='hs-varid'>attrName</span> <span class='hs-varop'>.</span> <span class='hs-varid'>non</span> <span class='hs-varid'>def</span> <span class='hs-varop'>.</span> <span class='hs-varid'>attrLens</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/src/TensorFlow-Ops.html b/docs/haddock/tensorflow-ops-0.1.0.0/src/TensorFlow-Ops.html
new file mode 100644
index 0000000..b26d3aa
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/src/TensorFlow-Ops.html
@@ -0,0 +1,307 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/Ops.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>-- | This module contains definitions for some built-in TensorFlow operations.</span>
+<a name="line-16"></a><span class='hs-comment'>--</span>
+<a name="line-17"></a><span class='hs-comment'>-- Note that certain, "stateful" ops like 'variable' and 'assign' return a</span>
+<a name="line-18"></a><span class='hs-comment'>-- 'Build' action (e.g., @Build (Tensor Ref a)@ instead of a pure value; the</span>
+<a name="line-19"></a><span class='hs-comment'>-- returned 'Tensor's are always rendered in the current 'Build' context.  This</span>
+<a name="line-20"></a><span class='hs-comment'>-- approach helps us avoid problems with inlining or common subexpression</span>
+<a name="line-21"></a><span class='hs-comment'>-- elimination, by writing</span>
+<a name="line-22"></a><span class='hs-comment'>--</span>
+<a name="line-23"></a><span class='hs-comment'>-- &gt; do</span>
+<a name="line-24"></a><span class='hs-comment'>-- &gt;     v &lt;- variable []</span>
+<a name="line-25"></a><span class='hs-comment'>-- &gt;     w &lt;- assign v 3</span>
+<a name="line-26"></a><span class='hs-comment'>-- &gt;     render $ w * w</span>
+<a name="line-27"></a><span class='hs-comment'>--</span>
+<a name="line-28"></a><span class='hs-comment'>-- instead of</span>
+<a name="line-29"></a><span class='hs-comment'>--</span>
+<a name="line-30"></a><span class='hs-comment'>-- &gt; let</span>
+<a name="line-31"></a><span class='hs-comment'>-- &gt;    v = variable []</span>
+<a name="line-32"></a><span class='hs-comment'>-- &gt;    w = assign v 3</span>
+<a name="line-33"></a><span class='hs-comment'>-- &gt; in w * w</span>
+<a name="line-34"></a><span class='hs-comment'>--</span>
+<a name="line-35"></a><span class='hs-comment'>-- since the latter could be reasonably transformed by the compiler into (or</span>
+<a name="line-36"></a><span class='hs-comment'>-- vice versa)</span>
+<a name="line-37"></a><span class='hs-comment'>--</span>
+<a name="line-38"></a><span class='hs-comment'>-- &gt; let</span>
+<a name="line-39"></a><span class='hs-comment'>-- &gt;    v = variable []</span>
+<a name="line-40"></a><span class='hs-comment'>-- &gt;    w = assign v 3</span>
+<a name="line-41"></a><span class='hs-comment'>-- &gt;    w' = assign v 3</span>
+<a name="line-42"></a><span class='hs-comment'>-- &gt; in w * w'</span>
+<a name="line-43"></a><span class='hs-comment'>--</span>
+<a name="line-44"></a><span class='hs-comment'>-- Ops should return a 'Build' action if their original 'OpDef' marks them as</span>
+<a name="line-45"></a><span class='hs-comment'>-- stateful, or if they take any Refs as input.  (This mirrors the rules that</span>
+<a name="line-46"></a><span class='hs-comment'>-- TensorFlow uses to avoid common subexpression elimination.)</span>
+<a name="line-47"></a><span class='hs-comment'>{-# LANGUAGE ConstraintKinds #-}</span>
+<a name="line-48"></a><span class='hs-comment'>{-# LANGUAGE DataKinds #-}</span>
+<a name="line-49"></a><span class='hs-comment'>{-# LANGUAGE FlexibleInstances #-}</span>
+<a name="line-50"></a><span class='hs-comment'>{-# LANGUAGE OverloadedLists #-}</span>
+<a name="line-51"></a><span class='hs-comment'>{-# LANGUAGE OverloadedStrings #-}</span>
+<a name="line-52"></a><span class='hs-comment'>{-# LANGUAGE RankNTypes #-}</span>
+<a name="line-53"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables #-}</span>
+<a name="line-54"></a><span class='hs-comment'>{-# LANGUAGE TypeFamilies #-}</span>
+<a name="line-55"></a><span class='hs-comment'>{-# LANGUAGE UndecidableInstances #-}</span>
+<a name="line-56"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-orphans #-}</span>
+<a name="line-57"></a>
+<a name="line-58"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Ops</span>
+<a name="line-59"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>add</span>
+<a name="line-60"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>abs</span>
+<a name="line-61"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>addN</span>
+<a name="line-62"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>argMax</span>
+<a name="line-63"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>assign</span>
+<a name="line-64"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>broadcastGradientArgs</span>
+<a name="line-65"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>cast</span>
+<a name="line-66"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>concat</span>
+<a name="line-67"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>constant</span>
+<a name="line-68"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>expandDims</span>
+<a name="line-69"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>initializedVariable</span>
+<a name="line-70"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>zeroInitializedVariable</span>
+<a name="line-71"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>fill</span>
+<a name="line-72"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>matMul</span>
+<a name="line-73"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>matTranspose</span>
+<a name="line-74"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>mul</span>
+<a name="line-75"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>neg</span>
+<a name="line-76"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>pack</span>
+<a name="line-77"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>placeholder</span>
+<a name="line-78"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>range</span>
+<a name="line-79"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>reducedShape</span>
+<a name="line-80"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>relu</span>
+<a name="line-81"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>reluGrad</span>
+<a name="line-82"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>reshape</span>
+<a name="line-83"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>restore</span>
+<a name="line-84"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>restoreFromName</span>
+<a name="line-85"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>save</span>
+<a name="line-86"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>scalar</span>
+<a name="line-87"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>shape</span>
+<a name="line-88"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>sign</span>
+<a name="line-89"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>size</span>
+<a name="line-90"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>softmax</span>
+<a name="line-91"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>softmaxCrossEntropyWithLogits</span>
+<a name="line-92"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>sparseToDense</span>
+<a name="line-93"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>sub</span>
+<a name="line-94"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>sum</span>
+<a name="line-95"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>topK</span>
+<a name="line-96"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>transpose</span>
+<a name="line-97"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>truncatedNormal</span>
+<a name="line-98"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>variable</span>
+<a name="line-99"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>vector</span>
+<a name="line-100"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>zeros</span>
+<a name="line-101"></a>    <span class='hs-layout'>,</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>zerosLike</span>
+<a name="line-102"></a>    <span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-103"></a>
+<a name="line-104"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-layout'>(</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>
+<a name="line-105"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Complex</span> <span class='hs-layout'>(</span><span class='hs-conid'>Complex</span><span class='hs-layout'>)</span>
+<a name="line-106"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span> <span class='hs-layout'>(</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-107"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Prelude</span> <span class='hs-varid'>hiding</span> <span class='hs-layout'>(</span><span class='hs-varid'>abs</span><span class='hs-layout'>,</span> <span class='hs-varid'>sum</span><span class='hs-layout'>,</span> <span class='hs-varid'>concat</span><span class='hs-layout'>)</span>
+<a name="line-108"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span> <span class='hs-layout'>(</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-109"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Encoding</span> <span class='hs-layout'>(</span><span class='hs-varid'>encodeUtf8</span><span class='hs-layout'>)</span>
+<a name="line-110"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varop'>.~</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>&amp;</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-111"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Printf</span> <span class='hs-layout'>(</span><span class='hs-varid'>printf</span><span class='hs-layout'>)</span>
+<a name="line-112"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span>
+<a name="line-113"></a>    <span class='hs-layout'>(</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-114"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>dtype</span>
+<a name="line-115"></a>    <span class='hs-layout'>,</span> <span class='hs-varid'>tensorShape</span>
+<a name="line-116"></a>    <span class='hs-layout'>)</span>
+<a name="line-117"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span>
+<a name="line-118"></a>  <span class='hs-keyword'>as</span> <span class='hs-conid'>TensorShape</span>
+<a name="line-119"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Build</span>
+<a name="line-120"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>BuildOp</span>
+<a name="line-121"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>ControlFlow</span> <span class='hs-layout'>(</span><span class='hs-varid'>group</span><span class='hs-layout'>)</span>
+<a name="line-122"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Output</span> <span class='hs-layout'>(</span><span class='hs-varid'>unNodeName</span><span class='hs-layout'>)</span>
+<a name="line-123"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span>
+<a name="line-124"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span>
+<a name="line-125"></a>
+<a name="line-126"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>GenOps</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>CoreOps</span>
+<a name="line-127"></a>
+<a name="line-128"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Prelude</span> <span class='hs-layout'>(</span><span class='hs-varid'>abs</span><span class='hs-layout'>)</span>
+<a name="line-129"></a>
+<a name="line-130"></a><a name="instance%20Num%20(Tensor%20v%20a)"></a><span class='hs-comment'>-- TODO: Look into hs-boot refactoring to allow mutually recursive imports.</span>
+<a name="line-131"></a><a name="instance%20Num%20(Tensor%20v%20a)"></a><span class='hs-comment'>-- | Must be defined as an orphan because of the dependency order between Ops</span>
+<a name="line-132"></a><a name="instance%20Num%20(Tensor%20v%20a)"></a><span class='hs-comment'>-- and Tensor.</span>
+<a name="line-133"></a><a name="instance%20Num%20(Tensor%20v%20a)"></a><span class='hs-comment'>--</span>
+<a name="line-134"></a><a name="instance%20Num%20(Tensor%20v%20a)"></a><span class='hs-comment'>-- The indirect constraint "v ~ Value" helps disambiguate types, for example in</span>
+<a name="line-135"></a><a name="instance%20Num%20(Tensor%20v%20a)"></a><span class='hs-comment'>-- "neg 1 :: Tensor Value Float", it helps find the type of the subexpression</span>
+<a name="line-136"></a><a name="instance%20Num%20(Tensor%20v%20a)"></a><span class='hs-comment'>-- "1".</span>
+<a name="line-137"></a><a name="instance%20Num%20(Tensor%20v%20a)"></a><span class='hs-keyword'>instance</span> <span class='hs-layout'>(</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span>
+<a name="line-138"></a>         <span class='hs-layout'>,</span> <span class='hs-conid'>Num</span> <span class='hs-varid'>a</span>
+<a name="line-139"></a>         <span class='hs-layout'>,</span> <span class='hs-varid'>v</span> <span class='hs-keyglyph'>~</span> <span class='hs-conid'>Value</span>
+<a name="line-140"></a>         <span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span> <span class='hs-conid'>Double</span><span class='hs-layout'>,</span> <span class='hs-conid'>Float</span><span class='hs-layout'>,</span> <span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Int64</span>
+<a name="line-141"></a>                  <span class='hs-layout'>,</span> <span class='hs-conid'>Complex</span> <span class='hs-conid'>Float</span><span class='hs-layout'>,</span> <span class='hs-conid'>Complex</span> <span class='hs-conid'>Double</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Num</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-142"></a>    <span class='hs-layout'>(</span><span class='hs-varop'>+</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>add</span>
+<a name="line-143"></a>    <span class='hs-layout'>(</span><span class='hs-varop'>*</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>mul</span>
+<a name="line-144"></a>    <span class='hs-layout'>(</span><span class='hs-comment'>-</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>sub</span>
+<a name="line-145"></a>    <span class='hs-varid'>abs</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>abs</span>
+<a name="line-146"></a>    <span class='hs-varid'>fromInteger</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>scalar</span> <span class='hs-varop'>.</span> <span class='hs-varid'>fromInteger</span>
+<a name="line-147"></a>    <span class='hs-varid'>signum</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>sign</span>
+<a name="line-148"></a>    <span class='hs-varid'>negate</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>neg</span>
+<a name="line-149"></a>
+<a name="line-150"></a><a name="matTranspose"></a><span class='hs-definition'>matTranspose</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varid'>v</span> <span class='hs-varop'>.</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span>
+<a name="line-151"></a>             <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span>
+<a name="line-152"></a><span class='hs-definition'>matTranspose</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>flip</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>transpose</span> <span class='hs-layout'>(</span><span class='hs-varid'>vector</span> <span class='hs-keyglyph'>[</span><span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-num'>0</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int32</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-153"></a>
+<a name="line-154"></a><a name="variable"></a><span class='hs-comment'>-- | Create a new, uninitialized stateful Tensor of the given shape.</span>
+<a name="line-155"></a><span class='hs-definition'>variable</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varop'>.</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Shape</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Ref</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-156"></a><span class='hs-definition'>variable</span> <span class='hs-varid'>shape'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>buildOp</span> <span class='hs-varop'>$</span> <span class='hs-varid'>opDef</span> <span class='hs-str'>"Variable"</span>
+<a name="line-157"></a>                          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"shape"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>shape'</span>
+<a name="line-158"></a>                          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-159"></a>
+<a name="line-160"></a><a name="placeholder"></a><span class='hs-definition'>placeholder</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varop'>.</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Shape</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-161"></a><span class='hs-definition'>placeholder</span> <span class='hs-varid'>shape'</span> <span class='hs-keyglyph'>=</span>
+<a name="line-162"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-varop'>$</span> <span class='hs-varid'>opDef</span> <span class='hs-str'>"Placeholder"</span>
+<a name="line-163"></a>            <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-164"></a>            <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"shape"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>shape'</span>
+<a name="line-165"></a>
+<a name="line-166"></a><a name="assign"></a><span class='hs-comment'>-- Assign returns the input ref.</span>
+<a name="line-167"></a><span class='hs-definition'>assign</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varid'>v</span> <span class='hs-varop'>.</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span>
+<a name="line-168"></a>       <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Ref</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Ref</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-169"></a><span class='hs-definition'>assign</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>buildOp</span> <span class='hs-varop'>$</span> <span class='hs-varid'>opDef</span> <span class='hs-str'>"Assign"</span>
+<a name="line-170"></a>                      <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-171"></a>                      <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"use_locking"</span> <span class='hs-varop'>.~</span> <span class='hs-conid'>True</span>
+<a name="line-172"></a>
+<a name="line-173"></a><a name="initializedVariable"></a><span class='hs-comment'>-- | Creates a variable initialized to the given value.</span>
+<a name="line-174"></a><span class='hs-comment'>-- Initialization happens next time session runs.</span>
+<a name="line-175"></a><span class='hs-definition'>initializedVariable</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varop'>.</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span>
+<a name="line-176"></a>                    <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Ref</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-177"></a><span class='hs-definition'>initializedVariable</span> <span class='hs-varid'>initializer</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-178"></a>    <span class='hs-varid'>v</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>variable</span> <span class='hs-conid'>[]</span>  <span class='hs-comment'>-- The shape is not known initially.</span>
+<a name="line-179"></a>    <span class='hs-layout'>(</span><span class='hs-varid'>i</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Ref</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>&lt;-</span>
+<a name="line-180"></a>        <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"Assign"</span>
+<a name="line-181"></a>                 <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-182"></a>                 <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"use_locking"</span> <span class='hs-varop'>.~</span> <span class='hs-conid'>True</span>
+<a name="line-183"></a>                 <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"validate_shape"</span> <span class='hs-varop'>.~</span> <span class='hs-conid'>False</span>
+<a name="line-184"></a>                 <span class='hs-layout'>)</span>
+<a name="line-185"></a>        <span class='hs-varid'>v</span> <span class='hs-varid'>initializer</span>
+<a name="line-186"></a>    <span class='hs-varid'>addInitializer</span> <span class='hs-varop'>=&lt;&lt;</span> <span class='hs-varid'>group</span> <span class='hs-varid'>i</span>
+<a name="line-187"></a>    <span class='hs-varid'>return</span> <span class='hs-varid'>v</span>
+<a name="line-188"></a>
+<a name="line-189"></a><a name="zeroInitializedVariable"></a><span class='hs-comment'>-- | Creates a zero-initialized variable with the given shape.</span>
+<a name="line-190"></a><span class='hs-definition'>zeroInitializedVariable</span>
+<a name="line-191"></a>  <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span><span class='hs-layout'>,</span> <span class='hs-conid'>Num</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-192"></a>     <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>Shape</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span><span class='hs-varop'>.</span><span class='hs-conid'>Ref</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-193"></a><span class='hs-definition'>zeroInitializedVariable</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>initializedVariable</span> <span class='hs-varop'>.</span> <span class='hs-varid'>zeros</span>
+<a name="line-194"></a>
+<a name="line-195"></a><a name="save"></a><span class='hs-comment'>-- TODO: Support heterogeneous list of tensors.</span>
+<a name="line-196"></a><span class='hs-definition'>save</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varid'>v</span> <span class='hs-varop'>.</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span>
+<a name="line-197"></a>        <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>ByteString</span>     <span class='hs-comment'>-- ^ File path.</span>
+<a name="line-198"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span>  <span class='hs-comment'>-- ^ Tensors to save.</span>
+<a name="line-199"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-200"></a><span class='hs-definition'>save</span> <span class='hs-varid'>path</span> <span class='hs-varid'>xs</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-201"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>toByteStringTensor</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>scalar</span> <span class='hs-varop'>.</span> <span class='hs-varid'>encodeUtf8</span> <span class='hs-varop'>.</span> <span class='hs-varid'>unNodeName</span>
+<a name="line-202"></a>    <span class='hs-varid'>names</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>mapM</span> <span class='hs-layout'>(</span><span class='hs-varid'>fmap</span> <span class='hs-varid'>toByteStringTensor</span> <span class='hs-varop'>.</span> <span class='hs-varid'>renderNodeName</span><span class='hs-layout'>)</span> <span class='hs-varid'>xs</span>
+<a name="line-203"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>types</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>replicate</span> <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>xs</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-204"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>saveOp</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>buildOp</span> <span class='hs-varop'>$</span> <span class='hs-varid'>opDef</span> <span class='hs-str'>"Save"</span>
+<a name="line-205"></a>                         <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>types</span>
+<a name="line-206"></a>    <span class='hs-varid'>saveOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>scalar</span> <span class='hs-varid'>path</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>pack</span> <span class='hs-varid'>names</span><span class='hs-layout'>)</span> <span class='hs-varid'>xs</span>
+<a name="line-207"></a>
+<a name="line-208"></a><a name="restoreFromName"></a><span class='hs-comment'>-- | Restore a tensor's value from a checkpoint file.</span>
+<a name="line-209"></a><span class='hs-comment'>--</span>
+<a name="line-210"></a><span class='hs-comment'>-- This version allows restoring from a checkpoint file that uses a different</span>
+<a name="line-211"></a><span class='hs-comment'>-- tensor name than the variable.</span>
+<a name="line-212"></a><span class='hs-definition'>restoreFromName</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varop'>.</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span>
+<a name="line-213"></a>                <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>ByteString</span>    <span class='hs-comment'>-- ^ File path.</span>
+<a name="line-214"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ByteString</span>    <span class='hs-comment'>-- ^ Tensor name override.</span>
+<a name="line-215"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Ref</span> <span class='hs-varid'>a</span>  <span class='hs-comment'>-- ^ Tensor to restore.</span>
+<a name="line-216"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-217"></a><span class='hs-definition'>restoreFromName</span> <span class='hs-varid'>path</span> <span class='hs-varid'>name</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-218"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>restoreOp</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>buildOp</span> <span class='hs-varop'>$</span> <span class='hs-varid'>opDef</span> <span class='hs-str'>"Restore"</span>
+<a name="line-219"></a>                            <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dt"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-220"></a>    <span class='hs-varid'>group</span> <span class='hs-varop'>=&lt;&lt;</span> <span class='hs-varid'>assign</span> <span class='hs-varid'>x</span> <span class='hs-layout'>(</span><span class='hs-varid'>restoreOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>scalar</span> <span class='hs-varid'>path</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>scalar</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-221"></a>
+<a name="line-222"></a><a name="restore"></a><span class='hs-comment'>-- | Restore a tensor's value from a checkpoint file.</span>
+<a name="line-223"></a><span class='hs-definition'>restore</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varop'>.</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span>
+<a name="line-224"></a>        <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>ByteString</span>    <span class='hs-comment'>-- ^ File path.</span>
+<a name="line-225"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Ref</span> <span class='hs-varid'>a</span>  <span class='hs-comment'>-- ^ Tensor to restore.</span>
+<a name="line-226"></a>        <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-227"></a><span class='hs-definition'>restore</span> <span class='hs-varid'>path</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-228"></a>    <span class='hs-varid'>name</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>encodeUtf8</span> <span class='hs-varop'>.</span> <span class='hs-varid'>unNodeName</span> <span class='hs-varop'>&lt;$&gt;</span> <span class='hs-varid'>renderNodeName</span> <span class='hs-varid'>x</span>
+<a name="line-229"></a>    <span class='hs-varid'>restoreFromName</span> <span class='hs-varid'>path</span> <span class='hs-varid'>name</span> <span class='hs-varid'>x</span>
+<a name="line-230"></a>
+<a name="line-231"></a><a name="constant"></a><span class='hs-comment'>-- | Create a constant tensor.</span>
+<a name="line-232"></a><span class='hs-comment'>--</span>
+<a name="line-233"></a><span class='hs-comment'>-- The values should be in row major order, e.g.,</span>
+<a name="line-234"></a><span class='hs-comment'>--</span>
+<a name="line-235"></a><span class='hs-comment'>--   element 0:   index (0, ..., 0)</span>
+<a name="line-236"></a><span class='hs-comment'>--   element 1:   index (0, ..., 1)</span>
+<a name="line-237"></a><span class='hs-comment'>--   ...</span>
+<a name="line-238"></a><span class='hs-definition'>constant</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varop'>.</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Shape</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span>
+<a name="line-239"></a><span class='hs-definition'>constant</span> <span class='hs-layout'>(</span><span class='hs-conid'>Shape</span> <span class='hs-varid'>shape'</span><span class='hs-layout'>)</span> <span class='hs-varid'>values</span>
+<a name="line-240"></a>    <span class='hs-keyglyph'>|</span> <span class='hs-varid'>invalidLength</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>error</span> <span class='hs-varid'>invalidLengthMsg</span>
+<a name="line-241"></a>    <span class='hs-keyglyph'>|</span> <span class='hs-varid'>otherwise</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>buildOp</span> <span class='hs-varop'>$</span> <span class='hs-varid'>opDef</span> <span class='hs-str'>"Const"</span>
+<a name="line-242"></a>                          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"value"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>typedNode</span>
+<a name="line-243"></a>                          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>nodeType</span>
+<a name="line-244"></a>  <span class='hs-keyword'>where</span>
+<a name="line-245"></a>    <span class='hs-varid'>invalidLength</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>product</span> <span class='hs-varid'>shape'</span> <span class='hs-varop'>/=</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>values</span><span class='hs-layout'>)</span>
+<a name="line-246"></a>    <span class='hs-varid'>invalidLengthMsg</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>printf</span> <span class='hs-str'>"invalid tensor length: expected %d got %d"</span>
+<a name="line-247"></a>                              <span class='hs-layout'>(</span><span class='hs-varid'>product</span> <span class='hs-varid'>shape'</span><span class='hs-layout'>)</span>
+<a name="line-248"></a>                              <span class='hs-layout'>(</span><span class='hs-varid'>length</span> <span class='hs-varid'>values</span><span class='hs-layout'>)</span>
+<a name="line-249"></a>    <span class='hs-varid'>nodeType</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-250"></a>    <span class='hs-varid'>typedNode</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-251"></a>    <span class='hs-varid'>typedNode</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>def</span>
+<a name="line-252"></a>                <span class='hs-varop'>&amp;</span> <span class='hs-varid'>dtype</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>nodeType</span>
+<a name="line-253"></a>                <span class='hs-varop'>&amp;</span> <span class='hs-varid'>tensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-varid'>dim</span> <span class='hs-varop'>.~</span>
+<a name="line-254"></a>                      <span class='hs-keyglyph'>[</span><span class='hs-varid'>def</span> <span class='hs-varop'>&amp;</span> <span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-varid'>size</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>|</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>shape'</span><span class='hs-keyglyph'>]</span>
+<a name="line-255"></a>                <span class='hs-varop'>&amp;</span> <span class='hs-varid'>tensorVal</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>values</span>
+<a name="line-256"></a>
+<a name="line-257"></a><a name="vector"></a><span class='hs-comment'>-- | Create a constant vector.</span>
+<a name="line-258"></a><span class='hs-definition'>vector</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>a</span><span class='hs-keyglyph'>]</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span>
+<a name="line-259"></a><span class='hs-definition'>vector</span> <span class='hs-varid'>xs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>constant</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>fromIntegral</span> <span class='hs-varop'>$</span> <span class='hs-varid'>length</span> <span class='hs-varid'>xs</span><span class='hs-keyglyph'>]</span> <span class='hs-varid'>xs</span>
+<a name="line-260"></a>
+<a name="line-261"></a><a name="scalar"></a><span class='hs-comment'>-- | Create a constant scalar.</span>
+<a name="line-262"></a><span class='hs-definition'>scalar</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varop'>.</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-varid'>a</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span>
+<a name="line-263"></a><span class='hs-definition'>scalar</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>constant</span> <span class='hs-conid'>[]</span> <span class='hs-keyglyph'>[</span><span class='hs-varid'>x</span><span class='hs-keyglyph'>]</span>
+<a name="line-264"></a>
+<a name="line-265"></a><a name="truncatedNormal"></a><span class='hs-comment'>-- Random tensor from the unit normal distribution with bounded values.</span>
+<a name="line-266"></a><span class='hs-definition'>truncatedNormal</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varid'>v</span> <span class='hs-varop'>.</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span>
+<a name="line-267"></a>                <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v</span> <span class='hs-conid'>Int64</span>  <span class='hs-comment'>-- ^ Shape.</span>
+<a name="line-268"></a>                <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-269"></a><span class='hs-definition'>truncatedNormal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>buildOp</span> <span class='hs-varop'>$</span> <span class='hs-varid'>opDef</span> <span class='hs-str'>"TruncatedNormal"</span>
+<a name="line-270"></a>                          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"dtype"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-271"></a>                          <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"T"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-272"></a>
+<a name="line-273"></a><a name="zeros"></a><span class='hs-definition'>zeros</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>Num</span> <span class='hs-varid'>a</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Shape</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>a</span>
+<a name="line-274"></a><span class='hs-definition'>zeros</span> <span class='hs-layout'>(</span><span class='hs-conid'>Shape</span> <span class='hs-varid'>shape'</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>fill</span> <span class='hs-layout'>(</span><span class='hs-varid'>vector</span> <span class='hs-varop'>$</span> <span class='hs-varid'>map</span> <span class='hs-varid'>fromIntegral</span> <span class='hs-varid'>shape'</span><span class='hs-layout'>)</span> <span class='hs-layout'>(</span><span class='hs-varid'>scalar</span> <span class='hs-num'>0</span><span class='hs-layout'>)</span>
+<a name="line-275"></a>
+<a name="line-276"></a><a name="shape"></a><span class='hs-definition'>shape</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Int32</span>
+<a name="line-277"></a><span class='hs-definition'>shape</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>shape</span>
+<a name="line-278"></a>
+<a name="line-279"></a><a name="expandDims"></a><span class='hs-definition'>expandDims</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>t</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-conid'>Int32</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-varid'>t</span>
+<a name="line-280"></a><span class='hs-definition'>expandDims</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>expandDims</span>
+<a name="line-281"></a>
+<a name="line-282"></a><a name="reducedShape"></a><span class='hs-comment'>-- | Helper function for reduction ops (translation of math_ops.reduced_shape).</span>
+<a name="line-283"></a><span class='hs-definition'>reducedShape</span> <span class='hs-keyglyph'>::</span> <span class='hs-layout'>(</span><span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span> <span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Int64</span> <span class='hs-keyglyph'>]</span> <span class='hs-varid'>t1</span><span class='hs-layout'>,</span> <span class='hs-conid'>OneOf</span> <span class='hs-chr'>'</span><span class='hs-keyglyph'>[</span> <span class='hs-conid'>Int32</span><span class='hs-layout'>,</span> <span class='hs-conid'>Int64</span> <span class='hs-keyglyph'>]</span> <span class='hs-varid'>t2</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-284"></a>                <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>t1</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>t2</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Int32</span>
+<a name="line-285"></a><span class='hs-definition'>reducedShape</span> <span class='hs-varid'>inputShape</span> <span class='hs-varid'>axes</span> <span class='hs-keyglyph'>=</span>
+<a name="line-286"></a>    <span class='hs-keyword'>let</span> <span class='hs-varid'>inputShape32</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>toInt32</span> <span class='hs-varid'>inputShape</span>         <span class='hs-comment'>-- [2, 3, 5, 7]</span>
+<a name="line-287"></a>        <span class='hs-varid'>axes32</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>toInt32</span> <span class='hs-varid'>axes</span>                     <span class='hs-comment'>-- [1, 2]</span>
+<a name="line-288"></a>        <span class='hs-varid'>toInt32</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>cast</span> <span class='hs-varid'>x</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Value</span> <span class='hs-conid'>Int32</span>
+<a name="line-289"></a>        <span class='hs-varid'>inputRank</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>size</span> <span class='hs-varid'>inputShape32</span>     <span class='hs-comment'>-- 4</span>
+<a name="line-290"></a>        <span class='hs-varid'>axesMod</span> <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-varid'>axes32</span> <span class='hs-varop'>+</span> <span class='hs-varid'>inputRank</span><span class='hs-layout'>)</span> <span class='hs-varop'>`</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>mod</span><span class='hs-varop'>`</span> <span class='hs-varid'>inputRank</span>
+<a name="line-291"></a>        <span class='hs-varid'>axesShape</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>shape</span> <span class='hs-varid'>axesMod</span>                 <span class='hs-comment'>-- [2]</span>
+<a name="line-292"></a>    <span class='hs-keyword'>in</span> <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>dynamicStitch</span>                      <span class='hs-comment'>-- [2, 1, 1, 7]</span>
+<a name="line-293"></a>         <span class='hs-keyglyph'>[</span><span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>range</span> <span class='hs-num'>0</span> <span class='hs-varid'>inputRank</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span>            <span class='hs-comment'>-- [0, 1, 2, 3]</span>
+<a name="line-294"></a>           <span class='hs-varid'>axesMod</span><span class='hs-keyglyph'>]</span>                               <span class='hs-comment'>-- [1, 2]</span>
+<a name="line-295"></a>         <span class='hs-keyglyph'>[</span><span class='hs-varid'>inputShape32</span><span class='hs-layout'>,</span>                           <span class='hs-comment'>-- [2, 3, 5, 7]</span>
+<a name="line-296"></a>           <span class='hs-conid'>CoreOps</span><span class='hs-varop'>.</span><span class='hs-varid'>fill</span> <span class='hs-varid'>axesShape</span> <span class='hs-num'>1</span><span class='hs-keyglyph'>]</span>              <span class='hs-comment'>-- [1, 1]</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/src/hscolour.css b/docs/haddock/tensorflow-ops-0.1.0.0/src/hscolour.css
new file mode 100644
index 0000000..c15919e
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/src/hscolour.css
@@ -0,0 +1,5 @@
+.hs-keyglyph, .hs-layout {color: red;}
+.hs-keyword {color: blue;}
+.hs-comment, .hs-comment a {color: green;}
+.hs-str, .hs-chr {color: teal;}
+.hs-keyword, .hs-conid, .hs-varid, .hs-conop, .hs-varop, .hs-num, .hs-cpp, .hs-sel, .hs-definition {}
diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/synopsis.png b/docs/haddock/tensorflow-ops-0.1.0.0/synopsis.png
new file mode 100644
index 0000000000000000000000000000000000000000..85fb86ec84907bcc86531dc82871948ff4d471fa
GIT binary patch
literal 11327
zcmV-FEWp!=P)<h;3K|Lk000e1NJLTq004jh003wR1^@s6IVt!g000+IX+uL$Nkc;*
zP;zf(X>4Tx0C)k_S!GyNTeqHT_l8Y(cXyX`gGi?cY`Qxn1VID|MJXwjPC)?)F$h6K
zMMOd+6hs7sqbPzXbr*U(-*=zy-hcPcUC*=TdiNM(jyd-lv&OpsU|J&v2m2!^0SE{T
z54F(O;<U4O;KWV=44?vZKnREv-NJ%3&CHI1|Ml9M1kj;!OzhCt|7!L>E2!K(!rTCW
z%wV;vdzf1QjBf#e&~gh74F>?Z4a=WLg$KhJ^$5nap>PLbJadS>e&h8+?D`9%QNL`g
zEVKbYGXj7k5Q(8)0Fd#*a?VIMFW3*64geVHKzE-&0BG!BtmfuTbO(T`0Jaeg2nagF
z{V*1E{Wm{e|AvV~*MEExiC+KU-~R=!2{)|c6Bg`GjQ;iG|FQ`1kAUCTuZtQk34#8{
z4r4<P-^=bq9b@>(3g7#|{=Z@d+d#}7f!3C=>=26vx*jwA8>@MS>RG@Tt_zt3hie^T
z_?<zX!Qp0Bf6L=zytR+Pm;lRjJ;HSV_$l=z8Xd-r9G16)Mp#(Gm>0%9VUd=)Fos7I
z^ghPh%Jy%YZ|)vCf6EaFPai$Q-!=$ppK!y&wrJs)bNdAuANB!m3n34Tfj{s75g-&U
z1A!Pg3bcXF-=!Gv1VmU93G2duANT;{0JugFTqg*|oPXPC|A$2HS3NJd-hcPV3EW`Y
zh=1Dr-5Mv{<{zIvz#Yba<V3iQL)~HT!km4Q{?+Cmi$j0EdL!8&xgxpyr%xz;lr%~X
zrG-*OsiBkrCyE}$48%~fC`FVeN*$#F>y&^Vcn^E_`qRfl{{bzYkp)4~$~NAx_VB;E
z{?P)PU)DbV{Qi#~0H0@T9czDj06@6MNq8OrpdAz(9qQxd9nPr<&s+~tPQySqaZyfb
zNh!%g_5YjeaLxMN*$s<dYY{|0@-La=c>v_p;d%b#U$Wpz0Geb0U>E+EOsEQ;I<Re%
zqFY30<Zpkd!v`9W019}7G68nL0|epsEDhv=3eW($zzCQED_{?t;PK-PkHJt71>!&=
zNC6q(BFF<p;2Nj^wV)BSfDUjM^n!=r33v`Bz%-Zxi(nP3gDtQFK@b)qgJ>Wohy&t-
zL?CHM5mJM6p`(xmWDmJOUQi$u0mVUQpbRJ*DuT+OI;a`C4fR4p&?xj8nuk`Puh35f
z5<x)FBG?f82nmEDLK9((utJ<bcp*X%F^FWu1w;X&9C01dj_5@^K};a#5UYqyBtViP
z>5*JWF{C0=8)=GkKzbrWk@3iMWInPS*@Wyu4kE{pbI3L14-^JPgW^Pq!Q<2bWsPz}
zg`nb5nW!REEvg;Wj~YYGqt;RTXfiY_S_G|(HbmQ@z0gtU6m&ki8r_B-Ku@3-(OVb{
zh8`n;QNS2r>@mKWSWG773g!l;2Q!LUz-(f%SSG9pRuyZCC1S&|DcC~nb!<2G1$Gg;
zjU&Zz;G}VSI0sxHE(w>9tH<5Py}&KucJP#VKD;vC6z`6Y#%JLx@m=^4{33pbgo;Ff
zM3uyf#Fr$Iq=2M}WPoIbWP_BHl$%tE)ST3Z^fYM!=}po{r1PXd2-E~&f;PdC5J9*=
zs3G(aUK2LR$jJD~G{_vt!pSa>)sa0QdqcKOPD3tEZb<G<o<Lqq-cJ64{1XM9f{#Lz
z!ige=BA=p_VuWIi5>LrbsZB|wjHfK7yiNI%a+8XNN{Y&qDu61Js-9|yYMB~K%}=dM
z?M|IcT|xbTdVvN>!$YG@<3@9arjllWW|0;{D?n>V>r0zK+erJ2cAbuzPL|Gw?j&6?
z-95TFdL%tRy&=6neHMKS{UrTQ1~vvw1`mcbh9-s=4Br`97&RC@7}FVVFitT3Wa4Df
zW%6UX#MHqw%Zy?cW;SPzV!p~ez`Vvn%c8>K#*)s`!ZO8*U=?PyV2x$1V13HE$;Qs6
z&lb#9$o7D3jh&udgWZ=sm;FBb3I`2`8ix-@E=M=VM@~9UO-_H#0?vNUbuLye1Fi_J
zGOlM_JKO@?*4#<lH@IhcNO%-^Jb7|?`gqoPIe1NY<9Hi*r}*%EihSODm-z<yzVi$4
z+w!OKcknL@FbWt7#0WG9%m|VRstbk)mJ5yxVT2Tg{DiIuJr_m@%L#i67YmPwAVlOv
zd_=B@jEbT~l|%zY%SB&`5yUjbBE=fS=Edp7kBTRWcZh$I;FWNYxFj(su`ekn=`UF+
zIW0vkWh9j()g|>+T3Fgmx>$N#hD=6JCPAiC=8LR|tcUDX*;jHjawc-Aa(!}p@(S{y
z@=fw93cLy~3MC3J6=@aC6f+ecDWR3LloFKgD*aHFR}NQhQU0tVrsAhkud;kZ;E2bO
z$|DP^+^R&?GSxXXPBj;`QnfjCE_I@Mx%xW|9u0SmYK<jLAx&S+>zbdmB(*}d+O)oF
zD{G(9?$JT&=<B5GJkzDtwbm`teWS;%=cU)6x2`XzAFJPEfHp8TxMc9skj>D|u+DJZ
zNWtioQNJ<4*wVPj_}x+AqoGH;Ob{kUCOIZE$M}u~9_ug#riP|Drn6=OW+7&G%rWL>
z=Ede8ETk;rECwxUES)XuEw`++tg@`8tp%+ktov*zY#eRsY`)v-*k;?#*-6-)vU_6B
zZ0}>=>40^xaj16KJg$2@@A#sloMVdP<OzWjCr=DHF+2G?-E*dJCOWsepj~WT>Ron;
zro?jMrmLZAiR-$Xw%cX5Rd)^dT=x|ZRgY|sB~Mk)Y|mvcRj(Yc6>oL#eD5_MZJ#2a
zFTMu8*L=VGnflfE9r)Y&-w413xCGn|qz?28><?lOIvF$?EE1d&JQt!Ik{7ZbY8+Y}
z28Epn>kOxb4~I`91S8Hy%txw47DsMJ*+jLTq&gXR@@ceibXxRMj9yGtEGpJ5wl9t=
zE-`NYl;)|jcqraAzAu3%Avt03wEpSZM3O|m#Ni~#r0k?`XKc@OC9@@;PF^^xf3_io
zJS8;cWvWW*wR5O*KIfjL$)pvg?Wen^KhBWM$j{i#bjy5vUg~_o`GX6d7oKIwXI;IB
zxfpnH@{;<c+H9)qnCyidlbrTk&fN6eZ+XsnL;14#*DjM>j<`HmaI~Pakhkz+;ck(4
z(L}LU@r@GJlC+ZVSKP0>xT<lr=^ER$jB7tjeM(=I>6f*a^OxsWU@9UjK2+LN4pu2v
z)m1ZBXH@Ui1lG*eTGaN}Db&@~v({%dAQ~bXR<1i<f8MCm*x4l3RCa^@M#fEWGvemQ
zW|!v47UPz_R>jt)TYR@l+GyI++oAU8_Vo_$j=4_z&e7XOxBI$Oy4voD->JFFb+`B)
z-My^)B=?i=A9TlbZ}tTDto3^JF7!F~O+T=EFy3$8|7^f`;L$_9hYtod2fH7sKDs-k
zJaqf9;^U4d@=w~I$~|oxmK$z+CjYE`L}8@!xzh8l(IcbxU#P$69n%?mIBq!pWa8Mw
z=%n@JtCx;1=U%zLT7K>S`pZ=0)Xwzj8T3s0Eahze8`d}FZ-w68n3JEoH?K4Q^qu9q
z=>@li)%RiVcNddCkbTHs;#jI%mR`QQqPOz=<MEG+tHG-~Ybk$G{Z;Zw@YAi&8lRu8
z+pRBt3H@^LHFJY$qvo6Jx4uoY%{N;CTRY#=x0$wUe<=PK{Av4hX(wtIyIZg)wAZz7
zygzdgaBvVz3?=?L3n1YV;pqv$*BUs7u!nPp=KzrO{LU$US-|NE1lz#a2I7HkB7Bib
zC_HKj-GeE>CgGy+9whdp4g`BLCvp!8U&;uov(!a2t+bEnRv6HXyi9t`-YglcEo`$K
zI8GTZXYLH1F5YE+b^&9-c%dfYc~N>X1MygiCdpZ8N*OKLV7W5+5rusvVP$KTgd_E;
zV`@J%*flk^Jhjj1)aX9cTQC5ItVZ(2W=FkE;*aH-)|+*kk6SET?pjmWaNEk+>D${o
z_#cmV%sNr-bj$gX%QW$m8{|&wA?SI;%go!uC))SCU%7vKz~jI-L0?1Ap^RZ7;i?hG
zB3+__P9{WW#uUa@#oavB8Q+`m==5;nXwvwZiR6j1<0+%5!{;8Q^`_s>XwIxTUvlAM
z)|rdpmprp=bM$iM@_6#8@((Vr7Q8HcP;{fXs3iGH;8nY8TBRaov}JqcixtC_ZB<vQ
z&(`?YTGlDob2eZb{<=Qi*ws{eBju)Fvqg(SE88vPt>w07?YBCLI#1vB=rX<|d6)j~
z?!9;SA9XkN<n~7QIX}?p7apJ-*nBuW*z>4rDD83J6N{$`!z{xG&lW}=KCd6md=WHe
zF)la3F!5t@`sLkMS6?Sg5vR3gcxTbGOK%>(y*_twKH{Cjg64anMViI^4{J-a%g<LH
ze(YYoyLRud-cLiHC*aZaV}tse#HQ(1*!QCCfgc+?yt~$WY5Tng2XOpFKtGg+ut(A%
z-=j*=VHi8C5{?;<#;=l0k&Y4u$@<72Q1np_Pz_U$(Jatz(CstOF!D30Fk7<(v!=0C
zv-fel=R$LH^XTz<^JVZi3XBSF3Nwow5p@<z60edNko+Kxkr9;Dk#m)gS143$Q5sTy
zqw-A^tHz`*q@kp#r)8miT*qBEKrc)`${^M-!6?x<@o1vS>0=3|@n*5+(H4=G;Z`Bm
z0XDw2UUnY#t`5ZG&W<NeoN#h<c62#GbaHieBf7hKczXJJ`FjWZMEFMgo$^l#ND0gg
zx)_`nQW$zAtSr1TqB^oJs{Uj{bVE#iY<*nqsjB$0ge#{D6LXTzpE;Wxe>ObDFO_)C
zCe0{aEki1k_dNXt+=U-mA1_W_8p^(%Q<IyQmz*DR*{Q&|P@#yYn4);6Wa-M-)$VKc
zrTJw^<pCA;mAX|@)oe8+HQ#IB*NxV9H#A-^X-sd5zTtDzzS*!vsa50_Ya2Pdf^Bpx
zbxz-Y-ZgON&fVsFb@$7<i+XZ;v-&a~r1qx_oPC%wnEEJVDC=?lldDf_hFhL>j|@Mb
z9sM+h7-yIepVWIvd=>Y)XzKR#)XeT1j<?!#-t!mU-CCGlggyu?nJkB_<bCW|eYv*z
ziR!cPy2cmFuWlPb-=a6;w@!ag*pB@X{?m8o#IEU{#=iIg)35lmfin;%)QHeRj3K>H
zI8-@&65hs?W6g0$Tn9b?K9MevmJ{6JljSOT6GbGYHWfM5G<6M41g#z&E8Qx6H$yI?
z50eHn6Z1ODBi1suSavH8F-{EUJXaTYHjh8AJ|73)7XPq7g<!OhsW6Z5mdKE3k(jTz
zvII_ILb6OMKw41-$PCI}kh79!lAl#5RdiS4Q~ID>t>OirQ5IDz)!g7S$y<#pnvPn`
zTCcP(>sag3>W=B<=vx}l7>pa{8`&AN7|$LpGx0noeC)GnyV)so9SefRgyl6WA8Q%w
zeVfO&`F8I1(hk7k+3~B6fhW|RD4pIpx4EPekGo2^q1>k2n?25Xx_BviQ+coYJoGK~
zi}SY&kPV~?{2VkK+z^r;>Jw%VE)ao<Sd8qCsyuln+BHTamLqmAZW>-y@)AN%A4?QY
z!X(X~xtpASHaNvFl_z!g+(cSqdP;^<ru=#G^Gg@HvkESTTr$e$&EC&>mD`$^mG5`i
zpn$&+Rk%>pUtCp^dd2Um*){o6wlZ|t=klqF!OHfk>gs};%-W>7nEHr@(CeX%5lwM7
zQg7xp*S7SwzHLLbOLn+*Uc0?`NAB*$d)wWCJsW)~{h|X4gV%@BpPU*_8L1qd8t0!(
zdySmVd!st{bK%K{=9Rj&=Ffv)KX1|hFxkC)82{hg(&3(fkq6-NB>?O?0kGBtAd?QJ
zm0$~|LIBLj0I*U5i1iA9XzK$|?dCuG2lOlFq=GX}9v}f{nuc(O=>uZH1yBw;!3bD_
zU{(i`gLA_m=mOLPjX+-zbO<GcGvX|w5itp8bfQQ*WD2qkUVrG}Of3%84DUniXj}9p
z^ivET&bv}EkFcb0hLww*!SUjPaCh){ydAy{ze{3HQcDVv9w+S}&=VpFugO%&D#;1t
zQRE9078DOC<tZDexT%V%>8W#QsA+O&>1m7Uxak_`<>>nu%o*kx!<ev4CCswS!z|vc
zXx2(L9kvhb=^T<AuQ(IA#JHxo(|J^QKJ!-bo#1ES9~a0Gv=U+xdMjKf5+SN1MlZG|
z-Y;=YGG59>T2DqomQ{`*59GHMHWa@qZ7S~^!Kl)z@vEz7SZjuAWovinywxMoS2FN7
zEH|1t%4A}H?2754xrD_j%Moi{n>gE7_6iP##}7_;J59Lg5Ifz(-D^B~y{dc!eQ)?H
z1`GsQ2d{)Cgfm98MOmHv9&;s5@6?xs(nO0hxa6LcxN|CLdl`M_GqP+i31t7w9nHU9
zkY40hVt!S*RG^%pl2DDR1@+)Ms)_U_Lks^c#r9*J-d)LeEAIFAEIl9{kQ}rbihXiz
zxOZfJbZ?wtQtXx5l+ld&8>=~scSi5kK8P(dtn9DO{nh=s_)Emb(M`^+uiKA)7VrA)
zEB#tO5ODlSVZM$P@WWh#2Fx+Iz|6u~m`%6|24UXdCqxG`1g0=2kOkd@#-Q&AR(P%P
zMdTpvAy(jBM;jT2tUyk{D~~EF3{{U>K(nFk;T(JdLx-`&6l3PF0@xsI7Y>87!d2q7
z@J9GD{0|aKlAELyq`{<P1SLWl8G$UCY?<7Fe1t-eqMK5MvWrTEsvFK%hiUBK+%$p?
zLsvvEML)#g&IlN<F{v}XV@_oeW0_<<%_hn=1;_jd$2w;fmm@bl_bAT=UK2hFz8Cxj
z0?vX$f}2A9!g(UzqH1CcVqe9dOEgR7NySNf$XLi~$;ru!DDWzBDsd|Fs0baAQPog0
zQ+L;h)y&gs)gIT`*5lMSHi$92ZZv<C*~A7u*Ss=kxA3yOWsS12wQaP=IJh3~J;CFY
z>in5#@A}YP&ZEYQ#XH-V)Gsvv6_^~14ao?j4lj=6k7|w9iW!UZJhhvUlPHq(FxfQ)
zq?V>>q`%8dxgeZ1aw#H*HTOZjUjc35y<*QR6jwV-iRB~}tyPXS=-S45n}+?ysv9OZ
zzqJ(K(rR1j$hs}xHG4PtzG(M&@2Lj@{VyISJQ5#z^W@U7{<DMUcV5Jft50HIPQ0$2
zPMURnYc#L4Ah{^9q_ASN>hV|l=i6Vte3RLV-yYuK+dKCw{z!laG%#N$3ABJM%p<0O
zYA^skKqQbP%m$r-WBwLFh0ujLomRwONMWQ8vL5*f<`CmhgJ?Rm2f718hVj63W7)9r
z*mpQXTq~XnpG|@xNg&xFjU_!Gq>|CVvs#J#1w}9=HDxE2J2egUAWZ`85!yYvKKcv>
zJ4PYKJ*G+KW|m8=VQlv7TJY|}%0<qN;Qq<;jdz`Io&T%Amf)T+R)kiRM@&}SK*Cw_
zq*ShSi_DnpraX&+wqk%%q4J>0wyKDli~41a=UN19Bb{{JVSQ=?d&3H&&qviwE*<+|
zre<Dfv2EpG-EAvqm*;>!9^?4cDF}{Txa*#Kx+jZQvyZXwvVVG@WYFu7)G)>HwaCho
zPBE;pGpDX4cqED@Z6)`nTsY^LE}F4-ek7|Lj+#LpTmF}Vfuf?4z^j_2v}GSEI;v7@
ztn0YySFg7=Mcq_r{?^*qM(m*I?Cd&z=li|$-7G!jeOwO;25=992S<m5AICiD8%90T
z7zuw~Iy(4bWgIcVFe&g-@|E0cxhctM!5QXRlG$%>X5MzsmCeV$vtN*Wk9q%cvGzm6
zlGZYQ`Nc~9M~79`)tR-DzwAEIeH!_EZe4SI`^$~5?i-97Prt=)N^Q<<UHq>3ePg@o
zht*Hi&(<B6UGm+oJ@<X`{k8*#!}y1J%19sX9snX}!Sv&sgM$rz0I=Br>|HuI*eO3a
z*sFk(4fq>KkN@xQ6^F(cm~$_2K14li9;XkV|9<@!M&f%8Nam8p00009a7bBm000XU
z000XU0RWnu7ytkil}SWFRCodHT?u#;Rkr@KbUNvfeG_5`YY-wNfPp{+o{ADgGcxep
z5O;8ydCWk3pWowCbe1RjK4lzy;4&jKqk}U-a1=+ud7z@;LLwlFC>S)v1jwFrI_XY2
zop;WyuIf%_F<t2n-Ic!IS6#QNZr{59Ip>~x?x|CCgE~7q5lBOq0>MKUdH^|7ARquk
zTn+*P5D<VNPJtl~KcE2u0x-lWFvQ`9_7{IpP!Ma2Ag8UrfAvN=8&%31Xc{{x^H{Co
z5Bf3s5%3VAbLh&IH$@c|uDl{TIw}*k&TyNplZ9bWQ*bafH>lHMG@8ELxbaVWHf?&T
znHpfF&E_pZ&^rD;1;7qozi0Q$(`V)7{8<+kI>wdbHk%E>!9AN2eO+^{$KB)hHtVU6
z4;0@%KYw`%{kM%aj|)L>`1``u*EM%B_Ep|f_<QCY-I>7iHT~t6&rZsneaT;XVt##n
z3*O&%0=#!k4Gq$@x_XoAC663)d$?Wm=U<vc73yj8s7C4ye?8%O?|sJH9hZdp8`Rda
zV)AbC6#gCsfWR{`**=IUEL^cLOs^a53WvcKmwX4axn1m?g`lRU8J`}Mu9cL0kEW(p
z*S~w@nA^2_^<7tUJ89F|+dbgAd&6H>XTrha?_sgD)BZa!4dhf)W5g$)o+5f!@!6p=
z7>#E6lGpa0z~7?)*juclePn!mT$U>W2F?VqT7?}(LqHHhL#3+DoNXk5_#Pb{(lwSP
zZ<=X|iSbjYeFoatR`H}3=!RdX3qeSTbc>FTPC<i4yKox}dW8l|7rAq|($3Sz1I6E)
z0PuMxfT6vuJ<7S!uFMY$3&G48V-OV;j??9JIDGgtTjt^~;o&;W${mB481A6nNfvWk
zG}X)k!0qh~kGnVT5kF&`L0DN3I5DFSlBQz~XE21}l1oOBfJjYXO~BgEYVC0M-fop2
zDE{6E#$wJW!`duuVRDQ5Vt#39HNO0^LN1EPNWEbGuwmiKId5c0K$-Du1N3`J+mCki
zh@bx#O+t{9<6v}qdxy{j&CM3X#uzYdYMN4g6Ac50s6((}=&dwR{Cx?4=mpk}4k7Lh
zja}8gbI(8i^%;C~;;e!(COSeeKO#br9p##*Gxqi|^SQmA8UnbUF8H0>&5WKoW3<xS
z+uAzF<~aqE$%3S$Xk2`8s#2WoA?J>vT<}n4p!j<e0B)WjD5-NnRG#WdwxQH=@&k}$
zOR8R0dom`Lc=?)q&RcuB7q?Ti)x-Dbg5R}n#wZVUe2wiO-Z*u-^!Q-=uplma_-CN_
z8Dw8c=~1luSn8d}KeqH68(WbvJ_VB|4Oi+hk~;kO6RAfiC;c;{;CF65PXUWd$`m2j
z>ve)Qtntp05&Y$`N~L&mauhNrjZlt#E%Rdnz*4RdA(~WsS0P~4Cker*^h9K3rID79
zAhx!)2_f*-6tD+E@|~5o_HbR*DQEm#fix64W;xPOIEsuwz3>ej`Mg}wlx+M?%^s;7
zt7<_1|D+24j|zb6{d*Duo)R*nQ%A&N`m}UK6}Gim#oV|jr-^I5{&3u6Y!z0&JjK=N
zf~iA{0UNr_&1RH*=FkdaRxmwXu@ih1pW6b!KwO1@&&hNB<m70YDveg=#B4Xn!v>f0
z=VYU~zns|bF>|Ig{pE8Oi&e4q8Sf>;d>$HnJ*g4^2E{@!BWJXj|MK2>t{)#4iCiKM
z_X3_Wd3!22SVWGECF_5t9Wx1ebd<xtk&rM9vu2GY%sRn*f5AMUxid3;-rS5H$2b;{
z-mVQ6!@O~ym}o<<xL6?O&7IIAJ-pEtPt>Ve1IRabo*K&M<b7Y)<TC!a?+aI7*L)G{
zfCa!8K?T~>e+mp(08G`jsI~A7O*rz=A?*I(Ym_y4*Z<vrc-}ABZsXaPWE_55X~M~}
zTIJaLcOUp224n-2zVikie7yn%YnEfm!5#SRe*5FmzkQB}p11+o{Fp|=>BHj<`2EIL
z@XCfeu<hfsczydWJ6bb}o?S)x0g-o5D5qWnbezFoiTRwunk%qdX1+L6i)|mA5jidY
z{0}jolhGWEzbnk=GvPS&`Ekzpd{Qp}jg2@(@Fp)^ip7aWa9kF~U5c$wF9JUgt8d{b
zE$t=?X?=dwwZ3ox^kn_>GtW8G6RGFlFM<@CjE-OtU#5a;0kB%yXw(N%<3n(~<Qsu8
zZ|+#=eP8Q&gdZK`A?MLQF(CmT9+sYjF8fKIN9rETN>sBeG(H{~)Y9EAyo%kT#Rg2j
zpdOnacnjrpoDswQL%S&=xD)LJZ^c?^7~tUKxVSW2U-+UJ`I8c2{Q|sd4FLUcTr-0M
zaqMa26wFKpz7U~s3AlNV<LRJWL`!?_{RP)vp+K58DGni_Hng{J{G%C}6UQspd^w|9
z18~65>^qhrHMbm9<`9gTLcVV_VCkY<jGCK;u}-q#%LEum+wQu00v693hqg8|E+F?e
zr#1I3!kny;lw%R`{p7X)SelGzum&R`J{*=hGlr!kVHo3VG@_P-B6~JTzO@CWjwnQm
zSR&sif;%k~Gk*Hv$KL#GDAqtA4UzOxf>cW$bp+1aV?*4j`n;5NQvl5P$NHC1)DVqF
ze?14Uta}S5dTDmrRR#Fn;tPAZ>c6M&cw`%zt17X5(`x+mXPZPMYENh$xHA{IIn#Q&
z<WuL|wgH?ccYQ_s*J*Sb=&abC^R8TNFm``&#I=ks9J{Fj!m;GPmSYjgeN40tkN)>^
zG}YF_5*3HIuofIEDMeLB1jc8M#;C+D(d52>)gx`#@~i9ZqkAV_+e~x*&R~QFvHtHw
zX=O8P?QIyJ9<S9QfZh~|rm@{l%9`SLom|1Yuu=8~?mQv6f;~cV->Ss9*B|&g;0hMp
z3Alm-uHb+xn7Ts16&!E{`__2XkJh+p1UhOAxPk+&;D9SQ;0g}7f`^<d*f9ujPKWd{
zNwA(ei(>~4p*<oBsnf?}Y_tY_z+HhWM{(ruZlLc|d{fqlC~^zWp3Kn@&XYx@s8428
z+W?HF@nkVpZ73FT0nX?mGjp-|=hGB_f%7#j<?C8Quwu<K-PpY#ZzNXSa*ba=b=wuJ
zprVklr=6G|NiR*v4kFXBPw?d4Dy67aGq*W0e|U1NJv0<}=fG*4{I=9DV5`;u*19?o
z5{B?pJoNC@$d2dNpxSZdpKs&0Z&je7$@!^)U(mv>Mp`Hum_uHM8Ep9TllPO>m-^Cs
zpVwg1bK6i`-w1z*2vDs7WXVaJJHyU=rk@Vk3#W^iKzdl}7D4^3u#E2B8*>%rGlt8u
z5=Bg)^vMF>N2OW-kTeo=C=#;#Uwg6hiz=At%UPznGuZL$9uX3jIcgXzEoL+}ne7De
zePX!NLIZ__1sfvp<d*$}QF`*n@c|qD`l$#Bn=o#aJJp@HvT{duzX6s;924PRO?YW@
z5nghs!8um&p{H&~w#Fia`1v1)e}&ROaMy!>aY5fTR<P*VpVM{qeXU#N9z3DGRUK^<
z`8ISD3xNMj_Il@&dJSMm!=vkNfHgFv_xE4u6KU4aaHMFg_cJ-U=WN?_Kka)L1Nw>(
zUH5HKQ7-^w@TCk-ATqS$+;^2Y-9Yg{p~En8>~LcE&~OCN2SO-y!qgT7qsff0kWR!$
z^D81!lBm$TfXL;}=Y9YJK+SF{!{d*<g%X#sAaChntiD3qVd2a){!IDu!avzV!TF4o
z9xGgkjB%soL!aKOw*koiw_`(h<}=y@MgPFg@A$O|ZKq#6{$@AkGvc*R?Si#UlQDr#
z3~i~w{&Resf;DUNF*}X?fEFV@D0-Vh!)meXqjD6C9^>=}ZDsk}pA}{0WdF3_)n|T5
zFNK7P(SF;zrP#jx9qieE2>F-K@p;gyHGt(@rI_!hEt)McpP}lbFn3v=a0JCAI=-Ld
z^HfmLKw}#PgVO)j-n&3BpR3@}{)WrPilHHGIK3w22T8R6=u<`rMwjnBh~jFy5zt}A
zN81hv!KkMXNNPDnh1mq7H@>uwma1@k3;2!wtQCOj+9tn%uigkWBw{AL|5)BofhX2&
zA+XZ302%fCsUzg9CimQPVv`f;C6O8|{n>ML#6sZcPqU_9<QKRdEq1%Pk+|_c^D$ZZ
zpTp?8<hlYZ8Sj`5DMd7vdH&JsA-;|ya5+f%=&CEBr`QO8Na{5Jql+JdN1k1QvXkYg
zs3w&k5rOovX-K7)MutMt)1dz*eGsClwG%b<Mnp!CZzRsYY@Y!MK;y}yczWw0oN9FR
ze!+NJMke*DMR;s+x~Aozt~D6zfBzb)s_VMoY^H}`58OE$vvRV#ocF~8^#ULQLiXc}
zUQkuE$PY|=DWHMVx2^eKu~LOA6rK$<8jY@}eezhf09Y%}*qLuY{&mxkmF%SY=ox@6
zF~uhVd#X`JdPSa>DPe!$!>g7c<pOq#ONp0~Ns93gcerW+2#u6Nz)abBxO;&=#J<Sv
z@ytFfgdg<)OXiP7*2GlLRC*q2ahR^|-aI}H3;8oRM~>oyleK6R!5=0O9Kit+4(r(6
ziv6QJ8-P(X4Sa3SakRGjFIv?a0G4_jZD3}d!^RD-cH>&cq5?d2jrKkeAp_;!Ur#;&
z9W7Y4e9epUX=T6m-g%gom8l&2YDT>Vpn#D2K2TLOYC9;D1)wkDRn>N#8T3J_^Lk0W
z2GEDo5^3Wxdgdfd9w7&WOIUcVywJ$#^9sz{H)rNATQUdN%*}+3f?}K#TL)6Cfb&`3
z%&Qjw3IaWJ_$1z;4dDsM&%YQ~=42pUgopbkSWmW!9lu+5e2Bl(Hp~!=)psw#l#5d7
z<59t4!9`Er%bRtn7l4p3WRMY9&31sf7Q0{HC$^-K>G(;07G_Pk5PmWfQbk{$>nD;C
z$aX+;iw(co_@<~Qn^p+B=a%_MiWA>XQ&sn1{z<(6(1#*dufHEF>#Fe8m!&8!F2%dw
zHlg}-8UFYJZG<8tdn)d^eHPNC3G-m$^7_440RBMV3*u1l6Q_-M<NrRRMWepLk*_Lo
z-`bbN{R`%2;?^Z|DbJ;SGxcT1^sB{@@yW=TI0_Rc#C0ow;>ckXuK!rmQ$k)#dR$sG
z@^U71!@qOSF|2)@pOpG;Qm+AE#NKTmpy<6aRJ-8I$ex7UR10>zRSMI&Dx4*+aC%oe
z$>ksZdHCl3@33X-u5M#~!F>8s>bP;(@Z1iZ5DQ57E<W`F&{Exn`H?9pD7b1W{*9JP
zOVBg7&1l$Lik;?q)RUhiU1Oo)Yy>(pe>^RmdH=2Rkv1Y;;r0f4a|kUQI?AO7tZbEf
zJ(*E203jiWBR5FKRnt*$=_L9l06hS)bRb+XpPQ(|6)W>G1u?i-W6WoCJgU<Q$tb+@
zD$Jffj@}LJ{)WH{zrK~CgV-P9o6Kh%ssy0+_)+}f)#J#erD6&w@S}9^J6Kv&%Joq)
zVCl;1aO2ciq)r?I|MikfXnOjsHU;7FXSU<j4Ywhj0zZnsHR6e<UbYvt;`zsJK|)+S
z62!-nogR|)RM~kx=eB4z9^a39Htt8sxmx;YG^zT!qj*qZKBH+t(F@!0k$*VC+$MjO
zDHQiTunA>lRkTWYJ9y;~2lKhQP~5|72z2_#^8q&npdI^OKWZnM4)jd~l<t#{4Y+Cf
zJmk?iep_h!z8~!6w6%M3%WgX$8wEf`;a??H_G2r4_wX%<Bi{h4{7*`b`2}z{1;B(K
z1YRBPTX!e&hj*a(pcMU@bm?3ynVpH1H&2p)w`(&X<ooZ-37-u`n^q8?b=dLG_XXa&
z|9vVMfYDf2c#}A8B~Ph8B=-cMr4A=W1b8$?C6oGZsl=!338XY{hCK?{ZdF8R@clrx
z<=e!HAY|AD>xFIKK%PKOm(9u+`!IG4P>PAtq9@Rh0<!HW6ucJiefk6H>JE!{0DuH!
zkK`y|6ZXDM&ju*fYcM2?dkd?0BQd?AvKl9=rI$l^%Bzo%82<d<-V^wnjLl;2>pwp_
z3!t@d`N^j}MPee&>2}gr!FRvB)4o^~UCPYDMfxiI>b@c+MsVI_ZG?n%#SdILF9)yD
z8iBv~&32h6$j=)^`5;_--)1F7aK==Pycf`JwRRcIa&EjD`NGhX@h9M+TM4YCmA;oJ
zrO3=nv3MeD1n(z%`&dZj&7(JU#eehVv~0XE^yJ%^arZ3+;^s6cinJi_LRv*8MlRsh
z{Xp^er2%-zvwii|iPQND<~cxwB;)S&_u$&{D%8_7aQMh%>8YP30yAe!z=De>;j*0J
zN>6b7(K|VAAJyy)=J$-BZpMp7n5{I{+sN@1<}jm{UYm<6az<l#QupT~-a!ux6n{`>
zC)2KLBDKeY!To$ha&qG2BZqfAotPNM^BbQ^H8u4$*;5z(vZ|_v=c1LgH4&aJ8cR)s
zhZ25=_;#ffO9d0sLd30K^&jiDoI6+3R|Htse-FYDw`bL=buUu;*yY6jR@v$9iMtOO
z{Jm)a77X@ba<Dr`c8-aSjUKyf*>%$<P8>f%7edh>l!!{woQDqvAyLn?wOiY*$B%zo
zv32X~pEWczvH$rLZ56cfy6vr`0a$epDA9d}4E`PkfT>4BU?%e$j!CrfB%e1P1~}M{
zuQ8DZRRHLI>|J6XE5CNbPoY`u^Tv~L_DESt0J@K9biv&;RPgs@1TwMtC4bqg&n_U&
z^RqpU@fmCZV8(Krcxd8Db|Y=v9v+%_sqO*ye5%7a4GH|cY5=AL^#T?U?(IAraOf}Z
znfd(s?_l?Sx}{(;kM%5!ES&ry9?r8?uz9NYQ(Ynr1^j&q08@d8z|&jaWMSaE-1`Sx
z2*lKk?$1KN8*2mJGw(g3`l+riN$dE3Q~;P7L<D9yC)d4ncJO{>Cd=wx?7hW&8J3pu
z_e%g|LIn2Oqk!C_wTCQ#s9zKa2tdEcq}@UR0njdQ`-LnZ0R1A9b_)drK)bx{7qWl=
z^ovZ|Eff#{?eex?$N~b;FEVMjP(T2*%iDe-`+v|7m{y$1dn*6{002ovPDHLkV1lnB
B5rhB$

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-ops-0.1.0.0/tensorflow-ops.txt b/docs/haddock/tensorflow-ops-0.1.0.0/tensorflow-ops.txt
new file mode 100644
index 0000000..c3c9ca3
--- /dev/null
+++ b/docs/haddock/tensorflow-ops-0.1.0.0/tensorflow-ops.txt
@@ -0,0 +1,378 @@
+-- Hoogle documentation, generated by Haddock
+-- See Hoogle, http://www.haskell.org/hoogle/
+
+
+-- | Friendly layer around TensorFlow bindings.
+--   
+--   Please see README.md
+@package tensorflow-ops
+@version 0.1.0.0
+
+
+-- | This module contains definitions for some built-in TensorFlow
+--   operations.
+--   
+--   Note that certain, "stateful" ops like <a>variable</a> and
+--   <a>assign</a> return a <a>Build</a> action (e.g., <tt>Build (Tensor
+--   Ref a)</tt> instead of a pure value; the returned <a>Tensor</a>s are
+--   always rendered in the current <a>Build</a> context. This approach
+--   helps us avoid problems with inlining or common subexpression
+--   elimination, by writing
+--   
+--   <pre>
+--   do
+--       v &lt;- variable []
+--       w &lt;- assign v 3
+--       render $ w * w
+--   </pre>
+--   
+--   instead of
+--   
+--   <pre>
+--   let
+--      v = variable []
+--      w = assign v 3
+--   in w * w
+--   </pre>
+--   
+--   since the latter could be reasonably transformed by the compiler into
+--   (or vice versa)
+--   
+--   <pre>
+--   let
+--      v = variable []
+--      w = assign v 3
+--      w' = assign v 3
+--   in w * w'
+--   </pre>
+--   
+--   Ops should return a <a>Build</a> action if their original
+--   <tt>OpDef</tt> marks them as stateful, or if they take any Refs as
+--   input. (This mirrors the rules that TensorFlow uses to avoid common
+--   subexpression elimination.)
+module TensorFlow.Ops
+
+-- | Returns x + y element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>Add</tt> supports broadcasting. <tt>AddN</tt> does not.
+--   More about broadcasting <a>here</a></li>
+--   </ul>
+add :: (TensorType t, OneOf ((:) * (Complex Double) ((:) * (Complex Float) ((:) * ByteString ((:) * Int16 ((:) * Int32 ((:) * Int64 ((:) * Int8 ((:) * Word16 ((:) * Word8 ((:) * Double ((:) * Float ([] *)))))))))))) t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Computes the absolute value of a tensor.
+--   
+--   Given a tensor <tt>x</tt>, this operation returns a tensor containing
+--   the absolute value of each element in <tt>x</tt>. For example, if x is
+--   an input element and y is an output element, this operation computes
+--   \(y = |x|\).
+abs :: (TensorType t, OneOf ((:) * Int32 ((:) * Int64 ((:) * Word16 ((:) * Double ((:) * Float ([] *)))))) t) => Tensor v1 t -> Tensor Value t
+
+-- | Add all input tensors element wise.
+addN :: (TensorType t, OneOf ((:) * (Complex Double) ((:) * (Complex Float) ((:) * Int16 ((:) * Int32 ((:) * Int64 ((:) * Int8 ((:) * Word16 ((:) * Word8 ((:) * Double ((:) * Float ([] *))))))))))) t) => [Tensor v1 t] -> Tensor Value t
+
+-- | Returns the index with the largest value across dimensions of a
+--   tensor.
+argMax :: (TensorType t, OneOf ((:) * (Complex Double) ((:) * (Complex Float) ((:) * Int16 ((:) * Int32 ((:) * Int64 ((:) * Int8 ((:) * Word16 ((:) * Word8 ((:) * Double ((:) * Float ([] *))))))))))) t, TensorType tidx, OneOf ((:) * Int32 ((:) * Int64 ([] *))) tidx) => Tensor v1 t -> Tensor v2 tidx -> Tensor Value Int64
+assign :: TensorType a => Tensor Ref a -> Tensor v a -> Build (Tensor Ref a)
+
+-- | Return the reduction indices for computing gradients of s0 op s1 with
+--   broadcast.
+--   
+--   This is typically used by gradient computations for a broadcasting
+--   operation.
+broadcastGradientArgs :: (TensorType t, OneOf ((:) * Int32 ((:) * Int64 ([] *))) t) => Tensor v1 t -> Tensor v2 t -> (Tensor Value t, Tensor Value t)
+
+-- | Cast x of type SrcT to y of DstT.
+cast :: (TensorType dstT, TensorType srcT) => Tensor v1 srcT -> Tensor Value dstT
+
+-- | Concatenates tensors along one dimension.
+concat :: TensorType t => Tensor v1 Int32 -> [Tensor v2 t] -> Tensor Value t
+
+-- | Create a constant tensor.
+--   
+--   The values should be in row major order, e.g.,
+--   
+--   element 0: index (0, ..., 0) element 1: index (0, ..., 1) ...
+constant :: TensorType a => Shape -> [a] -> Tensor Value a
+expandDims :: (TensorType t) => Tensor v1 t -> Tensor v2 Int32 -> Tensor Value t
+
+-- | Creates a variable initialized to the given value. Initialization
+--   happens next time session runs.
+initializedVariable :: TensorType a => Tensor Value a -> Build (Tensor Ref a)
+
+-- | Creates a zero-initialized variable with the given shape.
+zeroInitializedVariable :: (TensorType a, Num a) => Shape -> Build (Tensor Ref a)
+
+-- | Creates a tensor filled with a scalar value.
+--   
+--   This operation creates a tensor of shape <tt>dims</tt> and fills it
+--   with <a>value</a>.
+--   
+--   For example:
+--   
+--   ```prettyprint # Output tensor has shape [2, 3]. fill([2, 3], 9)
+--   ==&gt; [[9, 9, 9] [9, 9, 9]] ```
+fill :: TensorType t => Tensor v1 Int32 -> Tensor v2 t -> Tensor Value t
+
+-- | Multiply the matrix "a" by the matrix "b".
+--   
+--   The inputs must be two-dimensional matrices and the inner dimension of
+--   "a" (after being transposed if transpose_a is true) must match the
+--   outer dimension of "b" (after being transposed if transposed_b is
+--   true).
+--   
+--   <ul>
+--   <li>Note*: The default kernel implementation for MatMul on GPUs uses
+--   cublas.</li>
+--   </ul>
+matMul :: (TensorType t, OneOf ((:) * (Complex Double) ((:) * (Complex Float) ((:) * Int32 ((:) * Word16 ((:) * Double ((:) * Float ([] *))))))) t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+matTranspose :: TensorType a => Tensor v a -> Tensor Value a
+
+-- | Returns x * y element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>Mul</tt> supports broadcasting. More about broadcasting
+--   <a>here</a></li>
+--   </ul>
+mul :: (TensorType t, OneOf ((:) * (Complex Double) ((:) * (Complex Float) ((:) * Int16 ((:) * Int32 ((:) * Int64 ((:) * Int8 ((:) * Word16 ((:) * Word8 ((:) * Double ((:) * Float ([] *))))))))))) t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Computes numerical negative value element-wise.
+--   
+--   I.e., \(y = -x\).
+neg :: (TensorType t, OneOf ((:) * (Complex Double) ((:) * (Complex Float) ((:) * Int32 ((:) * Int64 ((:) * Word16 ((:) * Double ((:) * Float ([] *)))))))) t) => Tensor v1 t -> Tensor Value t
+
+-- | Packs a list of <tt>N</tt> rank-<tt>R</tt> tensors into one
+--   rank-`(R+1)` tensor.
+--   
+--   Packs the <tt>N</tt> tensors in <tt>values</tt> into a tensor with
+--   rank one higher than each tensor in <tt>values</tt>, by packing them
+--   along the <tt>axis</tt> dimension. Given a list of tensors of shape
+--   `(A, B, C)`;
+--   
+--   if `axis == 0` then the <tt>output</tt> tensor will have the shape
+--   `(N, A, B, C)`. if `axis == 1` then the <tt>output</tt> tensor will
+--   have the shape `(A, N, B, C)`. Etc.
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>x</tt> is [1, 4] # <tt>y</tt> is [2, 5] #
+--   <tt>z</tt> is [3, 6] pack([x, y, z]) =&gt; [[1, 4], [2, 5], [3, 6]] #
+--   Pack along first dim. pack([x, y, z], axis=1) =&gt; [[1, 2, 3], [4, 5,
+--   6]] ```
+--   
+--   This is the opposite of <a>unpack</a>.
+pack :: TensorType t => [Tensor v1 t] -> Tensor Value t
+placeholder :: TensorType a => Shape -> Build (Tensor Value a)
+
+-- | Creates a sequence of integers.
+--   
+--   This operation creates a sequence of integers that begins at
+--   <tt>start</tt> and extends by increments of <tt>delta</tt> up to but
+--   not including <tt>limit</tt>.
+--   
+--   For example:
+--   
+--   ``` # <tt>start</tt> is 3 # <tt>limit</tt> is 18 # <tt>delta</tt> is 3
+--   tf.range(start, limit, delta) ==&gt; [3, 6, 9, 12, 15] ```
+range :: (TensorType tidx, OneOf ((:) * Int32 ((:) * Int64 ([] *))) tidx) => Tensor v1 tidx -> Tensor v2 tidx -> Tensor v3 tidx -> Tensor Value tidx
+
+-- | Helper function for reduction ops (translation of
+--   math_ops.reduced_shape).
+reducedShape :: (OneOf '[Int32, Int64] t1, OneOf '[Int32, Int64] t2) => Tensor v1 t1 -> Tensor v2 t2 -> Tensor Value Int32
+
+-- | Computes rectified linear: `max(features, 0)`.
+relu :: (TensorType t, OneOf ((:) * Int16 ((:) * Int32 ((:) * Int64 ((:) * Int8 ((:) * Word16 ((:) * Word8 ((:) * Double ((:) * Float ([] *))))))))) t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes rectified linear gradients for a Relu operation.
+reluGrad :: (TensorType t, OneOf ((:) * Int16 ((:) * Int32 ((:) * Int64 ((:) * Int8 ((:) * Word16 ((:) * Word8 ((:) * Double ((:) * Float ([] *))))))))) t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Reshapes a tensor.
+--   
+--   Given <tt>tensor</tt>, this operation returns a tensor that has the
+--   same values as <tt>tensor</tt> with shape <a>shape</a>.
+--   
+--   If one component of <a>shape</a> is the special value -1, the size of
+--   that dimension is computed so that the total size remains constant. In
+--   particular, a <a>shape</a> of `[-1]` flattens into 1-D. At most one
+--   component of <a>shape</a> can be -1.
+--   
+--   If <a>shape</a> is 1-D or higher, then the operation returns a tensor
+--   with shape <a>shape</a> filled with the values of <tt>tensor</tt>. In
+--   this case, the number of elements implied by <a>shape</a> must be the
+--   same as the number of elements in <tt>tensor</tt>.
+--   
+--   For example:
+--   
+--   ```prettyprint # tensor <tt>t</tt> is [1, 2, 3, 4, 5, 6, 7, 8, 9] #
+--   tensor <tt>t</tt> has shape [9] reshape(t, [3, 3]) ==&gt; [[1, 2, 3],
+--   [4, 5, 6], [7, 8, 9]]
+--   
+--   # tensor <tt>t</tt> is [[[1, 1], [2, 2]], # [[3, 3], [4, 4]]] # tensor
+--   <tt>t</tt> has shape [2, 2, 2] reshape(t, [2, 4]) ==&gt; [[1, 1, 2,
+--   2], [3, 3, 4, 4]]
+--   
+--   # tensor <tt>t</tt> is [[[1, 1, 1], # [2, 2, 2]], # [[3, 3, 3], # [4,
+--   4, 4]], # [[5, 5, 5], # [6, 6, 6]]] # tensor <tt>t</tt> has shape [3,
+--   2, 3] # pass '[-1]' to flatten <tt>t</tt> reshape(t, [-1]) ==&gt; [1,
+--   1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6, 6, 6]
+--   
+--   # -1 can also be used to infer the shape
+--   
+--   # -1 is inferred to be 9: reshape(t, [2, -1]) ==&gt; [[1, 1, 1, 2, 2,
+--   2, 3, 3, 3], [4, 4, 4, 5, 5, 5, 6, 6, 6]] # -1 is inferred to be 2:
+--   reshape(t, [-1, 9]) ==&gt; [[1, 1, 1, 2, 2, 2, 3, 3, 3], [4, 4, 4, 5,
+--   5, 5, 6, 6, 6]] # -1 is inferred to be 3: reshape(t, [ 2, -1, 3])
+--   ==&gt; [[[1, 1, 1], [2, 2, 2], [3, 3, 3]], [[4, 4, 4], [5, 5, 5], [6,
+--   6, 6]]]
+--   
+--   # tensor <tt>t</tt> is [7] # shape `[]` reshapes to a scalar
+--   reshape(t, []) ==&gt; 7 ```
+reshape :: (TensorType t, TensorType tshape, OneOf ((:) * Int32 ((:) * Int64 ([] *))) tshape) => Tensor v1 t -> Tensor v2 tshape -> Tensor Value t
+
+-- | Restore a tensor's value from a checkpoint file.
+restore :: TensorType a => ByteString -> Tensor Ref a -> Build ControlNode
+
+-- | Restore a tensor's value from a checkpoint file.
+--   
+--   This version allows restoring from a checkpoint file that uses a
+--   different tensor name than the variable.
+restoreFromName :: TensorType a => ByteString -> ByteString -> Tensor Ref a -> Build ControlNode
+save :: TensorType a => ByteString -> [Tensor v a] -> Build ControlNode
+
+-- | Create a constant scalar.
+scalar :: TensorType a => a -> Tensor Value a
+shape :: (TensorType t) => Tensor v1 t -> Tensor Value Int32
+
+-- | Returns an element-wise indication of the sign of a number.
+--   
+--   `y = sign(x) = -1` if `x <a>0 if `x == 0`; 1 if `x</a> 0`.
+--   
+--   For complex numbers, `y = sign(x) = x / |x|` if `x != 0`, otherwise `y
+--   = 0`.
+sign :: (TensorType t, OneOf ((:) * (Complex Double) ((:) * (Complex Float) ((:) * Int32 ((:) * Int64 ((:) * Word16 ((:) * Double ((:) * Float ([] *)))))))) t) => Tensor v1 t -> Tensor Value t
+
+-- | Returns the size of a tensor.
+--   
+--   This operation returns an integer representing the number of elements
+--   in <tt>input</tt>.
+--   
+--   For example:
+--   
+--   ```prettyprint # <tt>t</tt> is [[[1, 1,, 1], [2, 2, 2]], [[3, 3, 3],
+--   [4, 4, 4]]]] size(t) ==&gt; 12 ```
+size :: (TensorType t, TensorType out_type, OneOf ((:) * Int32 ((:) * Int64 ([] *))) out_type) => Tensor v1 t -> Tensor Value out_type
+
+-- | Computes softmax activations.
+--   
+--   For each batch <tt>i</tt> and class <tt>j</tt> we have
+--   
+--   softmax[i, j] = exp(logits[i, j]) / sum_j(exp(logits[i, j]))
+softmax :: (TensorType t, OneOf ((:) * Word16 ((:) * Double ((:) * Float ([] *)))) t) => Tensor v1 t -> Tensor Value t
+
+-- | Computes softmax cross entropy cost and gradients to backpropagate.
+--   
+--   Inputs are the logits, not probabilities.
+softmaxCrossEntropyWithLogits :: (TensorType t, OneOf ((:) * Word16 ((:) * Double ((:) * Float ([] *)))) t) => Tensor v1 t -> Tensor v2 t -> (Tensor Value t, Tensor Value t)
+
+-- | Converts a sparse representation into a dense tensor.
+--   
+--   Builds an array <tt>dense</tt> with shape <tt>output_shape</tt> such
+--   that
+--   
+--   ```prettyprint # If sparse_indices is scalar dense[i] = (i ==
+--   sparse_indices ? sparse_values : default_value)
+--   
+--   # If sparse_indices is a vector, then for each i
+--   dense[sparse_indices[i]] = sparse_values[i]
+--   
+--   # If sparse_indices is an n by d matrix, then for each i in [0, n)
+--   dense[sparse_indices[i][0], ..., sparse_indices[i][d-1]] =
+--   sparse_values[i] ```
+--   
+--   All other values in <tt>dense</tt> are set to <tt>default_value</tt>.
+--   If <tt>sparse_values</tt> is a scalar, all sparse indices are set to
+--   this single value.
+--   
+--   Indices should be sorted in lexicographic order, and indices must not
+--   contain any repeats. If <tt>validate_indices</tt> is true, these
+--   properties are checked during execution.
+sparseToDense :: (TensorType t, TensorType tindices, OneOf ((:) * Int32 ((:) * Int64 ([] *))) tindices) => Tensor v1 tindices -> Tensor v2 tindices -> Tensor v3 t -> Tensor v4 t -> Tensor Value t
+
+-- | Returns x - y element-wise.
+--   
+--   <ul>
+--   <li>NOTE*: <tt>Sub</tt> supports broadcasting. More about broadcasting
+--   <a>here</a></li>
+--   </ul>
+sub :: (TensorType t, OneOf ((:) * (Complex Double) ((:) * (Complex Float) ((:) * Int32 ((:) * Int64 ((:) * Word16 ((:) * Double ((:) * Float ([] *)))))))) t) => Tensor v1 t -> Tensor v2 t -> Tensor Value t
+
+-- | Computes the sum of elements across dimensions of a tensor.
+--   
+--   Reduces <tt>input</tt> along the dimensions given in
+--   <tt>reduction_indices</tt>. Unless <tt>keep_dims</tt> is true, the
+--   rank of the tensor is reduced by 1 for each entry in
+--   <tt>reduction_indices</tt>. If <tt>keep_dims</tt> is true, the reduced
+--   dimensions are retained with length 1.
+sum :: (TensorType t, OneOf ((:) * (Complex Double) ((:) * (Complex Float) ((:) * Int16 ((:) * Int32 ((:) * Int64 ((:) * Int8 ((:) * Word16 ((:) * Word8 ((:) * Double ((:) * Float ([] *))))))))))) t, TensorType tidx, OneOf ((:) * Int32 ((:) * Int64 ([] *))) tidx) => Tensor v1 t -> Tensor v2 tidx -> Tensor Value t
+
+-- | Finds values and indices of the <tt>k</tt> largest elements for the
+--   last dimension.
+--   
+--   If the input is a vector (rank-1), finds the <tt>k</tt> largest
+--   entries in the vector and outputs their values and indices as vectors.
+--   Thus `values[j]` is the <tt>j</tt>-th largest entry in <tt>input</tt>,
+--   and its index is `indices[j]`.
+--   
+--   For matrices (resp. higher rank input), computes the top <tt>k</tt>
+--   entries in each row (resp. vector along the last dimension). Thus,
+--   
+--   values.shape = indices.shape = input.shape[:-1] + [k]
+--   
+--   If two elements are equal, the lower-index element appears first.
+--   
+--   If <tt>k</tt> varies dynamically, use <tt>TopKV2</tt> below.
+topK :: (TensorType t, OneOf ((:) * Int16 ((:) * Int32 ((:) * Int64 ((:) * Int8 ((:) * Word16 ((:) * Word8 ((:) * Double ((:) * Float ([] *))))))))) t) => Int64 -> Tensor v1 t -> (Tensor Value t, Tensor Value Int32)
+
+-- | Shuffle dimensions of x according to a permutation.
+--   
+--   The output <tt>y</tt> has the same rank as <tt>x</tt>. The shapes of
+--   <tt>x</tt> and <tt>y</tt> satisfy: `y.shape[i] == x.shape[perm[i]] for
+--   i in [0, 1, ..., rank(x) - 1]`
+transpose :: (TensorType t, TensorType tperm, OneOf ((:) * Int32 ((:) * Int64 ([] *))) tperm) => Tensor v1 t -> Tensor v2 tperm -> Tensor Value t
+truncatedNormal :: TensorType a => Tensor v Int64 -> Build (Tensor Value a)
+
+-- | Create a new, uninitialized stateful Tensor of the given shape.
+variable :: TensorType a => Shape -> Build (Tensor Ref a)
+
+-- | Create a constant vector.
+vector :: TensorType a => [a] -> Tensor Value a
+zeros :: (Num a, TensorType a) => Shape -> Tensor Value a
+
+-- | Returns a tensor of zeros with the same shape and type as x.
+zerosLike :: TensorType t => Tensor v1 t -> Tensor Value t
+instance (TensorFlow.Types.TensorType a, GHC.Num.Num a, v ~ TensorFlow.Tensor.Value, TensorFlow.Types.OneOf '[GHC.Types.Double, GHC.Types.Float, GHC.Int.Int32, GHC.Int.Int64, Data.Complex.Complex GHC.Types.Float, Data.Complex.Complex GHC.Types.Double] a) => GHC.Num.Num (TensorFlow.Tensor.Tensor v a)
+
+
+-- | Parallel lookups on the list of tensors.
+module TensorFlow.EmbeddingOps
+
+-- | Looks up <tt>ids</tt> in a list of embedding tensors.
+--   
+--   This function is used to perform parallel lookups on the list of
+--   tensors in <tt>params</tt>. It is a generalization of <a>gather</a>,
+--   where <tt>params</tt> is interpreted as a partition of a larger
+--   embedding tensor.
+--   
+--   The partition_strategy is "mod", we assign each id to partition `p =
+--   id % len(params)`. For instance, 13 ids are split across 5 partitions
+--   as: `[[0, 5, 10], [1, 6, 11], [2, 7, 12], [3, 8], [4, 9]]`
+--   
+--   The results of the lookup are concatenated into a dense tensor. The
+--   returned tensor has shape `shape(ids) + shape(params)[1:]`.
+embeddingLookup :: (TensorType a, OneOf '[Int64, Int32] b, Num b) => [Tensor v a] -> Tensor Value b -> Build (Tensor Value a)
+
+module TensorFlow.Gradient
+
+-- | Gradient of <tt>y</tt> w.r.t. each element of <tt>xs</tt>.
+gradients :: (Num (Tensor v1 a), v1 ~ Value, GradientCompatible a) => Tensor v1 a -> [Tensor v2 a] -> Build [Tensor Value a]
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html
new file mode 100644
index 0000000..57711ac
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-AttrValue.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Framework.AttrValue</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_Proto-Tensorflow-Core-Framework-AttrValue.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">Proto.Tensorflow.Core.Framework.AttrValue</p></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:AttrValue" class="def">AttrValue</a> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#AttrValue" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:AttrValue" class="def">AttrValue</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_AttrValue-39-s" class="def">_AttrValue's</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_AttrValue-39-i" class="def">_AttrValue'i</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_AttrValue-39-f" class="def">_AttrValue'f</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_AttrValue-39-b" class="def">_AttrValue'b</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_AttrValue-39-type-39-" class="def">_AttrValue'type'</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_AttrValue-39-shape" class="def">_AttrValue'shape</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_AttrValue-39-tensor" class="def">_AttrValue'tensor</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_AttrValue-39-list" class="def">_AttrValue'list</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_AttrValue-39-func" class="def">_AttrValue'func</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">NameAttrList</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_AttrValue-39-placeholder" class="def">_AttrValue'placeholder</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:AttrValue" class="caption collapser" onclick="toggleSection('i:AttrValue')">Instances</p><div id="section.i:AttrValue" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-44" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-44" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-214" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-225" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;b&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-97" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;f&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-81" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;func&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-182" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;i&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-65" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;list&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-166" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'b&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-105" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'f&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-89" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'func&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-190" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'i&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-73" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'list&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-174" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'placeholder&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-208" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe's&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-57" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'shape&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-140" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'tensor&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-157" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'type'&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-122" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;placeholder&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-199" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;s&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-49" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;shape&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-131" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;tensor&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-149" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;type'&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-114" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;b&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-95" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;f&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-79" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;func&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">NameAttrList</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-180" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;i&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-63" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;list&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-163" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'b&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-102" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'f&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-86" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'func&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">NameAttrList</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-187" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'i&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-70" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'list&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-171" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'placeholder&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-205" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe's&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-54" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'shape&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-136" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'tensor&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-154" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'type'&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-119" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;placeholder&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-196" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;s&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-46" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;shape&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-128" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;tensor&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-146" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;type'&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a> = <a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-111" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:AttrValue-39-ListValue" class="def">AttrValue'ListValue</a> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#AttrValue%27ListValue" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:AttrValue-39-ListValue" class="def">AttrValue'ListValue</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_AttrValue-39-ListValue-39-s" class="def">_AttrValue'ListValue's</a> :: [<a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_AttrValue-39-ListValue-39-i" class="def">_AttrValue'ListValue'i</a> :: [<a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_AttrValue-39-ListValue-39-f" class="def">_AttrValue'ListValue'f</a> :: [<a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_AttrValue-39-ListValue-39-b" class="def">_AttrValue'ListValue'b</a> :: [<a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_AttrValue-39-ListValue-39-type-39-" class="def">_AttrValue'ListValue'type'</a> :: [<a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_AttrValue-39-ListValue-39-shape" class="def">_AttrValue'ListValue'shape</a> :: [<a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_AttrValue-39-ListValue-39-tensor" class="def">_AttrValue'ListValue'tensor</a> :: [<a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a>]</dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:AttrValue-39-ListValue" class="caption collapser" onclick="toggleSection('i:AttrValue-39-ListValue')">Instances</p><div id="section.i:AttrValue-39-ListValue" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-313" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-313" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-378" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-385" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;b&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-345" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;f&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-336" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;i&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-327" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;s&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-318" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;shape&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-363" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;tensor&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-372" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;type'&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-354" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;b&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a> = [<a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-342" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;f&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a> = [<a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-333" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;i&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a> = [<a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-324" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;s&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a> = [<a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-315" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;shape&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a> = [<a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-360" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;tensor&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a> = [<a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-369" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;type'&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">AttrValue'ListValue</a> = [<a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-351" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:NameAttrList" class="def">NameAttrList</a> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#NameAttrList" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:NameAttrList" class="def">NameAttrList</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_NameAttrList-39-name" class="def">_NameAttrList'name</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_NameAttrList-39-attr" class="def">_NameAttrList'attr</a> :: <a href="../containers-0.5.6.2/Data-Map-Lazy.html#t:Map">Map</a> <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:NameAttrList" class="caption collapser" onclick="toggleSection('i:NameAttrList')">Instances</p><div id="section.i:NameAttrList" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">NameAttrList</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-445" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">NameAttrList</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-445" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">NameAttrList</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-465" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">NameAttrList</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-470" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;attr&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">NameAttrList</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">NameAttrList</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-459" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;name&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">NameAttrList</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">NameAttrList</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-450" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;attr&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">NameAttrList</a> = <a href="../containers-0.5.6.2/Data-Map-Lazy.html#t:Map">Map</a> <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-456" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;name&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">NameAttrList</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-447" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:NameAttrList-39-AttrEntry" class="def">NameAttrList'AttrEntry</a> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#NameAttrList%27AttrEntry" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:NameAttrList-39-AttrEntry" class="def">NameAttrList'AttrEntry</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_NameAttrList-39-AttrEntry-39-key" class="def">_NameAttrList'AttrEntry'key</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_NameAttrList-39-AttrEntry-39-value" class="def">_NameAttrList'AttrEntry'value</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:NameAttrList-39-AttrEntry" class="caption collapser" onclick="toggleSection('i:NameAttrList-39-AttrEntry')">Instances</p><div id="section.i:NameAttrList-39-AttrEntry" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">NameAttrList'AttrEntry</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-495" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">NameAttrList'AttrEntry</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-495" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">NameAttrList'AttrEntry</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-525" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">NameAttrList'AttrEntry</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-531" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;key&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">NameAttrList'AttrEntry</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">NameAttrList'AttrEntry</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-500" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'value&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">NameAttrList'AttrEntry</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">NameAttrList'AttrEntry</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-519" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;value&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">NameAttrList'AttrEntry</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">NameAttrList'AttrEntry</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-509" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;key&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">NameAttrList'AttrEntry</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-497" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'value&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">NameAttrList'AttrEntry</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-515" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;value&quot; <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">NameAttrList'AttrEntry</a> = <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#line-506" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><a name="v:attr" class="def">attr</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;attr&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;attr&quot; msg) (Field &quot;attr&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#attr" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:b" class="def">b</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;b&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;b&quot; msg) (Field &quot;b&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#b" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:f" class="def">f</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;f&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;f&quot; msg) (Field &quot;f&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#f" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:func" class="def">func</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;func&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;func&quot; msg) (Field &quot;func&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#func" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:i" class="def">i</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;i&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;i&quot; msg) (Field &quot;i&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#i" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:key" class="def">key</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;key&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;key&quot; msg) (Field &quot;key&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#key" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:list" class="def">list</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;list&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;list&quot; msg) (Field &quot;list&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#list" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-b" class="def">maybe'b</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'b&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'b&quot; msg) (Field &quot;maybe'b&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#maybe%27b" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-f" class="def">maybe'f</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'f&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'f&quot; msg) (Field &quot;maybe'f&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#maybe%27f" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-func" class="def">maybe'func</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'func&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'func&quot; msg) (Field &quot;maybe'func&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#maybe%27func" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-i" class="def">maybe'i</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'i&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'i&quot; msg) (Field &quot;maybe'i&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#maybe%27i" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-list" class="def">maybe'list</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'list&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'list&quot; msg) (Field &quot;maybe'list&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#maybe%27list" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-placeholder" class="def">maybe'placeholder</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'placeholder&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'placeholder&quot; msg) (Field &quot;maybe'placeholder&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#maybe%27placeholder" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-s" class="def">maybe's</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe's&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe's&quot; msg) (Field &quot;maybe's&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#maybe%27s" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-shape" class="def">maybe'shape</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'shape&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'shape&quot; msg) (Field &quot;maybe'shape&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#maybe%27shape" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-tensor" class="def">maybe'tensor</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'tensor&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'tensor&quot; msg) (Field &quot;maybe'tensor&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#maybe%27tensor" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-type-39-" class="def">maybe'type'</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'type'&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'type'&quot; msg) (Field &quot;maybe'type'&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#maybe%27type%27" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-value" class="def">maybe'value</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'value&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'value&quot; msg) (Field &quot;maybe'value&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#maybe%27value" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:name" class="def">name</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;name&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;name&quot; msg) (Field &quot;name&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#name" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:placeholder" class="def">placeholder</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;placeholder&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;placeholder&quot; msg) (Field &quot;placeholder&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#placeholder" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:s" class="def">s</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;s&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;s&quot; msg) (Field &quot;s&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#s" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:shape" class="def">shape</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;shape&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;shape&quot; msg) (Field &quot;shape&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#shape" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:tensor" class="def">tensor</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;tensor&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;tensor&quot; msg) (Field &quot;tensor&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#tensor" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:type-39-" class="def">type'</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;type'&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;type'&quot; msg) (Field &quot;type'&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#type%27" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:value" class="def">value</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;value&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;value&quot; msg) (Field &quot;value&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-AttrValue.html#value" class="link">Source</a></p></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html
new file mode 100644
index 0000000..d9a1521
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Graph.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Framework.Graph</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_Proto-Tensorflow-Core-Framework-Graph.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/Proto-Tensorflow-Core-Framework-Graph.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">Proto.Tensorflow.Core.Framework.Graph</p></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:GraphDef" class="def">GraphDef</a> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#GraphDef" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:GraphDef" class="def">GraphDef</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_GraphDef-39-node" class="def">_GraphDef'node</a> :: [<a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_GraphDef-39-versions" class="def">_GraphDef'versions</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> VersionDef</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_GraphDef-39-version" class="def">_GraphDef'version</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_GraphDef-39-library" class="def">_GraphDef'library</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> FunctionDefLibrary</dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:GraphDef" class="caption collapser" onclick="toggleSection('i:GraphDef')">Instances</p><div id="section.i:GraphDef" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#line-37" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#line-37" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#line-90" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#line-97" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;library&quot; <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a> <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#line-75" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'library&quot; <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a> <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#line-84" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'versions&quot; <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a> <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#line-58" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;node&quot; <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a> <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#line-42" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;version&quot; <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a> <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#line-67" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;versions&quot; <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a> <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#line-50" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;library&quot; <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#line-72" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'library&quot; <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#line-80" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'versions&quot; <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#line-55" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;node&quot; <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a> = [<a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#line-39" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;version&quot; <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#line-64" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;versions&quot; <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#line-47" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><a name="v:library" class="def">library</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;library&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;library&quot; msg) (Field &quot;library&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#library" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-library" class="def">maybe'library</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'library&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'library&quot; msg) (Field &quot;maybe'library&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#maybe%27library" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-versions" class="def">maybe'versions</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'versions&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'versions&quot; msg) (Field &quot;maybe'versions&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#maybe%27versions" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:node" class="def">node</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;node&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;node&quot; msg) (Field &quot;node&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#node" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:version" class="def">version</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;version&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;version&quot; msg) (Field &quot;version&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#version" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:versions" class="def">versions</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;versions&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;versions&quot; msg) (Field &quot;versions&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Graph.html#versions" class="link">Source</a></p></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html
new file mode 100644
index 0000000..fa368f3
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-NodeDef.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Framework.NodeDef</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_Proto-Tensorflow-Core-Framework-NodeDef.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">Proto.Tensorflow.Core.Framework.NodeDef</p></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:NodeDef" class="def">NodeDef</a> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#NodeDef" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:NodeDef" class="def">NodeDef</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_NodeDef-39-name" class="def">_NodeDef'name</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_NodeDef-39-op" class="def">_NodeDef'op</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_NodeDef-39-input" class="def">_NodeDef'input</a> :: [<a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_NodeDef-39-device" class="def">_NodeDef'device</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_NodeDef-39-attr" class="def">_NodeDef'attr</a> :: <a href="../containers-0.5.6.2/Data-Map-Lazy.html#t:Map">Map</a> <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:NodeDef" class="caption collapser" onclick="toggleSection('i:NodeDef')">Instances</p><div id="section.i:NodeDef" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-33" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-33" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-74" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-81" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;attr&quot; <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a> <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-69" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;device&quot; <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a> <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-60" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;input&quot; <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a> <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-52" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;name&quot; <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a> <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-37" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;op&quot; <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a> <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-44" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;attr&quot; <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a> = <a href="../containers-0.5.6.2/Data-Map-Lazy.html#t:Map">Map</a> <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-65" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;device&quot; <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-57" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;input&quot; <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a> = [<a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-49" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;name&quot; <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-35" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;op&quot; <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">NodeDef</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-42" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:NodeDef-39-AttrEntry" class="def">NodeDef'AttrEntry</a> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#NodeDef%27AttrEntry" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:NodeDef-39-AttrEntry" class="def">NodeDef'AttrEntry</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_NodeDef-39-AttrEntry-39-key" class="def">_NodeDef'AttrEntry'key</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_NodeDef-39-AttrEntry-39-value" class="def">_NodeDef'AttrEntry'value</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:NodeDef-39-AttrEntry" class="caption collapser" onclick="toggleSection('i:NodeDef-39-AttrEntry')">Instances</p><div id="section.i:NodeDef-39-AttrEntry" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">NodeDef'AttrEntry</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-127" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">NodeDef'AttrEntry</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-127" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">NodeDef'AttrEntry</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-156" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">NodeDef'AttrEntry</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-162" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;key&quot; <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">NodeDef'AttrEntry</a> <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">NodeDef'AttrEntry</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-132" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'value&quot; <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">NodeDef'AttrEntry</a> <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">NodeDef'AttrEntry</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-150" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;value&quot; <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">NodeDef'AttrEntry</a> <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">NodeDef'AttrEntry</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-141" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;key&quot; <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">NodeDef'AttrEntry</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-129" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'value&quot; <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">NodeDef'AttrEntry</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-147" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;value&quot; <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">NodeDef'AttrEntry</a> = <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#line-138" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><a name="v:attr" class="def">attr</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;attr&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;attr&quot; msg) (Field &quot;attr&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#attr" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:device" class="def">device</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;device&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;device&quot; msg) (Field &quot;device&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#device" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:input" class="def">input</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;input&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;input&quot; msg) (Field &quot;input&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#input" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:key" class="def">key</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;key&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;key&quot; msg) (Field &quot;key&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#key" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-value" class="def">maybe'value</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'value&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'value&quot; msg) (Field &quot;maybe'value&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#maybe%27value" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:name" class="def">name</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;name&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;name&quot; msg) (Field &quot;name&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#name" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:op" class="def">op</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;op&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;op&quot; msg) (Field &quot;op&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#op" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:value" class="def">value</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;value&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;value&quot; msg) (Field &quot;value&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-NodeDef.html#value" class="link">Source</a></p></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html
new file mode 100644
index 0000000..c656d2f
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-OpDef.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Framework.OpDef</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_Proto-Tensorflow-Core-Framework-OpDef.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/Proto-Tensorflow-Core-Framework-OpDef.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">Proto.Tensorflow.Core.Framework.OpDef</p></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:OpDef" class="def">OpDef</a> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#OpDef" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:OpDef" class="def">OpDef</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_OpDef-39-name" class="def">_OpDef'name</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-inputArg" class="def">_OpDef'inputArg</a> :: [<a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-outputArg" class="def">_OpDef'outputArg</a> :: [<a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-attr" class="def">_OpDef'attr</a> :: [<a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-deprecation" class="def">_OpDef'deprecation</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">OpDeprecation</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-summary" class="def">_OpDef'summary</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-description" class="def">_OpDef'description</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-isCommutative" class="def">_OpDef'isCommutative</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-isAggregate" class="def">_OpDef'isAggregate</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-isStateful" class="def">_OpDef'isStateful</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-allowsUninitializedInput" class="def">_OpDef'allowsUninitializedInput</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:OpDef" class="caption collapser" onclick="toggleSection('i:OpDef')">Instances</p><div id="section.i:OpDef" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-38" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-38" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-135" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-147" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;allowsUninitializedInput&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-129" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;attr&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-65" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;deprecation&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-73" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;description&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-97" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;inputArg&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-50" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;isAggregate&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-113" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;isCommutative&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-105" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;isStateful&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-121" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'deprecation&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-81" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;name&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-42" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;outputArg&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-58" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;summary&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-89" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;allowsUninitializedInput&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-126" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;attr&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> = [<a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-63" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;deprecation&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> = <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">OpDeprecation</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-70" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;description&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-94" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;inputArg&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> = [<a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-47" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;isAggregate&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-110" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;isCommutative&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-102" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;isStateful&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-118" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'deprecation&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">OpDeprecation</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-78" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;name&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-40" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;outputArg&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> = [<a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-55" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;summary&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-87" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:OpDef-39-ArgDef" class="def">OpDef'ArgDef</a> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#OpDef%27ArgDef" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:OpDef-39-ArgDef" class="def">OpDef'ArgDef</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_OpDef-39-ArgDef-39-name" class="def">_OpDef'ArgDef'name</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-ArgDef-39-description" class="def">_OpDef'ArgDef'description</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-ArgDef-39-type-39-" class="def">_OpDef'ArgDef'type'</a> :: <a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-ArgDef-39-typeAttr" class="def">_OpDef'ArgDef'typeAttr</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-ArgDef-39-numberAttr" class="def">_OpDef'ArgDef'numberAttr</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-ArgDef-39-typeListAttr" class="def">_OpDef'ArgDef'typeListAttr</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-ArgDef-39-isRef" class="def">_OpDef'ArgDef'isRef</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:OpDef-39-ArgDef" class="caption collapser" onclick="toggleSection('i:OpDef-39-ArgDef')">Instances</p><div id="section.i:OpDef-39-ArgDef" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-243" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-243" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-308" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-318" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;description&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-257" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;isRef&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-302" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;name&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-248" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;numberAttr&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-284" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;type'&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-266" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;typeAttr&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-275" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;typeListAttr&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-293" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;description&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-254" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;isRef&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-299" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;name&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-245" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;numberAttr&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-281" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;type'&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a> = <a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-263" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;typeAttr&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-272" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;typeListAttr&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">OpDef'ArgDef</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-290" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:OpDef-39-AttrDef" class="def">OpDef'AttrDef</a> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#OpDef%27AttrDef" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:OpDef-39-AttrDef" class="def">OpDef'AttrDef</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_OpDef-39-AttrDef-39-name" class="def">_OpDef'AttrDef'name</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-AttrDef-39-type-39-" class="def">_OpDef'AttrDef'type'</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-AttrDef-39-defaultValue" class="def">_OpDef'AttrDef'defaultValue</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-AttrDef-39-description" class="def">_OpDef'AttrDef'description</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-AttrDef-39-hasMinimum" class="def">_OpDef'AttrDef'hasMinimum</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-AttrDef-39-minimum" class="def">_OpDef'AttrDef'minimum</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDef-39-AttrDef-39-allowedValues" class="def">_OpDef'AttrDef'allowedValues</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:OpDef-39-AttrDef" class="caption collapser" onclick="toggleSection('i:OpDef-39-AttrDef')">Instances</p><div id="section.i:OpDef-39-AttrDef" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-387" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-387" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-472" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-482" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;allowedValues&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-456" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;defaultValue&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-410" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;description&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-429" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;hasMinimum&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-438" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'allowedValues&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-466" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'defaultValue&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-420" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;minimum&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-447" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;name&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-392" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;type'&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-401" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;allowedValues&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> = <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-453" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;defaultValue&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> = <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-407" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;description&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-426" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;hasMinimum&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-435" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'allowedValues&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-462" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'defaultValue&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">AttrValue</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-416" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;minimum&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-444" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;name&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-389" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;type'&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">OpDef'AttrDef</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-398" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:OpDeprecation" class="def">OpDeprecation</a> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#OpDeprecation" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:OpDeprecation" class="def">OpDeprecation</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_OpDeprecation-39-version" class="def">_OpDeprecation'version</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OpDeprecation-39-explanation" class="def">_OpDeprecation'explanation</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:OpDeprecation" class="caption collapser" onclick="toggleSection('i:OpDeprecation')">Instances</p><div id="section.i:OpDeprecation" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">OpDeprecation</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-543" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">OpDeprecation</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-543" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">OpDeprecation</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-563" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">OpDeprecation</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-569" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;explanation&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">OpDeprecation</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">OpDeprecation</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-557" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;version&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">OpDeprecation</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">OpDeprecation</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-548" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;explanation&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">OpDeprecation</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-554" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;version&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">OpDeprecation</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-545" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:OpList" class="def">OpList</a> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#OpList" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:OpList" class="def">OpList</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_OpList-39-op" class="def">_OpList'op</a> :: [<a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a>]</dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:OpList" class="caption collapser" onclick="toggleSection('i:OpList')">Instances</p><div id="section.i:OpList" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpList">OpList</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-591" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpList">OpList</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-591" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpList">OpList</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-600" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpList">OpList</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-603" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;op&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpList">OpList</a> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpList">OpList</a></span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-595" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;op&quot; <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpList">OpList</a> = [<a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">OpDef</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#line-593" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><a name="v:allowedValues" class="def">allowedValues</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;allowedValues&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;allowedValues&quot; msg) (Field &quot;allowedValues&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#allowedValues" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:allowsUninitializedInput" class="def">allowsUninitializedInput</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;allowsUninitializedInput&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;allowsUninitializedInput&quot; msg) (Field &quot;allowsUninitializedInput&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#allowsUninitializedInput" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:attr" class="def">attr</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;attr&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;attr&quot; msg) (Field &quot;attr&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#attr" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:defaultValue" class="def">defaultValue</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;defaultValue&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;defaultValue&quot; msg) (Field &quot;defaultValue&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#defaultValue" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:deprecation" class="def">deprecation</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;deprecation&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;deprecation&quot; msg) (Field &quot;deprecation&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#deprecation" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:description" class="def">description</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;description&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;description&quot; msg) (Field &quot;description&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#description" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:explanation" class="def">explanation</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;explanation&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;explanation&quot; msg) (Field &quot;explanation&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#explanation" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:hasMinimum" class="def">hasMinimum</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;hasMinimum&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;hasMinimum&quot; msg) (Field &quot;hasMinimum&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#hasMinimum" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:inputArg" class="def">inputArg</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;inputArg&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;inputArg&quot; msg) (Field &quot;inputArg&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#inputArg" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:isAggregate" class="def">isAggregate</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;isAggregate&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;isAggregate&quot; msg) (Field &quot;isAggregate&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#isAggregate" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:isCommutative" class="def">isCommutative</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;isCommutative&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;isCommutative&quot; msg) (Field &quot;isCommutative&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#isCommutative" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:isRef" class="def">isRef</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;isRef&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;isRef&quot; msg) (Field &quot;isRef&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#isRef" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:isStateful" class="def">isStateful</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;isStateful&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;isStateful&quot; msg) (Field &quot;isStateful&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#isStateful" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-allowedValues" class="def">maybe'allowedValues</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'allowedValues&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'allowedValues&quot; msg) (Field &quot;maybe'allowedValues&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#maybe%27allowedValues" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-defaultValue" class="def">maybe'defaultValue</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'defaultValue&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'defaultValue&quot; msg) (Field &quot;maybe'defaultValue&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#maybe%27defaultValue" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-deprecation" class="def">maybe'deprecation</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'deprecation&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'deprecation&quot; msg) (Field &quot;maybe'deprecation&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#maybe%27deprecation" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:minimum" class="def">minimum</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;minimum&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;minimum&quot; msg) (Field &quot;minimum&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#minimum" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:name" class="def">name</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;name&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;name&quot; msg) (Field &quot;name&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#name" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:numberAttr" class="def">numberAttr</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;numberAttr&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;numberAttr&quot; msg) (Field &quot;numberAttr&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#numberAttr" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:op" class="def">op</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;op&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;op&quot; msg) (Field &quot;op&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#op" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:outputArg" class="def">outputArg</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;outputArg&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;outputArg&quot; msg) (Field &quot;outputArg&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#outputArg" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:summary" class="def">summary</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;summary&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;summary&quot; msg) (Field &quot;summary&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#summary" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:type-39-" class="def">type'</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;type'&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;type'&quot; msg) (Field &quot;type'&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#type%27" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:typeAttr" class="def">typeAttr</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;typeAttr&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;typeAttr&quot; msg) (Field &quot;typeAttr&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#typeAttr" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:typeListAttr" class="def">typeListAttr</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;typeListAttr&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;typeListAttr&quot; msg) (Field &quot;typeListAttr&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#typeListAttr" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:version" class="def">version</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;version&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;version&quot; msg) (Field &quot;version&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-OpDef.html#version" class="link">Source</a></p></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html
new file mode 100644
index 0000000..b529bd1
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-ResourceHandle.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Framework.ResourceHandle</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_Proto-Tensorflow-Core-Framework-ResourceHandle.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">Proto.Tensorflow.Core.Framework.ResourceHandle</p></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:ResourceHandle" class="def">ResourceHandle</a> <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#ResourceHandle" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:ResourceHandle" class="def">ResourceHandle</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_ResourceHandle-39-device" class="def">_ResourceHandle'device</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_ResourceHandle-39-container" class="def">_ResourceHandle'container</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_ResourceHandle-39-name" class="def">_ResourceHandle'name</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_ResourceHandle-39-hashCode" class="def">_ResourceHandle'hashCode</a> :: <a href="../base-4.8.2.0/Data-Word.html#t:Word64">Word64</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_ResourceHandle-39-maybeTypeName" class="def">_ResourceHandle'maybeTypeName</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:ResourceHandle" class="caption collapser" onclick="toggleSection('i:ResourceHandle')">Instances</p><div id="section.i:ResourceHandle" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a></span> <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#line-32" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a></span> <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#line-32" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a></span> <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#line-79" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a></span> <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#line-88" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;container&quot; <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a> <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a></span> <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#line-46" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;device&quot; <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a> <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a></span> <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#line-37" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;hashCode&quot; <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a> <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a></span> <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#line-64" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybeTypeName&quot; <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a> <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a></span> <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#line-73" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;name&quot; <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a> <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a></span> <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#line-55" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;container&quot; <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#line-43" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;device&quot; <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#line-34" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;hashCode&quot; <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a> = <a href="../base-4.8.2.0/Data-Word.html#t:Word64">Word64</a></span> <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#line-61" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybeTypeName&quot; <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#line-70" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;name&quot; <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#line-52" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><a name="v:container" class="def">container</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;container&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;container&quot; msg) (Field &quot;container&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#container" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:device" class="def">device</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;device&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;device&quot; msg) (Field &quot;device&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#device" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:hashCode" class="def">hashCode</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;hashCode&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;hashCode&quot; msg) (Field &quot;hashCode&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#hashCode" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybeTypeName" class="def">maybeTypeName</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybeTypeName&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybeTypeName&quot; msg) (Field &quot;maybeTypeName&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#maybeTypeName" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:name" class="def">name</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;name&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;name&quot; msg) (Field &quot;name&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-ResourceHandle.html#name" class="link">Source</a></p></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html
new file mode 100644
index 0000000..f0dcc5c
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Tensor.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Framework.Tensor</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_Proto-Tensorflow-Core-Framework-Tensor.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/Proto-Tensorflow-Core-Framework-Tensor.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">Proto.Tensorflow.Core.Framework.Tensor</p></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:TensorProto" class="def">TensorProto</a> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#TensorProto" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:TensorProto" class="def">TensorProto</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_TensorProto-39-dtype" class="def">_TensorProto'dtype</a> :: <a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_TensorProto-39-tensorShape" class="def">_TensorProto'tensorShape</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_TensorProto-39-versionNumber" class="def">_TensorProto'versionNumber</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_TensorProto-39-tensorContent" class="def">_TensorProto'tensorContent</a> :: <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_TensorProto-39-halfVal" class="def">_TensorProto'halfVal</a> :: [<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_TensorProto-39-floatVal" class="def">_TensorProto'floatVal</a> :: [<a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_TensorProto-39-doubleVal" class="def">_TensorProto'doubleVal</a> :: [<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_TensorProto-39-intVal" class="def">_TensorProto'intVal</a> :: [<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_TensorProto-39-stringVal" class="def">_TensorProto'stringVal</a> :: [<a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_TensorProto-39-scomplexVal" class="def">_TensorProto'scomplexVal</a> :: [<a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_TensorProto-39-int64Val" class="def">_TensorProto'int64Val</a> :: [<a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_TensorProto-39-boolVal" class="def">_TensorProto'boolVal</a> :: [<a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_TensorProto-39-dcomplexVal" class="def">_TensorProto'dcomplexVal</a> :: [<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_TensorProto-39-resourceHandleVal" class="def">_TensorProto'resourceHandleVal</a> :: [<a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a>]</dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:TensorProto" class="caption collapser" onclick="toggleSection('i:TensorProto')">Instances</p><div id="section.i:TensorProto" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-47" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-47" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-186" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-198" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;boolVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-162" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;dcomplexVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-171" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;doubleVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-117" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;dtype&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-52" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;floatVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-108" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;halfVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-99" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;int64Val&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-153" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;intVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-126" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'tensorShape&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-72" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;resourceHandleVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-180" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;scomplexVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-144" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;stringVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-135" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;tensorContent&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-90" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;tensorShape&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-61" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;versionNumber&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-81" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;boolVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> = [<a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-159" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;dcomplexVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> = [<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-168" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;doubleVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> = [<a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-114" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;dtype&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> = <a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-49" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;floatVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> = [<a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-105" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;halfVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> = [<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-96" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;int64Val&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> = [<a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-150" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;intVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> = [<a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-123" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'tensorShape&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-67" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;resourceHandleVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> = [<a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">ResourceHandle</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-177" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;scomplexVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> = [<a href="../base-4.8.2.0/Prelude.html#t:Float">Float</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-141" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;stringVal&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> = [<a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-132" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;tensorContent&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> = <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-87" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;tensorShape&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> = <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-58" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;versionNumber&quot; <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">TensorProto</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#line-78" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><a name="v:boolVal" class="def">boolVal</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;boolVal&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;boolVal&quot; msg) (Field &quot;boolVal&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#boolVal" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:dcomplexVal" class="def">dcomplexVal</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;dcomplexVal&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;dcomplexVal&quot; msg) (Field &quot;dcomplexVal&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#dcomplexVal" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:doubleVal" class="def">doubleVal</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;doubleVal&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;doubleVal&quot; msg) (Field &quot;doubleVal&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#doubleVal" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:dtype" class="def">dtype</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;dtype&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;dtype&quot; msg) (Field &quot;dtype&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#dtype" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:floatVal" class="def">floatVal</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;floatVal&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;floatVal&quot; msg) (Field &quot;floatVal&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#floatVal" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:halfVal" class="def">halfVal</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;halfVal&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;halfVal&quot; msg) (Field &quot;halfVal&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#halfVal" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:int64Val" class="def">int64Val</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;int64Val&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;int64Val&quot; msg) (Field &quot;int64Val&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#int64Val" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:intVal" class="def">intVal</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;intVal&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;intVal&quot; msg) (Field &quot;intVal&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#intVal" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-tensorShape" class="def">maybe'tensorShape</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'tensorShape&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'tensorShape&quot; msg) (Field &quot;maybe'tensorShape&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#maybe%27tensorShape" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:resourceHandleVal" class="def">resourceHandleVal</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;resourceHandleVal&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;resourceHandleVal&quot; msg) (Field &quot;resourceHandleVal&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#resourceHandleVal" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:scomplexVal" class="def">scomplexVal</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;scomplexVal&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;scomplexVal&quot; msg) (Field &quot;scomplexVal&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#scomplexVal" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:stringVal" class="def">stringVal</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;stringVal&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;stringVal&quot; msg) (Field &quot;stringVal&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#stringVal" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:tensorContent" class="def">tensorContent</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;tensorContent&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;tensorContent&quot; msg) (Field &quot;tensorContent&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#tensorContent" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:tensorShape" class="def">tensorShape</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;tensorShape&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;tensorShape&quot; msg) (Field &quot;tensorShape&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#tensorShape" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:versionNumber" class="def">versionNumber</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;versionNumber&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;versionNumber&quot; msg) (Field &quot;versionNumber&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-Tensor.html#versionNumber" class="link">Source</a></p></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html
new file mode 100644
index 0000000..3db84a4
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-TensorShape.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Framework.TensorShape</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_Proto-Tensorflow-Core-Framework-TensorShape.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">Proto.Tensorflow.Core.Framework.TensorShape</p></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:TensorShapeProto" class="def">TensorShapeProto</a> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#TensorShapeProto" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:TensorShapeProto" class="def">TensorShapeProto</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_TensorShapeProto-39-dim" class="def">_TensorShapeProto'dim</a> :: [<a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim">TensorShapeProto'Dim</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_TensorShapeProto-39-unknownRank" class="def">_TensorShapeProto'unknownRank</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:TensorShapeProto" class="caption collapser" onclick="toggleSection('i:TensorShapeProto')">Instances</p><div id="section.i:TensorShapeProto" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#line-29" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#line-29" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#line-49" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#line-54" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;dim&quot; <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a> <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#line-34" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;unknownRank&quot; <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a> <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a></span> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#line-43" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;dim&quot; <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a> = [<a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim">TensorShapeProto'Dim</a>]</span> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#line-31" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;unknownRank&quot; <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">TensorShapeProto</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#line-40" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:TensorShapeProto-39-Dim" class="def">TensorShapeProto'Dim</a> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#TensorShapeProto%27Dim" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:TensorShapeProto-39-Dim" class="def">TensorShapeProto'Dim</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_TensorShapeProto-39-Dim-39-size" class="def">_TensorShapeProto'Dim'size</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_TensorShapeProto-39-Dim-39-name" class="def">_TensorShapeProto'Dim'name</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:TensorShapeProto-39-Dim" class="caption collapser" onclick="toggleSection('i:TensorShapeProto-39-Dim')">Instances</p><div id="section.i:TensorShapeProto-39-Dim" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim">TensorShapeProto'Dim</a></span> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#line-78" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim">TensorShapeProto'Dim</a></span> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#line-78" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim">TensorShapeProto'Dim</a></span> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#line-98" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim">TensorShapeProto'Dim</a></span> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#line-104" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;name&quot; <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim">TensorShapeProto'Dim</a> <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim">TensorShapeProto'Dim</a></span> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#line-92" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;size&quot; <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim">TensorShapeProto'Dim</a> <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim">TensorShapeProto'Dim</a></span> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#line-83" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;name&quot; <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim">TensorShapeProto'Dim</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#line-89" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;size&quot; <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim">TensorShapeProto'Dim</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></span> <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#line-80" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><a name="v:dim" class="def">dim</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;dim&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;dim&quot; msg) (Field &quot;dim&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#dim" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:name" class="def">name</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;name&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;name&quot; msg) (Field &quot;name&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#name" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:size" class="def">size</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;size&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;size&quot; msg) (Field &quot;size&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#size" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:unknownRank" class="def">unknownRank</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;unknownRank&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;unknownRank&quot; msg) (Field &quot;unknownRank&quot; msg') <a href="src/Proto-Tensorflow-Core-Framework-TensorShape.html#unknownRank" class="link">Source</a></p></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html
new file mode 100644
index 0000000..39ff85a
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Framework-Types.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Framework.Types</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_Proto-Tensorflow-Core-Framework-Types.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/Proto-Tensorflow-Core-Framework-Types.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">Proto.Tensorflow.Core.Framework.Types</p></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:DataType" class="def">DataType</a> <a href="src/Proto-Tensorflow-Core-Framework-Types.html#DataType" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:DT_INVALID" class="def">DT_INVALID</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_FLOAT" class="def">DT_FLOAT</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_DOUBLE" class="def">DT_DOUBLE</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_INT32" class="def">DT_INT32</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_UINT8" class="def">DT_UINT8</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_INT16" class="def">DT_INT16</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_INT8" class="def">DT_INT8</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_STRING" class="def">DT_STRING</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_COMPLEX64" class="def">DT_COMPLEX64</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_INT64" class="def">DT_INT64</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_BOOL" class="def">DT_BOOL</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_QINT8" class="def">DT_QINT8</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_QUINT8" class="def">DT_QUINT8</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_QINT32" class="def">DT_QINT32</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_BFLOAT16" class="def">DT_BFLOAT16</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_QINT16" class="def">DT_QINT16</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_QUINT16" class="def">DT_QUINT16</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_UINT16" class="def">DT_UINT16</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_COMPLEX128" class="def">DT_COMPLEX128</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_HALF" class="def">DT_HALF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_RESOURCE" class="def">DT_RESOURCE</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_FLOAT_REF" class="def">DT_FLOAT_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_DOUBLE_REF" class="def">DT_DOUBLE_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_INT32_REF" class="def">DT_INT32_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_UINT8_REF" class="def">DT_UINT8_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_INT16_REF" class="def">DT_INT16_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_INT8_REF" class="def">DT_INT8_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_STRING_REF" class="def">DT_STRING_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_COMPLEX64_REF" class="def">DT_COMPLEX64_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_INT64_REF" class="def">DT_INT64_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_BOOL_REF" class="def">DT_BOOL_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_QINT8_REF" class="def">DT_QINT8_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_QUINT8_REF" class="def">DT_QUINT8_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_QINT32_REF" class="def">DT_QINT32_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_BFLOAT16_REF" class="def">DT_BFLOAT16_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_QINT16_REF" class="def">DT_QINT16_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_QUINT16_REF" class="def">DT_QUINT16_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_UINT16_REF" class="def">DT_UINT16_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_COMPLEX128_REF" class="def">DT_COMPLEX128_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_HALF_REF" class="def">DT_HALF_REF</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:DT_RESOURCE_REF" class="def">DT_RESOURCE_REF</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="subs instances"><p id="control.i:DataType" class="caption collapser" onclick="toggleSection('i:DataType')">Instances</p><div id="section.i:DataType" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Prelude.html#t:Bounded">Bounded</a> <a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Types.html#line-342" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Prelude.html#t:Enum">Enum</a> <a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Types.html#line-202" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Types.html#line-67" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Types.html#line-67" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Types.html#line-69" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">MessageEnum <a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Types.html#line-75" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">FieldDefault <a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">DataType</a></span> <a href="src/Proto-Tensorflow-Core-Framework-Types.html#line-72" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html
new file mode 100644
index 0000000..c1f492f
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/Proto-Tensorflow-Core-Protobuf-Config.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Protobuf.Config</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_Proto-Tensorflow-Core-Protobuf-Config.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/Proto-Tensorflow-Core-Protobuf-Config.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">Proto.Tensorflow.Core.Protobuf.Config</p></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:ConfigProto" class="def">ConfigProto</a> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#ConfigProto" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:ConfigProto" class="def">ConfigProto</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_ConfigProto-39-deviceCount" class="def">_ConfigProto'deviceCount</a> :: <a href="../containers-0.5.6.2/Data-Map-Lazy.html#t:Map">Map</a> <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_ConfigProto-39-intraOpParallelismThreads" class="def">_ConfigProto'intraOpParallelismThreads</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_ConfigProto-39-interOpParallelismThreads" class="def">_ConfigProto'interOpParallelismThreads</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_ConfigProto-39-usePerSessionThreads" class="def">_ConfigProto'usePerSessionThreads</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_ConfigProto-39-sessionInterOpThreadPool" class="def">_ConfigProto'sessionInterOpThreadPool</a> :: [<a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ThreadPoolOptionProto">ThreadPoolOptionProto</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_ConfigProto-39-placementPeriod" class="def">_ConfigProto'placementPeriod</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_ConfigProto-39-deviceFilters" class="def">_ConfigProto'deviceFilters</a> :: [<a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_ConfigProto-39-gpuOptions" class="def">_ConfigProto'gpuOptions</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_ConfigProto-39-allowSoftPlacement" class="def">_ConfigProto'allowSoftPlacement</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_ConfigProto-39-logDevicePlacement" class="def">_ConfigProto'logDevicePlacement</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_ConfigProto-39-graphOptions" class="def">_ConfigProto'graphOptions</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_ConfigProto-39-operationTimeoutInMs" class="def">_ConfigProto'operationTimeoutInMs</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:ConfigProto" class="caption collapser" onclick="toggleSection('i:ConfigProto')">Instances</p><div id="section.i:ConfigProto" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-42" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-42" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-177" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-194" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;allowSoftPlacement&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-134" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;deviceCount&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-47" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;deviceFilters&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-107" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;gpuOptions&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-116" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;graphOptions&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-152" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;interOpParallelismThreads&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-68" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;intraOpParallelismThreads&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-57" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;logDevicePlacement&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-143" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'gpuOptions&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-125" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'graphOptions&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-161" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;operationTimeoutInMs&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-171" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;placementPeriod&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-98" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;sessionInterOpThreadPool&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-89" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;usePerSessionThreads&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-79" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;allowSoftPlacement&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-131" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;deviceCount&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> = <a href="../containers-0.5.6.2/Data-Map-Lazy.html#t:Map">Map</a> <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a> <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-44" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;deviceFilters&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> = [<a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a>]</span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-104" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;gpuOptions&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> = <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-113" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;graphOptions&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> = <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-149" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;interOpParallelismThreads&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-64" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;intraOpParallelismThreads&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-53" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;logDevicePlacement&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-140" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'gpuOptions&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-122" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'graphOptions&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-158" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;operationTimeoutInMs&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-167" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;placementPeriod&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-95" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;sessionInterOpThreadPool&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> = [<a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ThreadPoolOptionProto">ThreadPoolOptionProto</a>]</span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-85" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;usePerSessionThreads&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">ConfigProto</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-75" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:ConfigProto-39-DeviceCountEntry" class="def">ConfigProto'DeviceCountEntry</a> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#ConfigProto%27DeviceCountEntry" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:ConfigProto-39-DeviceCountEntry" class="def">ConfigProto'DeviceCountEntry</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_ConfigProto-39-DeviceCountEntry-39-key" class="def">_ConfigProto'DeviceCountEntry'key</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_ConfigProto-39-DeviceCountEntry-39-value" class="def">_ConfigProto'DeviceCountEntry'value</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:ConfigProto-39-DeviceCountEntry" class="caption collapser" onclick="toggleSection('i:ConfigProto-39-DeviceCountEntry')">Instances</p><div id="section.i:ConfigProto-39-DeviceCountEntry" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto-39-DeviceCountEntry">ConfigProto'DeviceCountEntry</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-305" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto-39-DeviceCountEntry">ConfigProto'DeviceCountEntry</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-305" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto-39-DeviceCountEntry">ConfigProto'DeviceCountEntry</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-327" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto-39-DeviceCountEntry">ConfigProto'DeviceCountEntry</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-335" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;key&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto-39-DeviceCountEntry">ConfigProto'DeviceCountEntry</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto-39-DeviceCountEntry">ConfigProto'DeviceCountEntry</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-311" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;value&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto-39-DeviceCountEntry">ConfigProto'DeviceCountEntry</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto-39-DeviceCountEntry">ConfigProto'DeviceCountEntry</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-321" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;key&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto-39-DeviceCountEntry">ConfigProto'DeviceCountEntry</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-307" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;value&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto-39-DeviceCountEntry">ConfigProto'DeviceCountEntry</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-317" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:DebugTensorWatch" class="def">DebugTensorWatch</a> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#DebugTensorWatch" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:DebugTensorWatch" class="def">DebugTensorWatch</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_DebugTensorWatch-39-nodeName" class="def">_DebugTensorWatch'nodeName</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_DebugTensorWatch-39-outputSlot" class="def">_DebugTensorWatch'outputSlot</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_DebugTensorWatch-39-debugOps" class="def">_DebugTensorWatch'debugOps</a> :: [<a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_DebugTensorWatch-39-debugUrls" class="def">_DebugTensorWatch'debugUrls</a> :: [<a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a>]</dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:DebugTensorWatch" class="caption collapser" onclick="toggleSection('i:DebugTensorWatch')">Instances</p><div id="section.i:DebugTensorWatch" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-361" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-361" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-399" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-406" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;debugOps&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-384" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;debugUrls&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-393" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;nodeName&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-366" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;outputSlot&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-375" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;debugOps&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a> = [<a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a>]</span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-381" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;debugUrls&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a> = [<a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a>]</span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-390" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;nodeName&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-363" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;outputSlot&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-372" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:GPUOptions" class="def">GPUOptions</a> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#GPUOptions" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:GPUOptions" class="def">GPUOptions</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_GPUOptions-39-perProcessGpuMemoryFraction" class="def">_GPUOptions'perProcessGpuMemoryFraction</a> :: <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_GPUOptions-39-allocatorType" class="def">_GPUOptions'allocatorType</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_GPUOptions-39-deferredDeletionBytes" class="def">_GPUOptions'deferredDeletionBytes</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_GPUOptions-39-allowGrowth" class="def">_GPUOptions'allowGrowth</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_GPUOptions-39-visibleDeviceList" class="def">_GPUOptions'visibleDeviceList</a> :: <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:GPUOptions" class="caption collapser" onclick="toggleSection('i:GPUOptions')">Instances</p><div id="section.i:GPUOptions" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-447" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-447" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-497" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-506" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;allocatorType&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-463" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;allowGrowth&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-482" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;deferredDeletionBytes&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-473" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;perProcessGpuMemoryFraction&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-453" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;visibleDeviceList&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-491" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;allocatorType&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-460" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;allowGrowth&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-479" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;deferredDeletionBytes&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-469" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;perProcessGpuMemoryFraction&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a> = <a href="../base-4.8.2.0/Prelude.html#t:Double">Double</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-449" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;visibleDeviceList&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">GPUOptions</a> = <a href="../text-1.2.2.1/Data-Text.html#t:Text">Text</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-488" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:GraphOptions" class="def">GraphOptions</a> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#GraphOptions" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:GraphOptions" class="def">GraphOptions</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_GraphOptions-39-enableRecvScheduling" class="def">_GraphOptions'enableRecvScheduling</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_GraphOptions-39-optimizerOptions" class="def">_GraphOptions'optimizerOptions</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_GraphOptions-39-buildCostModel" class="def">_GraphOptions'buildCostModel</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_GraphOptions-39-buildCostModelAfter" class="def">_GraphOptions'buildCostModelAfter</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_GraphOptions-39-inferShapes" class="def">_GraphOptions'inferShapes</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_GraphOptions-39-placePrunedGraph" class="def">_GraphOptions'placePrunedGraph</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_GraphOptions-39-enableBfloat16Sendrecv" class="def">_GraphOptions'enableBfloat16Sendrecv</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_GraphOptions-39-timelineStep" class="def">_GraphOptions'timelineStep</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:GraphOptions" class="caption collapser" onclick="toggleSection('i:GraphOptions')">Instances</p><div id="section.i:GraphOptions" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-563" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-563" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-650" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-662" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;buildCostModel&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-597" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;buildCostModelAfter&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-607" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;enableBfloat16Sendrecv&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-635" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;enableRecvScheduling&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-569" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;inferShapes&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-616" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'optimizerOptions&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-588" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;optimizerOptions&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-578" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;placePrunedGraph&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-625" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;timelineStep&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-644" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;buildCostModel&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-594" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;buildCostModelAfter&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-603" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;enableBfloat16Sendrecv&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-631" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;enableRecvScheduling&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-565" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;inferShapes&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-613" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'optimizerOptions&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> = <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-584" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;optimizerOptions&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> = <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-575" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;placePrunedGraph&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-622" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;timelineStep&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">GraphOptions</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-641" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:OptimizerOptions" class="def">OptimizerOptions</a> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#OptimizerOptions" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:OptimizerOptions" class="def">OptimizerOptions</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_OptimizerOptions-39-doCommonSubexpressionElimination" class="def">_OptimizerOptions'doCommonSubexpressionElimination</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OptimizerOptions-39-doConstantFolding" class="def">_OptimizerOptions'doConstantFolding</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OptimizerOptions-39-doFunctionInlining" class="def">_OptimizerOptions'doFunctionInlining</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_OptimizerOptions-39-optLevel" class="def">_OptimizerOptions'optLevel</a> :: <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions-39-Level">OptimizerOptions'Level</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:OptimizerOptions" class="caption collapser" onclick="toggleSection('i:OptimizerOptions')">Instances</p><div id="section.i:OptimizerOptions" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-736" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-736" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-780" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-788" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;doCommonSubexpressionElimination&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-743" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;doConstantFolding&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-755" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;doFunctionInlining&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-765" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;optLevel&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-774" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;doCommonSubexpressionElimination&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-738" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;doConstantFolding&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-751" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;doFunctionInlining&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-761" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;optLevel&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">OptimizerOptions</a> = <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions-39-Level">OptimizerOptions'Level</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-771" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:OptimizerOptions-39-Level" class="def">OptimizerOptions'Level</a> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#OptimizerOptions%27Level" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:OptimizerOptions-39-L0" class="def">OptimizerOptions'L0</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:OptimizerOptions-39-L1" class="def">OptimizerOptions'L1</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="subs instances"><p id="control.i:OptimizerOptions-39-Level" class="caption collapser" onclick="toggleSection('i:OptimizerOptions-39-Level')">Instances</p><div id="section.i:OptimizerOptions-39-Level" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Prelude.html#t:Bounded">Bounded</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions-39-Level">OptimizerOptions'Level</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-872" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Prelude.html#t:Enum">Enum</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions-39-Level">OptimizerOptions'Level</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-849" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions-39-Level">OptimizerOptions'Level</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-831" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions-39-Level">OptimizerOptions'Level</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-831" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions-39-Level">OptimizerOptions'Level</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-833" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">MessageEnum <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions-39-Level">OptimizerOptions'Level</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-839" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">FieldDefault <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions-39-Level">OptimizerOptions'Level</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-836" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:RunMetadata" class="def">RunMetadata</a> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#RunMetadata" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:RunMetadata" class="def">RunMetadata</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_RunMetadata-39-stepStats" class="def">_RunMetadata'stepStats</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> StepStats</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_RunMetadata-39-costGraph" class="def">_RunMetadata'costGraph</a> :: <a href="../base-4.8.2.0/Data-Maybe.html#t:Maybe">Maybe</a> CostGraphDef</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_RunMetadata-39-partitionGraphs" class="def">_RunMetadata'partitionGraphs</a> :: [<a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a>]</dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:RunMetadata" class="caption collapser" onclick="toggleSection('i:RunMetadata')">Instances</p><div id="section.i:RunMetadata" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-883" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-883" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-931" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-937" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;costGraph&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-906" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'costGraph&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-916" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;maybe'stepStats&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-897" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;partitionGraphs&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-925" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;stepStats&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-888" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;costGraph&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-903" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'costGraph&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-912" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;maybe'stepStats&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-894" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;partitionGraphs&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a> = [<a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">GraphDef</a>]</span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-922" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;stepStats&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">RunMetadata</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-885" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:RunOptions" class="def">RunOptions</a> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#RunOptions" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:RunOptions" class="def">RunOptions</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_RunOptions-39-traceLevel" class="def">_RunOptions'traceLevel</a> :: <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions-39-TraceLevel">RunOptions'TraceLevel</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_RunOptions-39-timeoutInMs" class="def">_RunOptions'timeoutInMs</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_RunOptions-39-interOpThreadPool" class="def">_RunOptions'interOpThreadPool</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_RunOptions-39-debugTensorWatchOpts" class="def">_RunOptions'debugTensorWatchOpts</a> :: [<a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a>]</dt><dd class="doc empty">&nbsp;</dd><dt class="src"><a name="v:_RunOptions-39-outputPartitionGraphs" class="def">_RunOptions'outputPartitionGraphs</a> :: <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:RunOptions" class="caption collapser" onclick="toggleSection('i:RunOptions')">Instances</p><div id="section.i:RunOptions" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-975" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-975" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1024" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1032" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;debugTensorWatchOpts&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1008" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;interOpThreadPool&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-998" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;outputPartitionGraphs&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1018" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;timeoutInMs&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-989" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;traceLevel&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-980" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;debugTensorWatchOpts&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a> = [<a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">DebugTensorWatch</a>]</span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1004" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;interOpThreadPool&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-995" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;outputPartitionGraphs&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a> = <a href="../base-4.8.2.0/Data-Bool.html#t:Bool">Bool</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1014" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;timeoutInMs&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-986" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;traceLevel&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">RunOptions</a> = <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions-39-TraceLevel">RunOptions'TraceLevel</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-977" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:RunOptions-39-TraceLevel" class="def">RunOptions'TraceLevel</a> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#RunOptions%27TraceLevel" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:RunOptions-39-NO_TRACE" class="def">RunOptions'NO_TRACE</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:RunOptions-39-SOFTWARE_TRACE" class="def">RunOptions'SOFTWARE_TRACE</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:RunOptions-39-HARDWARE_TRACE" class="def">RunOptions'HARDWARE_TRACE</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src"><a name="v:RunOptions-39-FULL_TRACE" class="def">RunOptions'FULL_TRACE</a></td><td class="doc empty">&nbsp;</td></tr></table></div><div class="subs instances"><p id="control.i:RunOptions-39-TraceLevel" class="caption collapser" onclick="toggleSection('i:RunOptions-39-TraceLevel')">Instances</p><div id="section.i:RunOptions-39-TraceLevel" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Prelude.html#t:Bounded">Bounded</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions-39-TraceLevel">RunOptions'TraceLevel</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1136" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Prelude.html#t:Enum">Enum</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions-39-TraceLevel">RunOptions'TraceLevel</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1107" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions-39-TraceLevel">RunOptions'TraceLevel</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1083" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions-39-TraceLevel">RunOptions'TraceLevel</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1083" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions-39-TraceLevel">RunOptions'TraceLevel</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1085" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">MessageEnum <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions-39-TraceLevel">RunOptions'TraceLevel</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1091" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">FieldDefault <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions-39-TraceLevel">RunOptions'TraceLevel</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1088" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:ThreadPoolOptionProto" class="def">ThreadPoolOptionProto</a> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#ThreadPoolOptionProto" class="link">Source</a></p><div class="subs constructors"><p class="caption">Constructors</p><table><tr><td class="src"><a name="v:ThreadPoolOptionProto" class="def">ThreadPoolOptionProto</a></td><td class="doc empty">&nbsp;</td></tr><tr><td colspan="2"><div class="subs fields"><p class="caption">Fields</p><dl><dt class="src"><a name="v:_ThreadPoolOptionProto-39-numThreads" class="def">_ThreadPoolOptionProto'numThreads</a> :: <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></dt><dd class="doc empty">&nbsp;</dd></dl><div class="clear"></div></div></td></tr></table></div><div class="subs instances"><p id="control.i:ThreadPoolOptionProto" class="caption collapser" onclick="toggleSection('i:ThreadPoolOptionProto')">Instances</p><div id="section.i:ThreadPoolOptionProto" class="show"><table><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Data-Eq.html#t:Eq">Eq</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ThreadPoolOptionProto">ThreadPoolOptionProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1142" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../base-4.8.2.0/Text-Show.html#t:Show">Show</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ThreadPoolOptionProto">ThreadPoolOptionProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1142" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><a href="../data-default-class-0.0.1/Data-Default-Class.html#t:Default">Default</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ThreadPoolOptionProto">ThreadPoolOptionProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1154" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">Message <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ThreadPoolOptionProto">ThreadPoolOptionProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1159" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left">HasField &quot;numThreads&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ThreadPoolOptionProto">ThreadPoolOptionProto</a> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ThreadPoolOptionProto">ThreadPoolOptionProto</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1148" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr><tr><td class="src clearfix"><span class="inst-left"><span class="keyword">type</span> Field &quot;numThreads&quot; <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ThreadPoolOptionProto">ThreadPoolOptionProto</a> = <a href="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></span> <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#line-1144" class="link">Source</a></td><td class="doc empty">&nbsp;</td></tr></table></div></div></div><div class="top"><p class="src"><a name="v:allocatorType" class="def">allocatorType</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;allocatorType&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;allocatorType&quot; msg) (Field &quot;allocatorType&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#allocatorType" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:allowGrowth" class="def">allowGrowth</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;allowGrowth&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;allowGrowth&quot; msg) (Field &quot;allowGrowth&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#allowGrowth" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:allowSoftPlacement" class="def">allowSoftPlacement</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;allowSoftPlacement&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;allowSoftPlacement&quot; msg) (Field &quot;allowSoftPlacement&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#allowSoftPlacement" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:buildCostModel" class="def">buildCostModel</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;buildCostModel&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;buildCostModel&quot; msg) (Field &quot;buildCostModel&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#buildCostModel" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:buildCostModelAfter" class="def">buildCostModelAfter</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;buildCostModelAfter&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;buildCostModelAfter&quot; msg) (Field &quot;buildCostModelAfter&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#buildCostModelAfter" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:costGraph" class="def">costGraph</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;costGraph&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;costGraph&quot; msg) (Field &quot;costGraph&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#costGraph" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:debugOps" class="def">debugOps</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;debugOps&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;debugOps&quot; msg) (Field &quot;debugOps&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#debugOps" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:debugTensorWatchOpts" class="def">debugTensorWatchOpts</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;debugTensorWatchOpts&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;debugTensorWatchOpts&quot; msg) (Field &quot;debugTensorWatchOpts&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#debugTensorWatchOpts" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:debugUrls" class="def">debugUrls</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;debugUrls&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;debugUrls&quot; msg) (Field &quot;debugUrls&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#debugUrls" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:deferredDeletionBytes" class="def">deferredDeletionBytes</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;deferredDeletionBytes&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;deferredDeletionBytes&quot; msg) (Field &quot;deferredDeletionBytes&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#deferredDeletionBytes" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:deviceCount" class="def">deviceCount</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;deviceCount&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;deviceCount&quot; msg) (Field &quot;deviceCount&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#deviceCount" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:deviceFilters" class="def">deviceFilters</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;deviceFilters&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;deviceFilters&quot; msg) (Field &quot;deviceFilters&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#deviceFilters" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:doCommonSubexpressionElimination" class="def">doCommonSubexpressionElimination</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;doCommonSubexpressionElimination&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;doCommonSubexpressionElimination&quot; msg) (Field &quot;doCommonSubexpressionElimination&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#doCommonSubexpressionElimination" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:doConstantFolding" class="def">doConstantFolding</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;doConstantFolding&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;doConstantFolding&quot; msg) (Field &quot;doConstantFolding&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#doConstantFolding" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:doFunctionInlining" class="def">doFunctionInlining</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;doFunctionInlining&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;doFunctionInlining&quot; msg) (Field &quot;doFunctionInlining&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#doFunctionInlining" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:enableBfloat16Sendrecv" class="def">enableBfloat16Sendrecv</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;enableBfloat16Sendrecv&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;enableBfloat16Sendrecv&quot; msg) (Field &quot;enableBfloat16Sendrecv&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#enableBfloat16Sendrecv" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:enableRecvScheduling" class="def">enableRecvScheduling</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;enableRecvScheduling&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;enableRecvScheduling&quot; msg) (Field &quot;enableRecvScheduling&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#enableRecvScheduling" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:gpuOptions" class="def">gpuOptions</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;gpuOptions&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;gpuOptions&quot; msg) (Field &quot;gpuOptions&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#gpuOptions" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:graphOptions" class="def">graphOptions</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;graphOptions&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;graphOptions&quot; msg) (Field &quot;graphOptions&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#graphOptions" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:inferShapes" class="def">inferShapes</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;inferShapes&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;inferShapes&quot; msg) (Field &quot;inferShapes&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#inferShapes" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:interOpParallelismThreads" class="def">interOpParallelismThreads</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;interOpParallelismThreads&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;interOpParallelismThreads&quot; msg) (Field &quot;interOpParallelismThreads&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#interOpParallelismThreads" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:interOpThreadPool" class="def">interOpThreadPool</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;interOpThreadPool&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;interOpThreadPool&quot; msg) (Field &quot;interOpThreadPool&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#interOpThreadPool" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:intraOpParallelismThreads" class="def">intraOpParallelismThreads</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;intraOpParallelismThreads&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;intraOpParallelismThreads&quot; msg) (Field &quot;intraOpParallelismThreads&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#intraOpParallelismThreads" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:key" class="def">key</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;key&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;key&quot; msg) (Field &quot;key&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#key" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:logDevicePlacement" class="def">logDevicePlacement</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;logDevicePlacement&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;logDevicePlacement&quot; msg) (Field &quot;logDevicePlacement&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#logDevicePlacement" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-costGraph" class="def">maybe'costGraph</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'costGraph&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'costGraph&quot; msg) (Field &quot;maybe'costGraph&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#maybe%27costGraph" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-gpuOptions" class="def">maybe'gpuOptions</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'gpuOptions&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'gpuOptions&quot; msg) (Field &quot;maybe'gpuOptions&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#maybe%27gpuOptions" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-graphOptions" class="def">maybe'graphOptions</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'graphOptions&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'graphOptions&quot; msg) (Field &quot;maybe'graphOptions&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#maybe%27graphOptions" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-optimizerOptions" class="def">maybe'optimizerOptions</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'optimizerOptions&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'optimizerOptions&quot; msg) (Field &quot;maybe'optimizerOptions&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#maybe%27optimizerOptions" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:maybe-39-stepStats" class="def">maybe'stepStats</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;maybe'stepStats&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;maybe'stepStats&quot; msg) (Field &quot;maybe'stepStats&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#maybe%27stepStats" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:nodeName" class="def">nodeName</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;nodeName&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;nodeName&quot; msg) (Field &quot;nodeName&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#nodeName" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:numThreads" class="def">numThreads</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;numThreads&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;numThreads&quot; msg) (Field &quot;numThreads&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#numThreads" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:operationTimeoutInMs" class="def">operationTimeoutInMs</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;operationTimeoutInMs&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;operationTimeoutInMs&quot; msg) (Field &quot;operationTimeoutInMs&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#operationTimeoutInMs" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:optLevel" class="def">optLevel</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;optLevel&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;optLevel&quot; msg) (Field &quot;optLevel&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#optLevel" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:optimizerOptions" class="def">optimizerOptions</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;optimizerOptions&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;optimizerOptions&quot; msg) (Field &quot;optimizerOptions&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#optimizerOptions" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:outputPartitionGraphs" class="def">outputPartitionGraphs</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;outputPartitionGraphs&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;outputPartitionGraphs&quot; msg) (Field &quot;outputPartitionGraphs&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#outputPartitionGraphs" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:outputSlot" class="def">outputSlot</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;outputSlot&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;outputSlot&quot; msg) (Field &quot;outputSlot&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#outputSlot" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:partitionGraphs" class="def">partitionGraphs</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;partitionGraphs&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;partitionGraphs&quot; msg) (Field &quot;partitionGraphs&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#partitionGraphs" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:perProcessGpuMemoryFraction" class="def">perProcessGpuMemoryFraction</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;perProcessGpuMemoryFraction&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;perProcessGpuMemoryFraction&quot; msg) (Field &quot;perProcessGpuMemoryFraction&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#perProcessGpuMemoryFraction" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:placePrunedGraph" class="def">placePrunedGraph</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;placePrunedGraph&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;placePrunedGraph&quot; msg) (Field &quot;placePrunedGraph&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#placePrunedGraph" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:placementPeriod" class="def">placementPeriod</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;placementPeriod&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;placementPeriod&quot; msg) (Field &quot;placementPeriod&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#placementPeriod" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:sessionInterOpThreadPool" class="def">sessionInterOpThreadPool</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;sessionInterOpThreadPool&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;sessionInterOpThreadPool&quot; msg) (Field &quot;sessionInterOpThreadPool&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#sessionInterOpThreadPool" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:stepStats" class="def">stepStats</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;stepStats&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;stepStats&quot; msg) (Field &quot;stepStats&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#stepStats" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:timelineStep" class="def">timelineStep</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;timelineStep&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;timelineStep&quot; msg) (Field &quot;timelineStep&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#timelineStep" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:timeoutInMs" class="def">timeoutInMs</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;timeoutInMs&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;timeoutInMs&quot; msg) (Field &quot;timeoutInMs&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#timeoutInMs" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:traceLevel" class="def">traceLevel</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;traceLevel&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;traceLevel&quot; msg) (Field &quot;traceLevel&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#traceLevel" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:usePerSessionThreads" class="def">usePerSessionThreads</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;usePerSessionThreads&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;usePerSessionThreads&quot; msg) (Field &quot;usePerSessionThreads&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#usePerSessionThreads" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:value" class="def">value</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;value&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;value&quot; msg) (Field &quot;value&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#value" class="link">Source</a></p></div><div class="top"><p class="src"><a name="v:visibleDeviceList" class="def">visibleDeviceList</a> :: <span class="keyword">forall</span> msg msg'. HasField &quot;visibleDeviceList&quot; msg msg' =&gt; <a href="../lens-family-1.2.0/Lens-Family2-Unchecked.html#t:Lens">Lens</a> msg msg' (Field &quot;visibleDeviceList&quot; msg) (Field &quot;visibleDeviceList&quot; msg') <a href="src/Proto-Tensorflow-Core-Protobuf-Config.html#visibleDeviceList" class="link">Source</a></p></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-95.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-95.html
new file mode 100644
index 0000000..141d60d
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-95.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - _)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - _</p><table><tr><td class="src">_AttrValue'b</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-b">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'f</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-f">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'func</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-func">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'i</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-i">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'list</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-list">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'b</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-b">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'f</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-f">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'i</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-i">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue's</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-s">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'shape</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-shape">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'tensor</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-tensor">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'type'</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-type-39-">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'placeholder</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-placeholder">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue's</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-s">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'shape</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-shape">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'tensor</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-tensor">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'type'</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-type-39-">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_ConfigProto'allowSoftPlacement</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-allowSoftPlacement">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'deviceCount</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-deviceCount">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'DeviceCountEntry'key</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-DeviceCountEntry-39-key">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'DeviceCountEntry'value</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-DeviceCountEntry-39-value">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'deviceFilters</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-deviceFilters">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'gpuOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-gpuOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'graphOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-graphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'interOpParallelismThreads</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-interOpParallelismThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'intraOpParallelismThreads</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-intraOpParallelismThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'logDevicePlacement</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-logDevicePlacement">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'operationTimeoutInMs</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-operationTimeoutInMs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'placementPeriod</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-placementPeriod">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'sessionInterOpThreadPool</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-sessionInterOpThreadPool">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'usePerSessionThreads</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-usePerSessionThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_DebugTensorWatch'debugOps</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_DebugTensorWatch-39-debugOps">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_DebugTensorWatch'debugUrls</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_DebugTensorWatch-39-debugUrls">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_DebugTensorWatch'nodeName</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_DebugTensorWatch-39-nodeName">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_DebugTensorWatch'outputSlot</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_DebugTensorWatch-39-outputSlot">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'allocatorType</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-allocatorType">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'allowGrowth</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-allowGrowth">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'deferredDeletionBytes</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-deferredDeletionBytes">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'perProcessGpuMemoryFraction</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-perProcessGpuMemoryFraction">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'visibleDeviceList</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-visibleDeviceList">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphDef'library</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:_GraphDef-39-library">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">_GraphDef'node</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:_GraphDef-39-node">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">_GraphDef'version</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:_GraphDef-39-version">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">_GraphDef'versions</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:_GraphDef-39-versions">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">_GraphOptions'buildCostModel</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-buildCostModel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'buildCostModelAfter</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-buildCostModelAfter">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'enableBfloat16Sendrecv</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-enableBfloat16Sendrecv">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'enableRecvScheduling</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-enableRecvScheduling">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'inferShapes</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-inferShapes">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'optimizerOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-optimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'placePrunedGraph</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-placePrunedGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'timelineStep</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-timelineStep">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_NameAttrList'attr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_NameAttrList-39-attr">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_NameAttrList'AttrEntry'key</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_NameAttrList-39-AttrEntry-39-key">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_NameAttrList'AttrEntry'value</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_NameAttrList-39-AttrEntry-39-value">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_NameAttrList'name</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_NameAttrList-39-name">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_NodeDef'attr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-attr">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'AttrEntry'key</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-AttrEntry-39-key">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'AttrEntry'value</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-AttrEntry-39-value">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'device</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-device">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'input</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-input">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'name</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-name">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'op</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-op">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_OpDef'allowsUninitializedInput</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-allowsUninitializedInput">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'description</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-description">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'isRef</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-isRef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'name</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-name">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'numberAttr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-numberAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'type'</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-type-39-">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'typeAttr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-typeAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'typeListAttr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-typeListAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'attr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-attr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'allowedValues</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-allowedValues">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'defaultValue</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-defaultValue">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'description</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-description">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'hasMinimum</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-hasMinimum">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'minimum</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-minimum">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'name</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-name">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'type'</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-type-39-">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'deprecation</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-deprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'description</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-description">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'inputArg</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-inputArg">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'isAggregate</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-isAggregate">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'isCommutative</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-isCommutative">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'isStateful</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-isStateful">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'name</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-name">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'outputArg</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-outputArg">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'summary</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-summary">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDeprecation'explanation</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDeprecation-39-explanation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDeprecation'version</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDeprecation-39-version">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpList'op</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpList-39-op">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OptimizerOptions'doCommonSubexpressionElimination</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_OptimizerOptions-39-doCommonSubexpressionElimination">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_OptimizerOptions'doConstantFolding</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_OptimizerOptions-39-doConstantFolding">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_OptimizerOptions'doFunctionInlining</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_OptimizerOptions-39-doFunctionInlining">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_OptimizerOptions'optLevel</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_OptimizerOptions-39-optLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ResourceHandle'container</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-container">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_ResourceHandle'device</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-device">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_ResourceHandle'hashCode</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-hashCode">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_ResourceHandle'maybeTypeName</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-maybeTypeName">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_ResourceHandle'name</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-name">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_RunMetadata'costGraph</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunMetadata-39-costGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunMetadata'partitionGraphs</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunMetadata-39-partitionGraphs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunMetadata'stepStats</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunMetadata-39-stepStats">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'debugTensorWatchOpts</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-debugTensorWatchOpts">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'interOpThreadPool</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-interOpThreadPool">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'outputPartitionGraphs</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-outputPartitionGraphs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'timeoutInMs</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-timeoutInMs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'traceLevel</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-traceLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_TensorProto'boolVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-boolVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'dcomplexVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-dcomplexVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'doubleVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-doubleVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'dtype</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-dtype">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'floatVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-floatVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'halfVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-halfVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'int64Val</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-int64Val">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'intVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-intVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'resourceHandleVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-resourceHandleVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'scomplexVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-scomplexVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'stringVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-stringVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'tensorContent</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-tensorContent">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'tensorShape</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-tensorShape">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'versionNumber</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-versionNumber">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorShapeProto'dim</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:_TensorShapeProto-39-dim">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">_TensorShapeProto'Dim'name</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:_TensorShapeProto-39-Dim-39-name">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">_TensorShapeProto'Dim'size</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:_TensorShapeProto-39-Dim-39-size">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">_TensorShapeProto'unknownRank</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:_TensorShapeProto-39-unknownRank">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">_ThreadPoolOptionProto'numThreads</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ThreadPoolOptionProto-39-numThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-A.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-A.html
new file mode 100644
index 0000000..ee12573
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-A.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - A)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - A</p><table><tr><td class="src">allocatorType</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:allocatorType">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">allowedValues</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:allowedValues">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">allowGrowth</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:allowGrowth">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">allowSoftPlacement</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:allowSoftPlacement">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">allowsUninitializedInput</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:allowsUninitializedInput">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">attr</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:attr">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:attr">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:attr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">AttrValue</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:AttrValue">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">AttrValue'ListValue</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:AttrValue-39-ListValue">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-All.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-All.html
new file mode 100644
index 0000000..2fa311b
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-All.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="index"><p class="caption">Index</p><table><tr><td class="src">allocatorType</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:allocatorType">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">allowedValues</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:allowedValues">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">allowGrowth</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:allowGrowth">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">allowSoftPlacement</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:allowSoftPlacement">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">allowsUninitializedInput</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:allowsUninitializedInput">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">attr</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:attr">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:attr">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:attr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">AttrValue</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:AttrValue">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">AttrValue'ListValue</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:AttrValue-39-ListValue">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">b</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:b">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">boolVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:boolVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">buildCostModel</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:buildCostModel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">buildCostModelAfter</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:buildCostModelAfter">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">ConfigProto</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:ConfigProto">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">ConfigProto'DeviceCountEntry</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto-39-DeviceCountEntry">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:ConfigProto-39-DeviceCountEntry">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">container</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:container">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">costGraph</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:costGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">DataType</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">dcomplexVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:dcomplexVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">debugOps</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:debugOps">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">DebugTensorWatch</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:DebugTensorWatch">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">debugTensorWatchOpts</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:debugTensorWatchOpts">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">debugUrls</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:debugUrls">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">defaultValue</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:defaultValue">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">deferredDeletionBytes</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:deferredDeletionBytes">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">deprecation</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:deprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">description</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:description">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">device</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:device">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:device">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">deviceCount</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:deviceCount">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">deviceFilters</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:deviceFilters">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">dim</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:dim">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">doCommonSubexpressionElimination</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:doCommonSubexpressionElimination">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">doConstantFolding</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:doConstantFolding">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">doFunctionInlining</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:doFunctionInlining">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">doubleVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:doubleVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">dtype</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:dtype">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">DT_BFLOAT16</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_BFLOAT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_BFLOAT16_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_BFLOAT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_BOOL</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_BOOL">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_BOOL_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_BOOL_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_COMPLEX128</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_COMPLEX128">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_COMPLEX128_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_COMPLEX128_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_COMPLEX64</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_COMPLEX64">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_COMPLEX64_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_COMPLEX64_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_DOUBLE</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_DOUBLE">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_DOUBLE_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_DOUBLE_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_FLOAT</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_FLOAT">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_FLOAT_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_FLOAT_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_HALF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_HALF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_HALF_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_HALF_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT16</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT16_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT32</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT32">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT32_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT32_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT64</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT64">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT64_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT64_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT8</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT8">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT8_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT8_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INVALID</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INVALID">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT16</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT16_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT32</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT32">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT32_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT32_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT8</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT8">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT8_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT8_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QUINT16</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QUINT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QUINT16_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QUINT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QUINT8</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QUINT8">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QUINT8_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QUINT8_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_RESOURCE</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_RESOURCE">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_RESOURCE_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_RESOURCE_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_STRING</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_STRING">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_STRING_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_STRING_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_UINT16</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_UINT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_UINT16_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_UINT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_UINT8</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_UINT8">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_UINT8_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_UINT8_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">enableBfloat16Sendrecv</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:enableBfloat16Sendrecv">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">enableRecvScheduling</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:enableRecvScheduling">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">explanation</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:explanation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">f</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:f">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">floatVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:floatVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">func</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:func">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">GPUOptions</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:GPUOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">gpuOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:gpuOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">GraphDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:GraphDef">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">GraphOptions</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:GraphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">graphOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:graphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">halfVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:halfVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">hashCode</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:hashCode">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">hasMinimum</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:hasMinimum">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">i</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:i">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">inferShapes</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:inferShapes">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">input</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:input">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">inputArg</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:inputArg">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">int64Val</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:int64Val">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">interOpParallelismThreads</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:interOpParallelismThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">interOpThreadPool</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:interOpThreadPool">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">intraOpParallelismThreads</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:intraOpParallelismThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">intVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:intVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">isAggregate</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:isAggregate">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">isCommutative</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:isCommutative">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">isRef</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:isRef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">isStateful</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:isStateful">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">key</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:key">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:key">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:key">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">library</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:library">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">list</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:list">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">logDevicePlacement</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:logDevicePlacement">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'allowedValues</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:maybe-39-allowedValues">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">maybe'b</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-b">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'costGraph</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-costGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'defaultValue</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:maybe-39-defaultValue">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">maybe'deprecation</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:maybe-39-deprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">maybe'f</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-f">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'func</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-func">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'gpuOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-gpuOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'graphOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-graphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'i</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-i">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'library</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:maybe-39-library">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">maybe'list</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-list">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'optimizerOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-optimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'placeholder</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-placeholder">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe's</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-s">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'shape</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-shape">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'stepStats</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-stepStats">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'tensor</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-tensor">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'tensorShape</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:maybe-39-tensorShape">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">maybe'type'</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-type-39-">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'value</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-value">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:maybe-39-value">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">maybe'versions</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:maybe-39-versions">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">maybeTypeName</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:maybeTypeName">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">minimum</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:minimum">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">name</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:name">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:name">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:name">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">4 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:name">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">5 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:name">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">NameAttrList</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:NameAttrList">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">NameAttrList'AttrEntry</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:NameAttrList-39-AttrEntry">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">node</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:node">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">NodeDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:NodeDef">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">NodeDef'AttrEntry</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:NodeDef-39-AttrEntry">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">nodeName</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:nodeName">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">numberAttr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:numberAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">numThreads</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:numThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">op</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:op">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:op">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">OpDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:OpDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">OpDef'ArgDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:OpDef-39-ArgDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">OpDef'AttrDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:OpDef-39-AttrDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">OpDeprecation</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:OpDeprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">operationTimeoutInMs</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:operationTimeoutInMs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OpList</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpList">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:OpList">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">OptimizerOptions</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:OptimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">optimizerOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:optimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OptimizerOptions'L0</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:OptimizerOptions-39-L0">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OptimizerOptions'L1</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:OptimizerOptions-39-L1">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OptimizerOptions'Level</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions-39-Level">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">optLevel</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:optLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">outputArg</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:outputArg">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">outputPartitionGraphs</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:outputPartitionGraphs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">outputSlot</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:outputSlot">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">partitionGraphs</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:partitionGraphs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">perProcessGpuMemoryFraction</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:perProcessGpuMemoryFraction">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">placeholder</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:placeholder">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">placementPeriod</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:placementPeriod">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">placePrunedGraph</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:placePrunedGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">ResourceHandle</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:ResourceHandle">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">resourceHandleVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:resourceHandleVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">RunMetadata</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:RunMetadata">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'FULL_TRACE</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions-39-FULL_TRACE">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'HARDWARE_TRACE</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions-39-HARDWARE_TRACE">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'NO_TRACE</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions-39-NO_TRACE">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'SOFTWARE_TRACE</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions-39-SOFTWARE_TRACE">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'TraceLevel</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions-39-TraceLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">s</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:s">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">scomplexVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:scomplexVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">sessionInterOpThreadPool</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:sessionInterOpThreadPool">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">shape</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:shape">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">size</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:size">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">stepStats</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:stepStats">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">stringVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:stringVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">summary</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:summary">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">tensor</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:tensor">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">tensorContent</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:tensorContent">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">TensorProto</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:TensorProto">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">tensorShape</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:tensorShape">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">TensorShapeProto</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:TensorShapeProto">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">TensorShapeProto'Dim</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:TensorShapeProto-39-Dim">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">ThreadPoolOptionProto</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ThreadPoolOptionProto">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:ThreadPoolOptionProto">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">timelineStep</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:timelineStep">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">timeoutInMs</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:timeoutInMs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">traceLevel</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:traceLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">type'</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:type-39-">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:type-39-">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">typeAttr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:typeAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">typeListAttr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:typeListAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">unknownRank</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:unknownRank">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">usePerSessionThreads</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:usePerSessionThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">value</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:value">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:value">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:value">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">version</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:version">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:version">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">versionNumber</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:versionNumber">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">versions</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:versions">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">visibleDeviceList</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:visibleDeviceList">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_AttrValue'b</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-b">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'f</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-f">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'func</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-func">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'i</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-i">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'list</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-list">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'b</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-b">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'f</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-f">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'i</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-i">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue's</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-s">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'shape</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-shape">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'tensor</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-tensor">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'ListValue'type'</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-ListValue-39-type-39-">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'placeholder</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-placeholder">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue's</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-s">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'shape</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-shape">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'tensor</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-tensor">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_AttrValue'type'</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_AttrValue-39-type-39-">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_ConfigProto'allowSoftPlacement</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-allowSoftPlacement">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'deviceCount</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-deviceCount">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'DeviceCountEntry'key</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-DeviceCountEntry-39-key">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'DeviceCountEntry'value</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-DeviceCountEntry-39-value">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'deviceFilters</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-deviceFilters">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'gpuOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-gpuOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'graphOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-graphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'interOpParallelismThreads</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-interOpParallelismThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'intraOpParallelismThreads</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-intraOpParallelismThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'logDevicePlacement</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-logDevicePlacement">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'operationTimeoutInMs</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-operationTimeoutInMs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'placementPeriod</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-placementPeriod">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'sessionInterOpThreadPool</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-sessionInterOpThreadPool">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ConfigProto'usePerSessionThreads</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ConfigProto-39-usePerSessionThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_DebugTensorWatch'debugOps</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_DebugTensorWatch-39-debugOps">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_DebugTensorWatch'debugUrls</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_DebugTensorWatch-39-debugUrls">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_DebugTensorWatch'nodeName</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_DebugTensorWatch-39-nodeName">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_DebugTensorWatch'outputSlot</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_DebugTensorWatch-39-outputSlot">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'allocatorType</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-allocatorType">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'allowGrowth</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-allowGrowth">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'deferredDeletionBytes</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-deferredDeletionBytes">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'perProcessGpuMemoryFraction</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-perProcessGpuMemoryFraction">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GPUOptions'visibleDeviceList</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GPUOptions-39-visibleDeviceList">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphDef'library</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:_GraphDef-39-library">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">_GraphDef'node</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:_GraphDef-39-node">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">_GraphDef'version</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:_GraphDef-39-version">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">_GraphDef'versions</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:_GraphDef-39-versions">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">_GraphOptions'buildCostModel</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-buildCostModel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'buildCostModelAfter</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-buildCostModelAfter">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'enableBfloat16Sendrecv</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-enableBfloat16Sendrecv">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'enableRecvScheduling</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-enableRecvScheduling">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'inferShapes</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-inferShapes">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'optimizerOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-optimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'placePrunedGraph</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-placePrunedGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_GraphOptions'timelineStep</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_GraphOptions-39-timelineStep">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_NameAttrList'attr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_NameAttrList-39-attr">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_NameAttrList'AttrEntry'key</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_NameAttrList-39-AttrEntry-39-key">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_NameAttrList'AttrEntry'value</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_NameAttrList-39-AttrEntry-39-value">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_NameAttrList'name</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:_NameAttrList-39-name">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">_NodeDef'attr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-attr">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'AttrEntry'key</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-AttrEntry-39-key">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'AttrEntry'value</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-AttrEntry-39-value">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'device</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-device">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'input</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-input">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'name</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-name">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_NodeDef'op</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:_NodeDef-39-op">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">_OpDef'allowsUninitializedInput</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-allowsUninitializedInput">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'description</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-description">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'isRef</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-isRef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'name</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-name">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'numberAttr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-numberAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'type'</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-type-39-">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'typeAttr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-typeAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'ArgDef'typeListAttr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-ArgDef-39-typeListAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'attr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-attr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'allowedValues</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-allowedValues">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'defaultValue</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-defaultValue">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'description</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-description">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'hasMinimum</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-hasMinimum">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'minimum</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-minimum">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'name</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-name">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'AttrDef'type'</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-AttrDef-39-type-39-">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'deprecation</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-deprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'description</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-description">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'inputArg</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-inputArg">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'isAggregate</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-isAggregate">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'isCommutative</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-isCommutative">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'isStateful</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-isStateful">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'name</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-name">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'outputArg</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-outputArg">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDef'summary</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDef-39-summary">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDeprecation'explanation</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDeprecation-39-explanation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpDeprecation'version</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpDeprecation-39-version">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OpList'op</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:_OpList-39-op">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">_OptimizerOptions'doCommonSubexpressionElimination</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_OptimizerOptions-39-doCommonSubexpressionElimination">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_OptimizerOptions'doConstantFolding</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_OptimizerOptions-39-doConstantFolding">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_OptimizerOptions'doFunctionInlining</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_OptimizerOptions-39-doFunctionInlining">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_OptimizerOptions'optLevel</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_OptimizerOptions-39-optLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_ResourceHandle'container</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-container">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_ResourceHandle'device</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-device">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_ResourceHandle'hashCode</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-hashCode">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_ResourceHandle'maybeTypeName</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-maybeTypeName">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_ResourceHandle'name</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:_ResourceHandle-39-name">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">_RunMetadata'costGraph</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunMetadata-39-costGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunMetadata'partitionGraphs</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunMetadata-39-partitionGraphs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunMetadata'stepStats</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunMetadata-39-stepStats">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'debugTensorWatchOpts</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-debugTensorWatchOpts">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'interOpThreadPool</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-interOpThreadPool">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'outputPartitionGraphs</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-outputPartitionGraphs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'timeoutInMs</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-timeoutInMs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_RunOptions'traceLevel</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_RunOptions-39-traceLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">_TensorProto'boolVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-boolVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'dcomplexVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-dcomplexVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'doubleVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-doubleVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'dtype</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-dtype">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'floatVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-floatVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'halfVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-halfVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'int64Val</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-int64Val">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'intVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-intVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'resourceHandleVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-resourceHandleVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'scomplexVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-scomplexVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'stringVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-stringVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'tensorContent</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-tensorContent">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'tensorShape</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-tensorShape">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorProto'versionNumber</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:_TensorProto-39-versionNumber">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">_TensorShapeProto'dim</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:_TensorShapeProto-39-dim">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">_TensorShapeProto'Dim'name</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:_TensorShapeProto-39-Dim-39-name">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">_TensorShapeProto'Dim'size</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:_TensorShapeProto-39-Dim-39-size">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">_TensorShapeProto'unknownRank</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:_TensorShapeProto-39-unknownRank">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">_ThreadPoolOptionProto'numThreads</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:_ThreadPoolOptionProto-39-numThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-B.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-B.html
new file mode 100644
index 0000000..ae55437
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-B.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - B)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - B</p><table><tr><td class="src">b</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:b">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">boolVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:boolVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">buildCostModel</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:buildCostModel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">buildCostModelAfter</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:buildCostModelAfter">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-C.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-C.html
new file mode 100644
index 0000000..f072b65
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-C.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - C)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - C</p><table><tr><td class="src">ConfigProto</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:ConfigProto">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">ConfigProto'DeviceCountEntry</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto-39-DeviceCountEntry">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:ConfigProto-39-DeviceCountEntry">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">container</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:container">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">costGraph</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:costGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-D.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-D.html
new file mode 100644
index 0000000..9d5266e
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-D.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - D)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - D</p><table><tr><td class="src">DataType</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">dcomplexVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:dcomplexVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">debugOps</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:debugOps">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">DebugTensorWatch</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:DebugTensorWatch">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">debugTensorWatchOpts</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:debugTensorWatchOpts">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">debugUrls</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:debugUrls">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">defaultValue</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:defaultValue">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">deferredDeletionBytes</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:deferredDeletionBytes">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">deprecation</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:deprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">description</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:description">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">device</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:device">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:device">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">deviceCount</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:deviceCount">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">deviceFilters</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:deviceFilters">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">dim</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:dim">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">doCommonSubexpressionElimination</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:doCommonSubexpressionElimination">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">doConstantFolding</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:doConstantFolding">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">doFunctionInlining</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:doFunctionInlining">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">doubleVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:doubleVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">dtype</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:dtype">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">DT_BFLOAT16</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_BFLOAT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_BFLOAT16_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_BFLOAT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_BOOL</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_BOOL">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_BOOL_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_BOOL_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_COMPLEX128</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_COMPLEX128">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_COMPLEX128_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_COMPLEX128_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_COMPLEX64</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_COMPLEX64">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_COMPLEX64_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_COMPLEX64_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_DOUBLE</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_DOUBLE">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_DOUBLE_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_DOUBLE_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_FLOAT</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_FLOAT">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_FLOAT_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_FLOAT_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_HALF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_HALF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_HALF_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_HALF_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT16</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT16_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT32</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT32">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT32_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT32_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT64</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT64">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT64_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT64_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT8</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT8">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INT8_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INT8_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_INVALID</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_INVALID">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT16</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT16_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT32</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT32">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT32_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT32_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT8</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT8">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QINT8_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QINT8_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QUINT16</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QUINT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QUINT16_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QUINT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QUINT8</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QUINT8">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_QUINT8_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_QUINT8_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_RESOURCE</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_RESOURCE">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_RESOURCE_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_RESOURCE_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_STRING</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_STRING">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_STRING_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_STRING_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_UINT16</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_UINT16">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_UINT16_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_UINT16_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_UINT8</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_UINT8">Proto.Tensorflow.Core.Framework.Types</a></td></tr><tr><td class="src">DT_UINT8_REF</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html#v:DT_UINT8_REF">Proto.Tensorflow.Core.Framework.Types</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-E.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-E.html
new file mode 100644
index 0000000..75c8281
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-E.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - E)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - E</p><table><tr><td class="src">enableBfloat16Sendrecv</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:enableBfloat16Sendrecv">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">enableRecvScheduling</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:enableRecvScheduling">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">explanation</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:explanation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-F.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-F.html
new file mode 100644
index 0000000..38b9899
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-F.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - F)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - F</p><table><tr><td class="src">f</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:f">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">floatVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:floatVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">func</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:func">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-G.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-G.html
new file mode 100644
index 0000000..48b269f
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-G.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - G)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - G</p><table><tr><td class="src">GPUOptions</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:GPUOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">gpuOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:gpuOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">GraphDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:GraphDef">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">GraphOptions</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:GraphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">graphOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:graphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-H.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-H.html
new file mode 100644
index 0000000..e71b627
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-H.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - H)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - H</p><table><tr><td class="src">halfVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:halfVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">hashCode</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:hashCode">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">hasMinimum</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:hasMinimum">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-I.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-I.html
new file mode 100644
index 0000000..a29b263
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-I.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - I)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - I</p><table><tr><td class="src">i</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:i">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">inferShapes</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:inferShapes">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">input</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:input">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">inputArg</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:inputArg">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">int64Val</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:int64Val">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">interOpParallelismThreads</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:interOpParallelismThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">interOpThreadPool</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:interOpThreadPool">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">intraOpParallelismThreads</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:intraOpParallelismThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">intVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:intVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">isAggregate</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:isAggregate">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">isCommutative</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:isCommutative">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">isRef</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:isRef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">isStateful</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:isStateful">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-K.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-K.html
new file mode 100644
index 0000000..8a70b07
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-K.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - K)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - K</p><table><tr><td class="src">key</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:key">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:key">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:key">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-L.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-L.html
new file mode 100644
index 0000000..e9e6c5c
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-L.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - L)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - L</p><table><tr><td class="src">library</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:library">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">list</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:list">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">logDevicePlacement</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:logDevicePlacement">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-M.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-M.html
new file mode 100644
index 0000000..3e02ce7
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-M.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - M)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - M</p><table><tr><td class="src">maybe'allowedValues</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:maybe-39-allowedValues">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">maybe'b</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-b">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'costGraph</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-costGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'defaultValue</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:maybe-39-defaultValue">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">maybe'deprecation</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:maybe-39-deprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">maybe'f</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-f">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'func</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-func">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'gpuOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-gpuOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'graphOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-graphOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'i</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-i">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'library</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:maybe-39-library">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">maybe'list</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-list">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'optimizerOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-optimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'placeholder</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-placeholder">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe's</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-s">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'shape</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-shape">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'stepStats</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-stepStats">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">maybe'tensor</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-tensor">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'tensorShape</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:maybe-39-tensorShape">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">maybe'type'</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-type-39-">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">maybe'value</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-value">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:maybe-39-value">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">maybe'versions</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:maybe-39-versions">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">maybeTypeName</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:maybeTypeName">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">minimum</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:minimum">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-N.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-N.html
new file mode 100644
index 0000000..89af93f
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-N.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - N)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - N</p><table><tr><td class="src">name</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:name">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:name">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:name">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">4 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:name">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">5 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:name">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">NameAttrList</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:NameAttrList">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">NameAttrList'AttrEntry</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:NameAttrList-39-AttrEntry">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">node</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:node">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">NodeDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:NodeDef">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">NodeDef'AttrEntry</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:NodeDef-39-AttrEntry">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="src">nodeName</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:nodeName">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">numberAttr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:numberAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">numThreads</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:numThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-O.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-O.html
new file mode 100644
index 0000000..417b5cd
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-O.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - O)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - O</p><table><tr><td class="src">op</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:op">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:op">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">OpDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:OpDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">OpDef'ArgDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:OpDef-39-ArgDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">OpDef'AttrDef</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:OpDef-39-AttrDef">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">OpDeprecation</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:OpDeprecation">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">operationTimeoutInMs</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:operationTimeoutInMs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OpList</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpList">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:OpList">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">OptimizerOptions</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:OptimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">optimizerOptions</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:optimizerOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OptimizerOptions'L0</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:OptimizerOptions-39-L0">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OptimizerOptions'L1</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:OptimizerOptions-39-L1">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">OptimizerOptions'Level</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions-39-Level">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">optLevel</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:optLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">outputArg</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:outputArg">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">outputPartitionGraphs</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:outputPartitionGraphs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">outputSlot</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:outputSlot">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-P.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-P.html
new file mode 100644
index 0000000..27bb76a
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-P.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - P)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - P</p><table><tr><td class="src">partitionGraphs</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:partitionGraphs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">perProcessGpuMemoryFraction</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:perProcessGpuMemoryFraction">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">placeholder</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:placeholder">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">placementPeriod</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:placementPeriod">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">placePrunedGraph</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:placePrunedGraph">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-R.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-R.html
new file mode 100644
index 0000000..2a83347
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-R.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - R)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - R</p><table><tr><td class="src">ResourceHandle</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:ResourceHandle">Proto.Tensorflow.Core.Framework.ResourceHandle</a></td></tr><tr><td class="src">resourceHandleVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:resourceHandleVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">RunMetadata</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:RunMetadata">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'FULL_TRACE</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions-39-FULL_TRACE">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'HARDWARE_TRACE</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions-39-HARDWARE_TRACE">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'NO_TRACE</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions-39-NO_TRACE">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'SOFTWARE_TRACE</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:RunOptions-39-SOFTWARE_TRACE">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">RunOptions'TraceLevel</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions-39-TraceLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-S.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-S.html
new file mode 100644
index 0000000..cb40582
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-S.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - S)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - S</p><table><tr><td class="src">s</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:s">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">scomplexVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:scomplexVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">sessionInterOpThreadPool</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:sessionInterOpThreadPool">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">shape</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:shape">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">size</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:size">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">stepStats</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:stepStats">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">stringVal</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:stringVal">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">summary</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:summary">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-T.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-T.html
new file mode 100644
index 0000000..ac03d34
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-T.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - T)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - T</p><table><tr><td class="src">tensor</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:tensor">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="src">tensorContent</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:tensorContent">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">TensorProto</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:TensorProto">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">tensorShape</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:tensorShape">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">TensorShapeProto</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:TensorShapeProto">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">TensorShapeProto'Dim</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:TensorShapeProto-39-Dim">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">ThreadPoolOptionProto</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Type/Class)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ThreadPoolOptionProto">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="alt">2 (Data Constructor)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:ThreadPoolOptionProto">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">timelineStep</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:timelineStep">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">timeoutInMs</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:timeoutInMs">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">traceLevel</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:traceLevel">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">type'</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:type-39-">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:type-39-">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">typeAttr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:typeAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="src">typeListAttr</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:typeListAttr">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-U.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-U.html
new file mode 100644
index 0000000..c48aaf7
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-U.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - U)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - U</p><table><tr><td class="src">unknownRank</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:unknownRank">Proto.Tensorflow.Core.Framework.TensorShape</a></td></tr><tr><td class="src">usePerSessionThreads</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:usePerSessionThreads">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-V.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-V.html
new file mode 100644
index 0000000..1f9d2fe
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index-V.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index - V)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div><div id="index"><p class="caption">Index - V</p><table><tr><td class="src">value</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:value">Proto.Tensorflow.Core.Framework.AttrValue</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:value">Proto.Tensorflow.Core.Framework.NodeDef</a></td></tr><tr><td class="alt">3 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:value">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr><tr><td class="src">version</td><td>&nbsp;</td></tr><tr><td class="alt">1 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:version">Proto.Tensorflow.Core.Framework.OpDef</a></td></tr><tr><td class="alt">2 (Function)</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:version">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">versionNumber</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:versionNumber">Proto.Tensorflow.Core.Framework.Tensor</a></td></tr><tr><td class="src">versions</td><td class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:versions">Proto.Tensorflow.Core.Framework.Graph</a></td></tr><tr><td class="src">visibleDeviceList</td><td class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:visibleDeviceList">Proto.Tensorflow.Core.Protobuf.Config</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/doc-index.html b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index.html
new file mode 100644
index 0000000..a016a72
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/doc-index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers. (Index)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="alphabet"><ul><li><a href="doc-index-A.html">A</a></li><li><a href="doc-index-B.html">B</a></li><li><a href="doc-index-C.html">C</a></li><li><a href="doc-index-D.html">D</a></li><li><a href="doc-index-E.html">E</a></li><li><a href="doc-index-F.html">F</a></li><li><a href="doc-index-G.html">G</a></li><li><a href="doc-index-H.html">H</a></li><li><a href="doc-index-I.html">I</a></li><li><a href="doc-index-K.html">K</a></li><li><a href="doc-index-L.html">L</a></li><li><a href="doc-index-M.html">M</a></li><li><a href="doc-index-N.html">N</a></li><li><a href="doc-index-O.html">O</a></li><li><a href="doc-index-P.html">P</a></li><li><a href="doc-index-R.html">R</a></li><li><a href="doc-index-S.html">S</a></li><li><a href="doc-index-T.html">T</a></li><li><a href="doc-index-U.html">U</a></li><li><a href="doc-index-V.html">V</a></li><li><a href="doc-index-95.html">_</a></li><li><a href="doc-index-All.html">All</a></li></ul></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/frames.html b/docs/haddock/tensorflow-proto-0.1.0.0/frames.html
new file mode 100644
index 0000000..1b4e38d
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/frames.html
@@ -0,0 +1,30 @@
+<!DOCTYPE html 
+     PUBLIC "-//W3C//DTD XHTML 1.0 Frameset//EN"
+     "http://www.w3.org/TR/xhtml1/DTD/xhtml1-frameset.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<title></title>
+<script src="haddock-util.js" type="text/javascript"></script>
+<script type="text/javascript"><!--
+/*
+
+  The synopsis frame needs to be updated using javascript, so we hide
+  it by default and only show it if javascript is enabled.
+
+  TODO: provide some means to disable it.
+*/
+function load() {
+  var d = document.getElementById("inner-fs");
+  d.rows = "50%,50%";
+  postReframe();
+}
+--></script>
+</head>
+<frameset id="outer-fs" cols="25%,75%" onload="load()">
+  <frameset id="inner-fs" rows="100%,0%">
+    <frame src="index-frames.html" name="modules" />
+    <frame src="" name="synopsis" />
+  </frameset>
+  <frame src="index.html" name="main" />
+</frameset>
+</html>
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/haddock-util.js b/docs/haddock/tensorflow-proto-0.1.0.0/haddock-util.js
new file mode 100644
index 0000000..9a6fccf
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/haddock-util.js
@@ -0,0 +1,344 @@
+// Haddock JavaScript utilities
+
+var rspace = /\s\s+/g,
+	  rtrim = /^\s+|\s+$/g;
+
+function spaced(s) { return (" " + s + " ").replace(rspace, " "); }
+function trim(s)   { return s.replace(rtrim, ""); }
+
+function hasClass(elem, value) {
+  var className = spaced(elem.className || "");
+  return className.indexOf( " " + value + " " ) >= 0;
+}
+
+function addClass(elem, value) {
+  var className = spaced(elem.className || "");
+  if ( className.indexOf( " " + value + " " ) < 0 ) {
+    elem.className = trim(className + " " + value);
+  }
+}
+
+function removeClass(elem, value) {
+  var className = spaced(elem.className || "");
+  className = className.replace(" " + value + " ", " ");
+  elem.className = trim(className);
+}
+
+function toggleClass(elem, valueOn, valueOff, bool) {
+  if (bool == null) { bool = ! hasClass(elem, valueOn); }
+  if (bool) {
+    removeClass(elem, valueOff);
+    addClass(elem, valueOn);
+  }
+  else {
+    removeClass(elem, valueOn);
+    addClass(elem, valueOff);
+  }
+  return bool;
+}
+
+
+function makeClassToggle(valueOn, valueOff)
+{
+  return function(elem, bool) {
+    return toggleClass(elem, valueOn, valueOff, bool);
+  }
+}
+
+toggleShow = makeClassToggle("show", "hide");
+toggleCollapser = makeClassToggle("collapser", "expander");
+
+function toggleSection(id)
+{
+  var b = toggleShow(document.getElementById("section." + id));
+  toggleCollapser(document.getElementById("control." + id), b);
+  rememberCollapsed(id, b);
+  return b;
+}
+
+var collapsed = {};
+function rememberCollapsed(id, b)
+{
+  if(b)
+    delete collapsed[id]
+  else
+    collapsed[id] = null;
+
+  var sections = [];
+  for(var i in collapsed)
+  {
+    if(collapsed.hasOwnProperty(i))
+      sections.push(i);
+  }
+  // cookie specific to this page; don't use setCookie which sets path=/
+  document.cookie = "collapsed=" + escape(sections.join('+'));
+}
+
+function restoreCollapsed()
+{
+  var cookie = getCookie("collapsed");
+  if(!cookie)
+    return;
+
+  var ids = cookie.split('+');
+  for(var i in ids)
+  {
+    if(document.getElementById("section." + ids[i]))
+      toggleSection(ids[i]);
+  }
+}
+
+function setCookie(name, value) {
+  document.cookie = name + "=" + escape(value) + ";path=/;";
+}
+
+function clearCookie(name) {
+  document.cookie = name + "=;path=/;expires=Thu, 01-Jan-1970 00:00:01 GMT;";
+}
+
+function getCookie(name) {
+  var nameEQ = name + "=";
+  var ca = document.cookie.split(';');
+  for(var i=0;i < ca.length;i++) {
+    var c = ca[i];
+    while (c.charAt(0)==' ') c = c.substring(1,c.length);
+    if (c.indexOf(nameEQ) == 0) {
+      return unescape(c.substring(nameEQ.length,c.length));
+    }
+  }
+  return null;
+}
+
+
+
+var max_results = 75; // 50 is not enough to search for map in the base libraries
+var shown_range = null;
+var last_search = null;
+
+function quick_search()
+{
+    perform_search(false);
+}
+
+function full_search()
+{
+    perform_search(true);
+}
+
+
+function perform_search(full)
+{
+    var text = document.getElementById("searchbox").value.toLowerCase();
+    if (text == last_search && !full) return;
+    last_search = text;
+    
+    var table = document.getElementById("indexlist");
+    var status = document.getElementById("searchmsg");
+    var children = table.firstChild.childNodes;
+    
+    // first figure out the first node with the prefix
+    var first = bisect(-1);
+    var last = (first == -1 ? -1 : bisect(1));
+
+    if (first == -1)
+    {
+        table.className = "";
+        status.innerHTML = "No results found, displaying all";
+    }
+    else if (first == 0 && last == children.length - 1)
+    {
+        table.className = "";
+        status.innerHTML = "";
+    }
+    else if (last - first >= max_results && !full)
+    {
+        table.className = "";
+        status.innerHTML = "More than " + max_results + ", press Search to display";
+    }
+    else
+    {
+        // decide what you need to clear/show
+        if (shown_range)
+            setclass(shown_range[0], shown_range[1], "indexrow");
+        setclass(first, last, "indexshow");
+        shown_range = [first, last];
+        table.className = "indexsearch";
+        status.innerHTML = "";
+    }
+
+    
+    function setclass(first, last, status)
+    {
+        for (var i = first; i <= last; i++)
+        {
+            children[i].className = status;
+        }
+    }
+    
+    
+    // do a binary search, treating 0 as ...
+    // return either -1 (no 0's found) or location of most far match
+    function bisect(dir)
+    {
+        var first = 0, finish = children.length - 1;
+        var mid, success = false;
+
+        while (finish - first > 3)
+        {
+            mid = Math.floor((finish + first) / 2);
+
+            var i = checkitem(mid);
+            if (i == 0) i = dir;
+            if (i == -1)
+                finish = mid;
+            else
+                first = mid;
+        }
+        var a = (dir == 1 ? first : finish);
+        var b = (dir == 1 ? finish : first);
+        for (var i = b; i != a - dir; i -= dir)
+        {
+            if (checkitem(i) == 0) return i;
+        }
+        return -1;
+    }    
+    
+    
+    // from an index, decide what the result is
+    // 0 = match, -1 is lower, 1 is higher
+    function checkitem(i)
+    {
+        var s = getitem(i).toLowerCase().substr(0, text.length);
+        if (s == text) return 0;
+        else return (s > text ? -1 : 1);
+    }
+    
+    
+    // from an index, get its string
+    // this abstracts over alternates
+    function getitem(i)
+    {
+        for ( ; i >= 0; i--)
+        {
+            var s = children[i].firstChild.firstChild.data;
+            if (s.indexOf(' ') == -1)
+                return s;
+        }
+        return ""; // should never be reached
+    }
+}
+
+function setSynopsis(filename) {
+    if (parent.window.synopsis) {
+        if (parent.window.synopsis.location.replace) {
+            // In Firefox this avoids adding the change to the history.
+            parent.window.synopsis.location.replace(filename);
+        } else {
+            parent.window.synopsis.location = filename;
+        }
+    }
+}
+
+function addMenuItem(html) {
+  var menu = document.getElementById("page-menu");
+  if (menu) {
+    var btn = menu.firstChild.cloneNode(false);
+    btn.innerHTML = html;
+    menu.appendChild(btn);
+  }
+}
+
+function adjustForFrames() {
+  var bodyCls;
+  
+  if (parent.location.href == window.location.href) {
+    // not in frames, so add Frames button
+    addMenuItem("<a href='#' onclick='reframe();return true;'>Frames</a>");
+    bodyCls = "no-frame";
+  }
+  else {
+    bodyCls = "in-frame";
+  }
+  addClass(document.body, bodyCls);
+}
+
+function reframe() {
+  setCookie("haddock-reframe", document.URL);
+  window.location = "frames.html";
+}
+
+function postReframe() {
+  var s = getCookie("haddock-reframe");
+  if (s) {
+    parent.window.main.location = s;
+    clearCookie("haddock-reframe");
+  }
+}
+
+function styles() {
+  var i, a, es = document.getElementsByTagName("link"), rs = [];
+  for (i = 0; a = es[i]; i++) {
+    if(a.rel.indexOf("style") != -1 && a.title) {
+      rs.push(a);
+    }
+  }
+  return rs;
+}
+
+function addStyleMenu() {
+  var as = styles();
+  var i, a, btns = "";
+  for(i=0; a = as[i]; i++) {
+    btns += "<li><a href='#' onclick=\"setActiveStyleSheet('"
+      + a.title + "'); return false;\">"
+      + a.title + "</a></li>"
+  }
+  if (as.length > 1) {
+    var h = "<div id='style-menu-holder'>"
+      + "<a href='#' onclick='styleMenu(); return false;'>Style &#9662;</a>"
+      + "<ul id='style-menu' class='hide'>" + btns + "</ul>"
+      + "</div>";
+    addMenuItem(h);
+  }
+}
+
+function setActiveStyleSheet(title) {
+  var as = styles();
+  var i, a, found;
+  for(i=0; a = as[i]; i++) {
+    a.disabled = true;
+          // need to do this always, some browsers are edge triggered
+    if(a.title == title) {
+      found = a;
+    }
+  }
+  if (found) {
+    found.disabled = false;
+    setCookie("haddock-style", title);
+  }
+  else {
+    as[0].disabled = false;
+    clearCookie("haddock-style");
+  }
+  styleMenu(false);
+}
+
+function resetStyle() {
+  var s = getCookie("haddock-style");
+  if (s) setActiveStyleSheet(s);
+}
+
+
+function styleMenu(show) {
+  var m = document.getElementById('style-menu');
+  if (m) toggleShow(m, show);
+}
+
+
+function pageLoad() {
+  addStyleMenu();
+  adjustForFrames();
+  resetStyle();
+  restoreCollapsed();
+}
+
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/hslogo-16.png b/docs/haddock/tensorflow-proto-0.1.0.0/hslogo-16.png
new file mode 100644
index 0000000000000000000000000000000000000000..0ff8579fbd897417b0d6dad6e920f8882138a7c0
GIT binary patch
literal 1684
zcmV;F25b3=P)<h;3K|Lk000e1NJLTq000*N000mO1^@s6{V(Jg0008+X+uL$Nkc;*
zP;zf(X>4Tx0C)j~RL^S@K@|QrZmG~B2wH0nvUrdpNm;9CMbtL^5n^i$+aI<?kYqDS
z)^0ZI?k2URh~(nYD&9ryp$Gqf9(wR(FG_`4MJV2@C}N=(E%D86Otz)!z|Ma2-Z$@k
zZ+71R4RX>n^?(HA4aZWV5ov6ELTdbo0FI&wK{O>*+w4vx20?>!`FrQsdJlnHR>OPy
zcd~b_n$otK2Za4V;76L-DzNVtaSB-y0*E}{p()372;bw_^6ZZ}PI-92wGS&j#91PI
zKs7DSe@(bk%_Y-7gGe}(^>I=@oY#w#*Bu9GZf3^F5WP>3rn}7Ut74&?PWBFvy`A)a
zPP5)V!Xd&78LdA?xQ(9mjMYElVd13a#D+Z_7&Y|xU=_C-srWU*6kiZcC!$nw*)9$7
zn6CX+@=AhmkT}X@VSsa5NKe;HZuq)~1$`#h6R+ZTR#D-3j}vF!)ZOnz+5)dI4jl{{
z44Mr{P!L4~VVJN`K!!XTF*LGrKO?IK8<Tr7btG!LbYeuYL3=jbJ-1P$-8}v%B5{;M
zwFr{@LH;VQ$xr2Z`O93e*jD$Ht(%&<^58qg<(at}9@o>z<8w`3e3jI8lUGNUta*C8
zn(P`s>{pjD=7Kek#B;Fw@hxAK%$F&Q6vg9J^Xf~4by_hu-=A!MJ3Znq&n~srbFGPs
zH&&aMXZ>nO`|hf|ljc?VPhR!${AbO?W8x_>CU%PFA&Hm8F7cAsOREdwU~R_;ot1_u
z(ruCYB-LPGn!NQdT|ZlRy+(fw^-+`=%+gee_kY4FWHg<*4sZI8+sFJD<oAl_pC|$^
zY~aY5x@}W&?+~G7rEYVs0vEs0eekw!YomTR`~+A$s}`+NHJ>270UUORdLHO0nA4V)
z%{fwsET5CQ>B?eK%uw4yQc~9?*JVo<vzb+5>2}ze(;aRcp*ceL#HUJSllrgm5wQKR
zQu+C;QrUh^8rFfA`ftFz{YAidi-`aL010qNS#tmY4c7nw4c7reD4Tcy00T@(L_t(I
z5sj2vNEA^R$7gqDc6T=2^@fUA2(c`MltuL5<|<Z1YDGjBEr^N`3VJpQ$`C4&9(*Z6
z=|*B$y;-1^F6vXRAd7+(;)77Qps2euv)AeWmj>KW>RWz$&YbU@|M|{$E*8Tu-Ux!w
z1Y*Dr&Ubfr&v-nZaaB{3ilRumrjPmk{sZvQEWlW+{o~IH|8)=s6c#X9S5s5d%<Agu
zgM$153Ttk++ZPA~ZqgqN1|^pNV-|?9-zo48n9+Yiy$*su5czgp*VFj6!1h&DRZR>J
z4@)&QH5|xQY-)^L1n<EM!01(Q6kM8}ojndgRD~2KVIa@>0pTRu0Lx9`08YTjTwn^6
z0;b1+aQ@)n;Em$q;=7BBi)v0zj&o^g>0Whp^_^5IbxIUP8C@y9;R?*Ouu}rmfxbU=
zwtWVNke-m!=`7bYEhWpcI5#)9qp`8E0lr6IQ)ARL3Ui}Af@grj8aN1=r>Cb+prlzO
zNfJs*N_tUm2ZL<f%*@E?H`Ak_J`Q#Qri<;K{{H@Bg@uKWS5{WE+}vCx91geD*4DNm
zJ?D^~dGz7lWtbWH`o<0O#|{R6h9M*GMpjnV?Y_RgatOb~1le#ikPKTNIn%<;=>%5*
zPmL2??da$TR904gL(VDAQ<BE%k@5ZA>-Fv_Dk}Pdw*4T(%*f4MKLRg=4ekMjhe2mW
zMFsBwg%ftWT}0kxRaIk1k7qJ8*#cKB;Ft{i`zVIs-Nqge;!!Ld7#O&Qqu7e0sJmP)
z$MW*>L$vSB&dxp@iA3U9fo)-7!Czlr{|o7<LYh`Fd@X2-b0fIa*Wdy$({Zt@t7`&L
zhoaGF!l5{1P1Qm;<GaSj#=Oa7atf1JZ8n=F9*@uC?b(X48O9kPoH~h;!o2=WM?ZLl
z$Lj;4K~)fAdW@^N%nzT>Hv{1oyg3xsu%gn@(b1>$;SM-ZaQ`HV=V0s;lr%d8bd;xY
zGw<i+<$17ay36gdrqk(Lxc1|{y}dSeM5OIV(FjhBg5#OM<!Z*8(@<7crlVw+B-8R1
zl{3J6w9fYQ^vojauaRgZ=~SE&CK!R3pCe!v3>Nvm3=Iu=tyXIgtJnf@A(2S@M140N
ew{UA~tMxaJq;$xaSSi*30000<MNUMnLSTa0p)&^n

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/index-frames.html b/docs/haddock/tensorflow-proto-0.1.0.0/index-frames.html
new file mode 100644
index 0000000..a3a63f9
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/index-frames.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-list"><p class="caption">Modules</p><ul><li class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html" target="main">Proto.Tensorflow.Core.Framework.AttrValue</a></li><li class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html" target="main">Proto.Tensorflow.Core.Framework.Graph</a></li><li class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html" target="main">Proto.Tensorflow.Core.Framework.NodeDef</a></li><li class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html" target="main">Proto.Tensorflow.Core.Framework.OpDef</a></li><li class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html" target="main">Proto.Tensorflow.Core.Framework.ResourceHandle</a></li><li class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html" target="main">Proto.Tensorflow.Core.Framework.Tensor</a></li><li class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html" target="main">Proto.Tensorflow.Core.Framework.TensorShape</a></li><li class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html" target="main">Proto.Tensorflow.Core.Framework.Types</a></li><li class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html" target="main">Proto.Tensorflow.Core.Protobuf.Config</a></li></ul></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/index.html b/docs/haddock/tensorflow-proto-0.1.0.0/index.html
new file mode 100644
index 0000000..fb07a5b
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</p></div><div id="content"><div id="description"><h1>tensorflow-proto-0.1.0.0: TensorFlow protocol buffers.</h1><div class="doc"><p>Please see README.md</p></div></div><div id="module-list"><p class="caption">Modules</p><ul><li><span id="control.n.1" class="module collapser" onclick="toggleSection('n.1')">Proto</span><ul id="section.n.1" class="show"><li><span id="control.n.1.1" class="module collapser" onclick="toggleSection('n.1.1')">Tensorflow</span><ul id="section.n.1.1" class="show"><li><span id="control.n.1.1.1" class="module collapser" onclick="toggleSection('n.1.1.1')">Core</span><ul id="section.n.1.1.1" class="show"><li><span id="control.n.1.1.1.1" class="module collapser" onclick="toggleSection('n.1.1.1.1')">Framework</span><ul id="section.n.1.1.1.1" class="show"><li><span class="module"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html">Proto.Tensorflow.Core.Framework.AttrValue</a></span></li><li><span class="module"><a href="Proto-Tensorflow-Core-Framework-Graph.html">Proto.Tensorflow.Core.Framework.Graph</a></span></li><li><span class="module"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html">Proto.Tensorflow.Core.Framework.NodeDef</a></span></li><li><span class="module"><a href="Proto-Tensorflow-Core-Framework-OpDef.html">Proto.Tensorflow.Core.Framework.OpDef</a></span></li><li><span class="module"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html">Proto.Tensorflow.Core.Framework.ResourceHandle</a></span></li><li><span class="module"><a href="Proto-Tensorflow-Core-Framework-Tensor.html">Proto.Tensorflow.Core.Framework.Tensor</a></span></li><li><span class="module"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html">Proto.Tensorflow.Core.Framework.TensorShape</a></span></li><li><span class="module"><a href="Proto-Tensorflow-Core-Framework-Types.html">Proto.Tensorflow.Core.Framework.Types</a></span></li></ul></li><li><span id="control.n.1.1.1.2" class="module collapser" onclick="toggleSection('n.1.1.1.2')">Protobuf</span><ul id="section.n.1.1.1.2" class="show"><li><span class="module"><a href="Proto-Tensorflow-Core-Protobuf-Config.html">Proto.Tensorflow.Core.Protobuf.Config</a></span></li></ul></li></ul></li></ul></li></ul></li></ul></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-AttrValue.html b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-AttrValue.html
new file mode 100644
index 0000000..a05d215
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-AttrValue.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Framework.AttrValue</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">Proto.Tensorflow.Core.Framework.AttrValue</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue" target="main">AttrValue</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:AttrValue-39-ListValue" target="main">AttrValue'ListValue</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList" target="main">NameAttrList</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-AttrValue.html#t:NameAttrList-39-AttrEntry" target="main">NameAttrList'AttrEntry</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:attr" target="main">attr</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:b" target="main">b</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:f" target="main">f</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:func" target="main">func</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:i" target="main">i</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:key" target="main">key</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:list" target="main">list</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-b" target="main">maybe'b</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-f" target="main">maybe'f</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-func" target="main">maybe'func</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-i" target="main">maybe'i</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-list" target="main">maybe'list</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-placeholder" target="main">maybe'placeholder</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-s" target="main">maybe's</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-shape" target="main">maybe'shape</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-tensor" target="main">maybe'tensor</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-type-39-" target="main">maybe'type'</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:maybe-39-value" target="main">maybe'value</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:name" target="main">name</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:placeholder" target="main">placeholder</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:s" target="main">s</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:shape" target="main">shape</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:tensor" target="main">tensor</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:type-39-" target="main">type'</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-AttrValue.html#v:value" target="main">value</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-Graph.html b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-Graph.html
new file mode 100644
index 0000000..14f4318
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-Graph.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Framework.Graph</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">Proto.Tensorflow.Core.Framework.Graph</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-Graph.html#t:GraphDef" target="main">GraphDef</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:library" target="main">library</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:maybe-39-library" target="main">maybe'library</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:maybe-39-versions" target="main">maybe'versions</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:node" target="main">node</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:version" target="main">version</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Graph.html#v:versions" target="main">versions</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-NodeDef.html b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-NodeDef.html
new file mode 100644
index 0000000..7d65c77
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-NodeDef.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Framework.NodeDef</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">Proto.Tensorflow.Core.Framework.NodeDef</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef" target="main">NodeDef</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-NodeDef.html#t:NodeDef-39-AttrEntry" target="main">NodeDef'AttrEntry</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:attr" target="main">attr</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:device" target="main">device</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:input" target="main">input</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:key" target="main">key</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:maybe-39-value" target="main">maybe'value</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:name" target="main">name</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:op" target="main">op</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-NodeDef.html#v:value" target="main">value</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-OpDef.html b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-OpDef.html
new file mode 100644
index 0000000..2cd795c
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-OpDef.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Framework.OpDef</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">Proto.Tensorflow.Core.Framework.OpDef</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef" target="main">OpDef</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-ArgDef" target="main">OpDef'ArgDef</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDef-39-AttrDef" target="main">OpDef'AttrDef</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpDeprecation" target="main">OpDeprecation</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-OpDef.html#t:OpList" target="main">OpList</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:allowedValues" target="main">allowedValues</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:allowsUninitializedInput" target="main">allowsUninitializedInput</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:attr" target="main">attr</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:defaultValue" target="main">defaultValue</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:deprecation" target="main">deprecation</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:description" target="main">description</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:explanation" target="main">explanation</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:hasMinimum" target="main">hasMinimum</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:inputArg" target="main">inputArg</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:isAggregate" target="main">isAggregate</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:isCommutative" target="main">isCommutative</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:isRef" target="main">isRef</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:isStateful" target="main">isStateful</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:maybe-39-allowedValues" target="main">maybe'allowedValues</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:maybe-39-defaultValue" target="main">maybe'defaultValue</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:maybe-39-deprecation" target="main">maybe'deprecation</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:minimum" target="main">minimum</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:name" target="main">name</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:numberAttr" target="main">numberAttr</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:op" target="main">op</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:outputArg" target="main">outputArg</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:summary" target="main">summary</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:type-39-" target="main">type'</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:typeAttr" target="main">typeAttr</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:typeListAttr" target="main">typeListAttr</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-OpDef.html#v:version" target="main">version</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-ResourceHandle.html b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-ResourceHandle.html
new file mode 100644
index 0000000..521fe5c
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-ResourceHandle.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Framework.ResourceHandle</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">Proto.Tensorflow.Core.Framework.ResourceHandle</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#t:ResourceHandle" target="main">ResourceHandle</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:container" target="main">container</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:device" target="main">device</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:hashCode" target="main">hashCode</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:maybeTypeName" target="main">maybeTypeName</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-ResourceHandle.html#v:name" target="main">name</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-Tensor.html b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-Tensor.html
new file mode 100644
index 0000000..275f22f
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-Tensor.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Framework.Tensor</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">Proto.Tensorflow.Core.Framework.Tensor</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-Tensor.html#t:TensorProto" target="main">TensorProto</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:boolVal" target="main">boolVal</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:dcomplexVal" target="main">dcomplexVal</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:doubleVal" target="main">doubleVal</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:dtype" target="main">dtype</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:floatVal" target="main">floatVal</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:halfVal" target="main">halfVal</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:int64Val" target="main">int64Val</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:intVal" target="main">intVal</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:maybe-39-tensorShape" target="main">maybe'tensorShape</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:resourceHandleVal" target="main">resourceHandleVal</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:scomplexVal" target="main">scomplexVal</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:stringVal" target="main">stringVal</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:tensorContent" target="main">tensorContent</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:tensorShape" target="main">tensorShape</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-Tensor.html#v:versionNumber" target="main">versionNumber</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-TensorShape.html b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-TensorShape.html
new file mode 100644
index 0000000..29f6d63
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-TensorShape.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Framework.TensorShape</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">Proto.Tensorflow.Core.Framework.TensorShape</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto" target="main">TensorShapeProto</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-TensorShape.html#t:TensorShapeProto-39-Dim" target="main">TensorShapeProto'Dim</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:dim" target="main">dim</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:name" target="main">name</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:size" target="main">size</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Framework-TensorShape.html#v:unknownRank" target="main">unknownRank</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-Types.html b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-Types.html
new file mode 100644
index 0000000..ab4e0a2
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Framework-Types.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Framework.Types</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">Proto.Tensorflow.Core.Framework.Types</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Framework-Types.html#t:DataType" target="main">DataType</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Protobuf-Config.html b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Protobuf-Config.html
new file mode 100644
index 0000000..6854d28
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/mini_Proto-Tensorflow-Core-Protobuf-Config.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Proto.Tensorflow.Core.Protobuf.Config</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">Proto.Tensorflow.Core.Protobuf.Config</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto" target="main">ConfigProto</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ConfigProto-39-DeviceCountEntry" target="main">ConfigProto'DeviceCountEntry</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:DebugTensorWatch" target="main">DebugTensorWatch</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GPUOptions" target="main">GPUOptions</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:GraphOptions" target="main">GraphOptions</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions" target="main">OptimizerOptions</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:OptimizerOptions-39-Level" target="main">OptimizerOptions'Level</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunMetadata" target="main">RunMetadata</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions" target="main">RunOptions</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:RunOptions-39-TraceLevel" target="main">RunOptions'TraceLevel</a></p></div><div class="top"><p class="src"><span class="keyword">data</span> <a href="Proto-Tensorflow-Core-Protobuf-Config.html#t:ThreadPoolOptionProto" target="main">ThreadPoolOptionProto</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:allocatorType" target="main">allocatorType</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:allowGrowth" target="main">allowGrowth</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:allowSoftPlacement" target="main">allowSoftPlacement</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:buildCostModel" target="main">buildCostModel</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:buildCostModelAfter" target="main">buildCostModelAfter</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:costGraph" target="main">costGraph</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:debugOps" target="main">debugOps</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:debugTensorWatchOpts" target="main">debugTensorWatchOpts</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:debugUrls" target="main">debugUrls</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:deferredDeletionBytes" target="main">deferredDeletionBytes</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:deviceCount" target="main">deviceCount</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:deviceFilters" target="main">deviceFilters</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:doCommonSubexpressionElimination" target="main">doCommonSubexpressionElimination</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:doConstantFolding" target="main">doConstantFolding</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:doFunctionInlining" target="main">doFunctionInlining</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:enableBfloat16Sendrecv" target="main">enableBfloat16Sendrecv</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:enableRecvScheduling" target="main">enableRecvScheduling</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:gpuOptions" target="main">gpuOptions</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:graphOptions" target="main">graphOptions</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:inferShapes" target="main">inferShapes</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:interOpParallelismThreads" target="main">interOpParallelismThreads</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:interOpThreadPool" target="main">interOpThreadPool</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:intraOpParallelismThreads" target="main">intraOpParallelismThreads</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:key" target="main">key</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:logDevicePlacement" target="main">logDevicePlacement</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-costGraph" target="main">maybe'costGraph</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-gpuOptions" target="main">maybe'gpuOptions</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-graphOptions" target="main">maybe'graphOptions</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-optimizerOptions" target="main">maybe'optimizerOptions</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:maybe-39-stepStats" target="main">maybe'stepStats</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:nodeName" target="main">nodeName</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:numThreads" target="main">numThreads</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:operationTimeoutInMs" target="main">operationTimeoutInMs</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:optLevel" target="main">optLevel</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:optimizerOptions" target="main">optimizerOptions</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:outputPartitionGraphs" target="main">outputPartitionGraphs</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:outputSlot" target="main">outputSlot</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:partitionGraphs" target="main">partitionGraphs</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:perProcessGpuMemoryFraction" target="main">perProcessGpuMemoryFraction</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:placePrunedGraph" target="main">placePrunedGraph</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:placementPeriod" target="main">placementPeriod</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:sessionInterOpThreadPool" target="main">sessionInterOpThreadPool</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:stepStats" target="main">stepStats</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:timelineStep" target="main">timelineStep</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:timeoutInMs" target="main">timeoutInMs</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:traceLevel" target="main">traceLevel</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:usePerSessionThreads" target="main">usePerSessionThreads</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:value" target="main">value</a></p></div><div class="top"><p class="src"><a href="Proto-Tensorflow-Core-Protobuf-Config.html#v:visibleDeviceList" target="main">visibleDeviceList</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/minus.gif b/docs/haddock/tensorflow-proto-0.1.0.0/minus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..1deac2fe1a42e35b994f1b855488f392c50f6a89
GIT binary patch
literal 56
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS53HJ0a;6C;`j@u!^I(6E|
Jh4LZ{)&P#)4$%Mr

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/ocean.css b/docs/haddock/tensorflow-proto-0.1.0.0/ocean.css
new file mode 100644
index 0000000..1110b40
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/ocean.css
@@ -0,0 +1,600 @@
+/* @group Fundamentals */
+
+* { margin: 0; padding: 0 }
+
+/* Is this portable? */
+html {
+  background-color: white;
+  width: 100%;
+  height: 100%;
+}
+
+body {
+  background: white;
+  color: black;
+  text-align: left;
+  min-height: 100%;
+  position: relative;
+}
+
+p {
+  margin: 0.8em 0;
+}
+
+ul, ol {
+  margin: 0.8em 0 0.8em 2em;
+}
+
+dl {
+  margin: 0.8em 0;
+}
+
+dt {
+  font-weight: bold;
+}
+dd {
+  margin-left: 2em;
+}
+
+a { text-decoration: none; }
+a[href]:link { color: rgb(196,69,29); }
+a[href]:visited { color: rgb(171,105,84); }
+a[href]:hover { text-decoration:underline; }
+
+/* @end */
+
+/* @group Fonts & Sizes */
+
+/* Basic technique & IE workarounds from YUI 3
+   For reasons, see:
+      http://yui.yahooapis.com/3.1.1/build/cssfonts/fonts.css
+ */
+
+body {
+	font:13px/1.4 sans-serif;
+	*font-size:small; /* for IE */
+	*font:x-small; /* for IE in quirks mode */
+}
+
+h1 { font-size: 146.5%; /* 19pt */ }
+h2 { font-size: 131%;   /* 17pt */ }
+h3 { font-size: 116%;   /* 15pt */ }
+h4 { font-size: 100%;   /* 13pt */ }
+h5 { font-size: 100%;   /* 13pt */ }
+
+select, input, button, textarea {
+	font:99% sans-serif;
+}
+
+table {
+	font-size:inherit;
+	font:100%;
+}
+
+pre, code, kbd, samp, tt, .src {
+	font-family:monospace;
+	*font-size:108%;
+	line-height: 124%;
+}
+
+.links, .link {
+  font-size: 85%; /* 11pt */
+}
+
+#module-header .caption {
+  font-size: 182%; /* 24pt */
+}
+
+.info  {
+  font-size: 85%; /* 11pt */
+}
+
+#table-of-contents, #synopsis  {
+  /* font-size: 85%; /* 11pt */
+}
+
+
+/* @end */
+
+/* @group Common */
+
+.caption, h1, h2, h3, h4, h5, h6 {
+  font-weight: bold;
+  color: rgb(78,98,114);
+  margin: 0.8em 0 0.4em;
+}
+
+* + h1, * + h2, * + h3, * + h4, * + h5, * + h6 {
+  margin-top: 2em;
+}
+
+h1 + h2, h2 + h3, h3 + h4, h4 + h5, h5 + h6 {
+  margin-top: inherit;
+}
+
+ul.links {
+  list-style: none;
+  text-align: left;
+  float: right;
+  display: inline-table;
+  margin: 0 0 0 1em;
+}
+
+ul.links li {
+  display: inline;
+  border-left: 1px solid #d5d5d5;
+  white-space: nowrap;
+  padding: 0;
+}
+
+ul.links li a {
+  padding: 0.2em 0.5em;
+}
+
+.hide { display: none; }
+.show { display: inherit; }
+.clear { clear: both; }
+
+.collapser {
+  background-image: url(minus.gif);
+  background-repeat: no-repeat;
+}
+.expander {
+  background-image: url(plus.gif);
+  background-repeat: no-repeat;
+}
+p.caption.collapser,
+p.caption.expander {
+  background-position: 0 0.4em;
+}
+.collapser, .expander {
+  padding-left: 14px;
+  margin-left: -14px;
+  cursor: pointer;
+}
+
+pre {
+  padding: 0.25em;
+  margin: 0.8em 0;
+  background: rgb(229,237,244);
+  overflow: auto;
+  border-bottom: 0.25em solid white;
+  /* white border adds some space below the box to compensate
+     for visual extra space that paragraphs have between baseline
+     and the bounding box */
+}
+
+.src {
+  background: #f0f0f0;
+  padding: 0.2em 0.5em;
+}
+
+.keyword { font-weight: normal; }
+.def { font-weight: bold; }
+
+
+/* @end */
+
+/* @group Page Structure */
+
+#content {
+  margin: 0 auto;
+  padding: 0 2em 6em;
+}
+
+#package-header {
+  background: rgb(41,56,69);
+  border-top: 5px solid rgb(78,98,114);
+  color: #ddd;
+  padding: 0.2em;
+  position: relative;
+  text-align: left;
+}
+
+#package-header .caption {
+  background: url(hslogo-16.png) no-repeat 0em;
+  color: white;
+  margin: 0 2em;
+  font-weight: normal;
+  font-style: normal;
+  padding-left: 2em;
+}
+
+#package-header a:link, #package-header a:visited { color: white; }
+#package-header a:hover { background: rgb(78,98,114); }
+
+#module-header .caption {
+  color: rgb(78,98,114);
+  font-weight: bold;
+  border-bottom: 1px solid #ddd;
+}
+
+table.info {
+  float: right;
+  padding: 0.5em 1em;
+  border: 1px solid #ddd;
+  color: rgb(78,98,114);
+  background-color: #fff;
+  max-width: 40%;
+  border-spacing: 0;
+  position: relative;
+  top: -0.5em;
+  margin: 0 0 0 2em;
+}
+
+.info th {
+	padding: 0 1em 0 0;
+}
+
+div#style-menu-holder {
+  position: relative;
+  z-index: 2;
+  display: inline;
+}
+
+#style-menu {
+  position: absolute;
+  z-index: 1;
+  overflow: visible;
+  background: #374c5e;
+  margin: 0;
+  text-align: center;
+  right: 0;
+  padding: 0;
+  top: 1.25em;
+}
+
+#style-menu li {
+	display: list-item;
+	border-style: none;
+	margin: 0;
+	padding: 0;
+	color: #000;
+	list-style-type: none;
+}
+
+#style-menu li + li {
+	border-top: 1px solid #919191;
+}
+
+#style-menu a {
+  width: 6em;
+  padding: 3px;
+  display: block;
+}
+
+#footer {
+  background: #ddd;
+  border-top: 1px solid #aaa;
+  padding: 0.5em 0;
+  color: #666;
+  text-align: center;
+  position: absolute;
+  bottom: 0;
+  width: 100%;
+  height: 3em;
+}
+
+/* @end */
+
+/* @group Front Matter */
+
+#table-of-contents {
+  float: right;
+  clear: right;
+  background: #faf9dc;
+  border: 1px solid #d8d7ad;
+  padding: 0.5em 1em;
+  max-width: 20em;
+  margin: 0.5em 0 1em 1em;
+}
+
+#table-of-contents .caption {
+  text-align: center;
+  margin: 0;
+}
+
+#table-of-contents ul {
+  list-style: none;
+  margin: 0;
+}
+
+#table-of-contents ul ul {
+  margin-left: 2em;
+}
+
+#description .caption {
+  display: none;
+}
+
+#synopsis {
+  display: none;
+}
+
+.no-frame #synopsis {
+  display: block;
+  position: fixed;
+  right: 0;
+  height: 80%;
+  top: 10%;
+  padding: 0;
+  max-width: 75%;
+}
+
+#synopsis .caption {
+  float: left;
+  width: 29px;
+  color: rgba(255,255,255,0);
+  height: 110px;
+  margin: 0;
+  font-size: 1px;
+  padding: 0;
+}
+
+#synopsis p.caption.collapser {
+  background: url(synopsis.png) no-repeat -64px -8px;
+}
+
+#synopsis p.caption.expander {
+  background: url(synopsis.png) no-repeat 0px -8px;
+}
+
+#synopsis ul {
+  height: 100%;
+  overflow: auto;
+  padding: 0.5em;
+  margin: 0;
+}
+
+#synopsis ul ul {
+  overflow: hidden;
+}
+
+#synopsis ul,
+#synopsis ul li.src {
+  background-color: #faf9dc;
+  white-space: nowrap;
+  list-style: none;
+  margin-left: 0;
+}
+
+/* @end */
+
+/* @group Main Content */
+
+#interface div.top { margin: 2em 0; }
+#interface h1 + div.top,
+#interface h2 + div.top,
+#interface h3 + div.top,
+#interface h4 + div.top,
+#interface h5 + div.top {
+ 	margin-top: 1em;
+}
+#interface p.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface td.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface span.fixity {
+  color: #919191;
+  border-left: 1px solid #919191;
+  padding: 0.2em 0.5em 0.2em 0.5em;
+  margin: 0 -1em 0 1em;
+}
+
+#interface span.rightedge {
+  border-left: 1px solid #919191;
+  padding: 0.2em 0 0.2em 0;
+  margin: 0 0 0 1em;
+}
+
+#interface table { border-spacing: 2px; }
+#interface td {
+  vertical-align: top;
+  padding-left: 0.5em;
+}
+#interface td.src {
+  white-space: nowrap;
+}
+#interface td.doc p {
+  margin: 0;
+}
+#interface td.doc p + p {
+  margin-top: 0.8em;
+}
+
+.clearfix:after {
+  clear: both;
+  content: " ";
+  display: block;
+  height: 0;
+  visibility: hidden;
+}
+
+.subs dl {
+  margin: 0;
+}
+
+.subs dt {
+  float: left;
+  clear: left;
+  display: block;
+  margin: 1px 0;
+}
+
+.subs dd {
+  float: right;
+  width: 90%;
+  display: block;
+  padding-left: 0.5em;
+  margin-bottom: 0.5em;
+}
+
+.subs dd.empty {
+  display: none;
+}
+
+.subs dd p {
+  margin: 0;
+}
+
+/* Render short-style data instances */
+.inst ul {
+  height: 100%;
+  padding: 0.5em;
+  margin: 0;
+}
+
+.inst, .inst li {
+  list-style: none;
+  margin-left: 1em;
+}
+
+/* Workaround for bug in Firefox (issue #384) */
+.inst-left {
+  float: left;
+}
+
+.top p.src {
+  border-top: 1px solid #ccc;
+}
+
+.subs, .doc {
+  /* use this selector for one level of indent */
+  padding-left: 2em;
+}
+
+.warning {
+  color: red;
+}
+
+.arguments {
+  margin-top: -0.4em;
+}
+.arguments .caption {
+  display: none;
+}
+
+.fields { padding-left: 1em; }
+
+.fields .caption { display: none; }
+
+.fields p { margin: 0 0; }
+
+/* this seems bulky to me
+.methods, .constructors {
+  background: #f8f8f8;
+  border: 1px solid #eee;
+}
+*/
+
+/* @end */
+
+/* @group Auxillary Pages */
+
+
+.extension-list {
+    list-style-type: none;
+    margin-left: 0;
+}
+
+#mini {
+  margin: 0 auto;
+  padding: 0 1em 1em;
+}
+
+#mini > * {
+  font-size: 93%; /* 12pt */
+}
+
+#mini #module-list .caption,
+#mini #module-header .caption {
+  font-size: 125%; /* 15pt */
+}
+
+#mini #interface h1,
+#mini #interface h2,
+#mini #interface h3,
+#mini #interface h4 {
+  font-size: 109%; /* 13pt */
+  margin: 1em 0 0;
+}
+
+#mini #interface .top,
+#mini #interface .src {
+  margin: 0;
+}
+
+#mini #module-list ul {
+  list-style: none;
+  margin: 0;
+}
+
+#alphabet ul {
+	list-style: none;
+	padding: 0;
+	margin: 0.5em 0 0;
+	text-align: center;
+}
+
+#alphabet li {
+	display: inline;
+	margin: 0 0.25em;
+}
+
+#alphabet a {
+	font-weight: bold;
+}
+
+#index .caption,
+#module-list .caption { font-size: 131%; /* 17pt */ }
+
+#index table {
+  margin-left: 2em;
+}
+
+#index .src {
+  font-weight: bold;
+}
+#index .alt {
+  font-size: 77%; /* 10pt */
+  font-style: italic;
+  padding-left: 2em;
+}
+
+#index td + td {
+  padding-left: 1em;
+}
+
+#module-list ul {
+  list-style: none;
+  margin: 0 0 0 2em;
+}
+
+#module-list li {
+  clear: right;
+}
+
+#module-list span.collapser,
+#module-list span.expander {
+  background-position: 0 0.3em;
+}
+
+#module-list .package {
+  float: right;
+}
+
+/* @end */
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/plus.gif b/docs/haddock/tensorflow-proto-0.1.0.0/plus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..2d15c14173d23f664b955cd24f51c82f5f09d91d
GIT binary patch
literal 59
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS5iS+a@;NG@rPSdH*>gbBX
M^XE!9f*2UA0nx1yDgXcg

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-AllocationDescription.html b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-AllocationDescription.html
new file mode 100644
index 0000000..403c9e4
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-AllocationDescription.html
@@ -0,0 +1,233 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Proto/Tensorflow/Core/Framework/AllocationDescription.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>{- This file was auto-generated from tensorflow/core/framework/allocation_description.proto by the proto-lens-protoc program. -}</span>
+<a name="line-2"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables, DataKinds, TypeFamilies,
+<a name="line-3"></a>  MultiParamTypeClasses, FlexibleContexts, FlexibleInstances,
+<a name="line-4"></a>  PatternSynonyms #-}</span>
+<a name="line-5"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-unused-imports#-}</span>
+<a name="line-6"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Prelude</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-11"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-12"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-13"></a>       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-14"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-15"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-16"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-17"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-18"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-19"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-20"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-24"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-25"></a>
+<a name="line-26"></a><a name="AllocationDescription"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>AllocationDescription</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AllocationDescription</span><span class='hs-layout'>{</span><span class='hs-sel'>_AllocationDescription'requestedBytes</span>
+<a name="line-27"></a>                                                   <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-28"></a>                                                   <span class='hs-sel'>_AllocationDescription'allocatedBytes</span> <span class='hs-keyglyph'>::</span>
+<a name="line-29"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-30"></a>                                                   <span class='hs-sel'>_AllocationDescription'allocatorName</span> <span class='hs-keyglyph'>::</span>
+<a name="line-31"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-32"></a>                                                   <span class='hs-sel'>_AllocationDescription'allocationId</span> <span class='hs-keyglyph'>::</span>
+<a name="line-33"></a>                                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-34"></a>                                                   <span class='hs-sel'>_AllocationDescription'hasSingleReference</span> <span class='hs-keyglyph'>::</span>
+<a name="line-35"></a>                                                   <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-36"></a>                                                   <span class='hs-sel'>_AllocationDescription'ptr</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word64</span><span class='hs-layout'>}</span>
+<a name="line-37"></a>                           <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-38"></a>
+<a name="line-39"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-40"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"requestedBytes"</span> <span class='hs-conid'>AllocationDescription</span> <span class='hs-keyglyph'>=</span>
+<a name="line-41"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-42"></a>
+<a name="line-43"></a><a name="instance%20Data.ProtoLens.HasField%20%22requestedBytes%22%20AllocationDescription%20AllocationDescription"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"requestedBytes"</span>
+<a name="line-44"></a>         <span class='hs-conid'>AllocationDescription</span> <span class='hs-conid'>AllocationDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-45"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-46"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AllocationDescription'requestedBytes</span>
+<a name="line-47"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AllocationDescription'requestedBytes</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-48"></a>
+<a name="line-49"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-50"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocatedBytes"</span> <span class='hs-conid'>AllocationDescription</span> <span class='hs-keyglyph'>=</span>
+<a name="line-51"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-52"></a>
+<a name="line-53"></a><a name="instance%20Data.ProtoLens.HasField%20%22allocatedBytes%22%20AllocationDescription%20AllocationDescription"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allocatedBytes"</span>
+<a name="line-54"></a>         <span class='hs-conid'>AllocationDescription</span> <span class='hs-conid'>AllocationDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-55"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-56"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AllocationDescription'allocatedBytes</span>
+<a name="line-57"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AllocationDescription'allocatedBytes</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-58"></a>
+<a name="line-59"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-60"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocatorName"</span> <span class='hs-conid'>AllocationDescription</span> <span class='hs-keyglyph'>=</span>
+<a name="line-61"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-62"></a>
+<a name="line-63"></a><a name="instance%20Data.ProtoLens.HasField%20%22allocatorName%22%20AllocationDescription%20AllocationDescription"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allocatorName"</span>
+<a name="line-64"></a>         <span class='hs-conid'>AllocationDescription</span> <span class='hs-conid'>AllocationDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-65"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-66"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AllocationDescription'allocatorName</span>
+<a name="line-67"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AllocationDescription'allocatorName</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-68"></a>
+<a name="line-69"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-70"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocationId"</span> <span class='hs-conid'>AllocationDescription</span> <span class='hs-keyglyph'>=</span>
+<a name="line-71"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-72"></a>
+<a name="line-73"></a><a name="instance%20Data.ProtoLens.HasField%20%22allocationId%22%20AllocationDescription%20AllocationDescription"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allocationId"</span>
+<a name="line-74"></a>         <span class='hs-conid'>AllocationDescription</span> <span class='hs-conid'>AllocationDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-75"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-76"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AllocationDescription'allocationId</span>
+<a name="line-77"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AllocationDescription'allocationId</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-78"></a>
+<a name="line-79"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-80"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"hasSingleReference"</span> <span class='hs-conid'>AllocationDescription</span> <span class='hs-keyglyph'>=</span>
+<a name="line-81"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-82"></a>
+<a name="line-83"></a><a name="instance%20Data.ProtoLens.HasField%20%22hasSingleReference%22%20AllocationDescription%20AllocationDescription"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"hasSingleReference"</span>
+<a name="line-84"></a>         <span class='hs-conid'>AllocationDescription</span> <span class='hs-conid'>AllocationDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-85"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-86"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span>
+<a name="line-87"></a>              <span class='hs-sel'>_AllocationDescription'hasSingleReference</span>
+<a name="line-88"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AllocationDescription'hasSingleReference</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-89"></a>
+<a name="line-90"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"ptr"</span> <span class='hs-conid'>AllocationDescription</span> <span class='hs-keyglyph'>=</span>
+<a name="line-91"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word64</span>
+<a name="line-92"></a>
+<a name="line-93"></a><a name="instance%20Data.ProtoLens.HasField%20%22ptr%22%20AllocationDescription%20AllocationDescription"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"ptr"</span> <span class='hs-conid'>AllocationDescription</span>
+<a name="line-94"></a>         <span class='hs-conid'>AllocationDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-95"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-96"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AllocationDescription'ptr</span>
+<a name="line-97"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AllocationDescription'ptr</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-98"></a>
+<a name="line-99"></a><a name="instance%20Data.Default.Class.Default%20AllocationDescription"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>AllocationDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-100"></a>        <span class='hs-varid'>def</span>
+<a name="line-101"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AllocationDescription</span><span class='hs-layout'>{</span><span class='hs-sel'>_AllocationDescription'requestedBytes</span> <span class='hs-keyglyph'>=</span>
+<a name="line-102"></a>                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-103"></a>                                  <span class='hs-sel'>_AllocationDescription'allocatedBytes</span> <span class='hs-keyglyph'>=</span>
+<a name="line-104"></a>                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-105"></a>                                  <span class='hs-sel'>_AllocationDescription'allocatorName</span> <span class='hs-keyglyph'>=</span>
+<a name="line-106"></a>                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-107"></a>                                  <span class='hs-sel'>_AllocationDescription'allocationId</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-108"></a>                                  <span class='hs-sel'>_AllocationDescription'hasSingleReference</span> <span class='hs-keyglyph'>=</span>
+<a name="line-109"></a>                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-110"></a>                                  <span class='hs-sel'>_AllocationDescription'ptr</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-111"></a>
+<a name="line-112"></a><a name="instance%20Data.ProtoLens.Message%20AllocationDescription"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>AllocationDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-113"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-114"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>requestedBytes__field_descriptor</span>
+<a name="line-115"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"requested_bytes"</span>
+<a name="line-116"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-117"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-118"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>requestedBytes</span><span class='hs-layout'>)</span>
+<a name="line-119"></a>                <span class='hs-varid'>allocatedBytes__field_descriptor</span>
+<a name="line-120"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"allocated_bytes"</span>
+<a name="line-121"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-122"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-123"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>allocatedBytes</span><span class='hs-layout'>)</span>
+<a name="line-124"></a>                <span class='hs-varid'>allocatorName__field_descriptor</span>
+<a name="line-125"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"allocator_name"</span>
+<a name="line-126"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-127"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-128"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>allocatorName</span><span class='hs-layout'>)</span>
+<a name="line-129"></a>                <span class='hs-varid'>allocationId__field_descriptor</span>
+<a name="line-130"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"allocation_id"</span>
+<a name="line-131"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-132"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-133"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>allocationId</span><span class='hs-layout'>)</span>
+<a name="line-134"></a>                <span class='hs-varid'>hasSingleReference__field_descriptor</span>
+<a name="line-135"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"has_single_reference"</span>
+<a name="line-136"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-137"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-138"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-139"></a>                         <span class='hs-varid'>hasSingleReference</span><span class='hs-layout'>)</span>
+<a name="line-140"></a>                <span class='hs-varid'>ptr__field_descriptor</span>
+<a name="line-141"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"ptr"</span>
+<a name="line-142"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>UInt64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-143"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word64</span><span class='hs-layout'>)</span>
+<a name="line-144"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>ptr</span><span class='hs-layout'>)</span>
+<a name="line-145"></a>              <span class='hs-keyword'>in</span>
+<a name="line-146"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-147"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-148"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>requestedBytes__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-149"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>allocatedBytes__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-150"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>allocatorName__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-151"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>allocationId__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-152"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>5</span><span class='hs-layout'>,</span> <span class='hs-varid'>hasSingleReference__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-153"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>6</span><span class='hs-layout'>,</span> <span class='hs-varid'>ptr__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-154"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-155"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"requested_bytes"</span><span class='hs-layout'>,</span> <span class='hs-varid'>requestedBytes__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-156"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"allocated_bytes"</span><span class='hs-layout'>,</span> <span class='hs-varid'>allocatedBytes__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-157"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"allocator_name"</span><span class='hs-layout'>,</span> <span class='hs-varid'>allocatorName__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-158"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"allocation_id"</span><span class='hs-layout'>,</span> <span class='hs-varid'>allocationId__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-159"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"has_single_reference"</span><span class='hs-layout'>,</span> <span class='hs-varid'>hasSingleReference__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-160"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"ptr"</span><span class='hs-layout'>,</span> <span class='hs-varid'>ptr__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-161"></a>
+<a name="line-162"></a><a name="allocatedBytes"></a><span class='hs-definition'>allocatedBytes</span> <span class='hs-keyglyph'>::</span>
+<a name="line-163"></a>               <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-164"></a>                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allocatedBytes"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-165"></a>                 <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-166"></a>                   <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocatedBytes"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-167"></a>                   <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocatedBytes"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-168"></a><span class='hs-definition'>allocatedBytes</span>
+<a name="line-169"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-170"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-171"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"allocatedBytes"</span><span class='hs-layout'>)</span>
+<a name="line-172"></a>
+<a name="line-173"></a><a name="allocationId"></a><span class='hs-definition'>allocationId</span> <span class='hs-keyglyph'>::</span>
+<a name="line-174"></a>             <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-175"></a>               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allocationId"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-176"></a>               <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-177"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocationId"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-178"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocationId"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-179"></a><span class='hs-definition'>allocationId</span>
+<a name="line-180"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-181"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"allocationId"</span><span class='hs-layout'>)</span>
+<a name="line-182"></a>
+<a name="line-183"></a><a name="allocatorName"></a><span class='hs-definition'>allocatorName</span> <span class='hs-keyglyph'>::</span>
+<a name="line-184"></a>              <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-185"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allocatorName"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-186"></a>                <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-187"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocatorName"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-188"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocatorName"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-189"></a><span class='hs-definition'>allocatorName</span>
+<a name="line-190"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-191"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-192"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"allocatorName"</span><span class='hs-layout'>)</span>
+<a name="line-193"></a>
+<a name="line-194"></a><a name="hasSingleReference"></a><span class='hs-definition'>hasSingleReference</span> <span class='hs-keyglyph'>::</span>
+<a name="line-195"></a>                   <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-196"></a>                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"hasSingleReference"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-197"></a>                     <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-198"></a>                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"hasSingleReference"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-199"></a>                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"hasSingleReference"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-200"></a><span class='hs-definition'>hasSingleReference</span>
+<a name="line-201"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-202"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-203"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"hasSingleReference"</span><span class='hs-layout'>)</span>
+<a name="line-204"></a>
+<a name="line-205"></a><a name="ptr"></a><span class='hs-definition'>ptr</span> <span class='hs-keyglyph'>::</span>
+<a name="line-206"></a>    <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"ptr"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-207"></a>      <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"ptr"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-208"></a>        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"ptr"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-209"></a><span class='hs-definition'>ptr</span>
+<a name="line-210"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-211"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"ptr"</span><span class='hs-layout'>)</span>
+<a name="line-212"></a>
+<a name="line-213"></a><a name="requestedBytes"></a><span class='hs-definition'>requestedBytes</span> <span class='hs-keyglyph'>::</span>
+<a name="line-214"></a>               <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-215"></a>                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"requestedBytes"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-216"></a>                 <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-217"></a>                   <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"requestedBytes"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-218"></a>                   <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"requestedBytes"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-219"></a><span class='hs-definition'>requestedBytes</span>
+<a name="line-220"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-221"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-222"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"requestedBytes"</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-AttrValue.html b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-AttrValue.html
new file mode 100644
index 0000000..7cdd141
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-AttrValue.html
@@ -0,0 +1,762 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Proto/Tensorflow/Core/Framework/AttrValue.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>{- This file was auto-generated from tensorflow/core/framework/attr_value.proto by the proto-lens-protoc program. -}</span>
+<a name="line-2"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables, DataKinds, TypeFamilies,
+<a name="line-3"></a>  MultiParamTypeClasses, FlexibleContexts, FlexibleInstances,
+<a name="line-4"></a>  PatternSynonyms #-}</span>
+<a name="line-5"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-unused-imports#-}</span>
+<a name="line-6"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span> <span class='hs-keyword'>where</span>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Prelude</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-11"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-12"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-13"></a>       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-14"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-15"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-16"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-17"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-18"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-19"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-20"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-24"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-25"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span>
+<a name="line-26"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span>
+<a name="line-27"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span>
+<a name="line-28"></a>
+<a name="line-29"></a><a name="AttrValue"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AttrValue</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue's</span> <span class='hs-keyglyph'>::</span>
+<a name="line-30"></a>                           <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>,</span>
+<a name="line-31"></a>                           <span class='hs-sel'>_AttrValue'i</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-32"></a>                           <span class='hs-sel'>_AttrValue'f</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Float</span><span class='hs-layout'>,</span>
+<a name="line-33"></a>                           <span class='hs-sel'>_AttrValue'b</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-34"></a>                           <span class='hs-sel'>_AttrValue'type'</span> <span class='hs-keyglyph'>::</span>
+<a name="line-35"></a>                           <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span><span class='hs-layout'>,</span>
+<a name="line-36"></a>                           <span class='hs-sel'>_AttrValue'shape</span> <span class='hs-keyglyph'>::</span>
+<a name="line-37"></a>                           <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-38"></a>                             <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span><span class='hs-layout'>,</span>
+<a name="line-39"></a>                           <span class='hs-sel'>_AttrValue'tensor</span> <span class='hs-keyglyph'>::</span>
+<a name="line-40"></a>                           <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorProto</span><span class='hs-layout'>,</span>
+<a name="line-41"></a>                           <span class='hs-sel'>_AttrValue'list</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>AttrValue'ListValue</span><span class='hs-layout'>,</span>
+<a name="line-42"></a>                           <span class='hs-sel'>_AttrValue'func</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>NameAttrList</span><span class='hs-layout'>,</span>
+<a name="line-43"></a>                           <span class='hs-sel'>_AttrValue'placeholder</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>}</span>
+<a name="line-44"></a>               <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-45"></a>
+<a name="line-46"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"s"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-47"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-48"></a>
+<a name="line-49"></a><a name="instance%20Data.ProtoLens.HasField%20%22s%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"s"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyword'>where</span>
+<a name="line-50"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-51"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe's</span>
+<a name="line-52"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>)</span>
+<a name="line-53"></a>
+<a name="line-54"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe's"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-55"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-56"></a>
+<a name="line-57"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe's%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe's"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span>
+<a name="line-58"></a>         <span class='hs-keyword'>where</span>
+<a name="line-59"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-60"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue's</span>
+<a name="line-61"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue's</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-62"></a>
+<a name="line-63"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"i"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-64"></a>
+<a name="line-65"></a><a name="instance%20Data.ProtoLens.HasField%20%22i%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"i"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyword'>where</span>
+<a name="line-66"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-67"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'i</span>
+<a name="line-68"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>)</span>
+<a name="line-69"></a>
+<a name="line-70"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'i"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-71"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-72"></a>
+<a name="line-73"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'i%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'i"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span>
+<a name="line-74"></a>         <span class='hs-keyword'>where</span>
+<a name="line-75"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-76"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue'i</span>
+<a name="line-77"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'i</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-78"></a>
+<a name="line-79"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"f"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Float</span>
+<a name="line-80"></a>
+<a name="line-81"></a><a name="instance%20Data.ProtoLens.HasField%20%22f%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"f"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyword'>where</span>
+<a name="line-82"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-83"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'f</span>
+<a name="line-84"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>)</span>
+<a name="line-85"></a>
+<a name="line-86"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'f"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-87"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Float</span>
+<a name="line-88"></a>
+<a name="line-89"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'f%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'f"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span>
+<a name="line-90"></a>         <span class='hs-keyword'>where</span>
+<a name="line-91"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-92"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue'f</span>
+<a name="line-93"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'f</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-94"></a>
+<a name="line-95"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"b"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-96"></a>
+<a name="line-97"></a><a name="instance%20Data.ProtoLens.HasField%20%22b%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"b"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyword'>where</span>
+<a name="line-98"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-99"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'b</span>
+<a name="line-100"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>)</span>
+<a name="line-101"></a>
+<a name="line-102"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'b"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-103"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-104"></a>
+<a name="line-105"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'b%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'b"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span>
+<a name="line-106"></a>         <span class='hs-keyword'>where</span>
+<a name="line-107"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-108"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue'b</span>
+<a name="line-109"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'b</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-110"></a>
+<a name="line-111"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"type'"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-112"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span>
+<a name="line-113"></a>
+<a name="line-114"></a><a name="instance%20Data.ProtoLens.HasField%20%22type'%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"type'"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyword'>where</span>
+<a name="line-115"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-116"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'type'</span>
+<a name="line-117"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-118"></a>
+<a name="line-119"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'type'"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-120"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span>
+<a name="line-121"></a>
+<a name="line-122"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'type'%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'type'"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span>
+<a name="line-123"></a>         <span class='hs-keyword'>where</span>
+<a name="line-124"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-125"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue'type'</span>
+<a name="line-126"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'type'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-127"></a>
+<a name="line-128"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"shape"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-129"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span>
+<a name="line-130"></a>
+<a name="line-131"></a><a name="instance%20Data.ProtoLens.HasField%20%22shape%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"shape"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyword'>where</span>
+<a name="line-132"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-133"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'shape</span>
+<a name="line-134"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-135"></a>
+<a name="line-136"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'shape"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-137"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-138"></a>       <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span>
+<a name="line-139"></a>
+<a name="line-140"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'shape%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'shape"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span>
+<a name="line-141"></a>         <span class='hs-keyword'>where</span>
+<a name="line-142"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-143"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue'shape</span>
+<a name="line-144"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'shape</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-145"></a>
+<a name="line-146"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"tensor"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-147"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorProto</span>
+<a name="line-148"></a>
+<a name="line-149"></a><a name="instance%20Data.ProtoLens.HasField%20%22tensor%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"tensor"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyword'>where</span>
+<a name="line-150"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-151"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'tensor</span>
+<a name="line-152"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-153"></a>
+<a name="line-154"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'tensor"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-155"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorProto</span>
+<a name="line-156"></a>
+<a name="line-157"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'tensor%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'tensor"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span>
+<a name="line-158"></a>         <span class='hs-keyword'>where</span>
+<a name="line-159"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-160"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue'tensor</span>
+<a name="line-161"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'tensor</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-162"></a>
+<a name="line-163"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"list"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-164"></a>     <span class='hs-conid'>AttrValue'ListValue</span>
+<a name="line-165"></a>
+<a name="line-166"></a><a name="instance%20Data.ProtoLens.HasField%20%22list%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"list"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyword'>where</span>
+<a name="line-167"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-168"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'list</span>
+<a name="line-169"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-170"></a>
+<a name="line-171"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'list"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-172"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>AttrValue'ListValue</span>
+<a name="line-173"></a>
+<a name="line-174"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'list%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'list"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span>
+<a name="line-175"></a>         <span class='hs-keyword'>where</span>
+<a name="line-176"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-177"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue'list</span>
+<a name="line-178"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'list</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-179"></a>
+<a name="line-180"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"func"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NameAttrList</span>
+<a name="line-181"></a>
+<a name="line-182"></a><a name="instance%20Data.ProtoLens.HasField%20%22func%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"func"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyword'>where</span>
+<a name="line-183"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-184"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'func</span>
+<a name="line-185"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-186"></a>
+<a name="line-187"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'func"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-188"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>NameAttrList</span>
+<a name="line-189"></a>
+<a name="line-190"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'func%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'func"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span>
+<a name="line-191"></a>         <span class='hs-keyword'>where</span>
+<a name="line-192"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-193"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue'func</span>
+<a name="line-194"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'func</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-195"></a>
+<a name="line-196"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"placeholder"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-197"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-198"></a>
+<a name="line-199"></a><a name="instance%20Data.ProtoLens.HasField%20%22placeholder%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"placeholder"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-conid'>AttrValue</span>
+<a name="line-200"></a>         <span class='hs-keyword'>where</span>
+<a name="line-201"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-202"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'placeholder</span>
+<a name="line-203"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>)</span>
+<a name="line-204"></a>
+<a name="line-205"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'placeholder"</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-206"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-207"></a>
+<a name="line-208"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'placeholder%22%20AttrValue%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'placeholder"</span> <span class='hs-conid'>AttrValue</span>
+<a name="line-209"></a>         <span class='hs-conid'>AttrValue</span> <span class='hs-keyword'>where</span>
+<a name="line-210"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-211"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue'placeholder</span>
+<a name="line-212"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'placeholder</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-213"></a>
+<a name="line-214"></a><a name="instance%20Data.Default.Class.Default%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyword'>where</span>
+<a name="line-215"></a>        <span class='hs-varid'>def</span>
+<a name="line-216"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AttrValue</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue's</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-217"></a>                      <span class='hs-sel'>_AttrValue'i</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span> <span class='hs-sel'>_AttrValue'f</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-218"></a>                      <span class='hs-sel'>_AttrValue'b</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span> <span class='hs-sel'>_AttrValue'type'</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-219"></a>                      <span class='hs-sel'>_AttrValue'shape</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-220"></a>                      <span class='hs-sel'>_AttrValue'tensor</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-221"></a>                      <span class='hs-sel'>_AttrValue'list</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-222"></a>                      <span class='hs-sel'>_AttrValue'func</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-223"></a>                      <span class='hs-sel'>_AttrValue'placeholder</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>}</span>
+<a name="line-224"></a>
+<a name="line-225"></a><a name="instance%20Data.ProtoLens.Message%20AttrValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>AttrValue</span> <span class='hs-keyword'>where</span>
+<a name="line-226"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-227"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>s__field_descriptor</span>
+<a name="line-228"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"s"</span>
+<a name="line-229"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BytesField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-230"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>
+<a name="line-231"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe's</span><span class='hs-layout'>)</span>
+<a name="line-232"></a>                <span class='hs-varid'>i__field_descriptor</span>
+<a name="line-233"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"i"</span>
+<a name="line-234"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-235"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-236"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'i</span><span class='hs-layout'>)</span>
+<a name="line-237"></a>                <span class='hs-varid'>f__field_descriptor</span>
+<a name="line-238"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"f"</span>
+<a name="line-239"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FloatField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-240"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Float</span><span class='hs-layout'>)</span>
+<a name="line-241"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'f</span><span class='hs-layout'>)</span>
+<a name="line-242"></a>                <span class='hs-varid'>b__field_descriptor</span>
+<a name="line-243"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"b"</span>
+<a name="line-244"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-245"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-246"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'b</span><span class='hs-layout'>)</span>
+<a name="line-247"></a>                <span class='hs-varid'>type'__field_descriptor</span>
+<a name="line-248"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"type"</span>
+<a name="line-249"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>EnumField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-250"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-251"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span><span class='hs-layout'>)</span>
+<a name="line-252"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'type'</span><span class='hs-layout'>)</span>
+<a name="line-253"></a>                <span class='hs-varid'>shape__field_descriptor</span>
+<a name="line-254"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"shape"</span>
+<a name="line-255"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-256"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-257"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span><span class='hs-layout'>)</span>
+<a name="line-258"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'shape</span><span class='hs-layout'>)</span>
+<a name="line-259"></a>                <span class='hs-varid'>tensor__field_descriptor</span>
+<a name="line-260"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"tensor"</span>
+<a name="line-261"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-262"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-263"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorProto</span><span class='hs-layout'>)</span>
+<a name="line-264"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'tensor</span><span class='hs-layout'>)</span>
+<a name="line-265"></a>                <span class='hs-varid'>list__field_descriptor</span>
+<a name="line-266"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"list"</span>
+<a name="line-267"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-268"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>AttrValue'ListValue</span><span class='hs-layout'>)</span>
+<a name="line-269"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'list</span><span class='hs-layout'>)</span>
+<a name="line-270"></a>                <span class='hs-varid'>func__field_descriptor</span>
+<a name="line-271"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"func"</span>
+<a name="line-272"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-273"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>NameAttrList</span><span class='hs-layout'>)</span>
+<a name="line-274"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'func</span><span class='hs-layout'>)</span>
+<a name="line-275"></a>                <span class='hs-varid'>placeholder__field_descriptor</span>
+<a name="line-276"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"placeholder"</span>
+<a name="line-277"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-278"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-279"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'placeholder</span><span class='hs-layout'>)</span>
+<a name="line-280"></a>              <span class='hs-keyword'>in</span>
+<a name="line-281"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-282"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-283"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>s__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-284"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>i__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-285"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>f__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-286"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>5</span><span class='hs-layout'>,</span> <span class='hs-varid'>b__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-287"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>6</span><span class='hs-layout'>,</span> <span class='hs-varid'>type'__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-288"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>7</span><span class='hs-layout'>,</span> <span class='hs-varid'>shape__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-289"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>8</span><span class='hs-layout'>,</span> <span class='hs-varid'>tensor__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-290"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>list__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-291"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>10</span><span class='hs-layout'>,</span> <span class='hs-varid'>func__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-292"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>9</span><span class='hs-layout'>,</span> <span class='hs-varid'>placeholder__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-293"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-294"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"s"</span><span class='hs-layout'>,</span> <span class='hs-varid'>s__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-str'>"i"</span><span class='hs-layout'>,</span> <span class='hs-varid'>i__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-295"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"f"</span><span class='hs-layout'>,</span> <span class='hs-varid'>f__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-str'>"b"</span><span class='hs-layout'>,</span> <span class='hs-varid'>b__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-296"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"type"</span><span class='hs-layout'>,</span> <span class='hs-varid'>type'__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-297"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"shape"</span><span class='hs-layout'>,</span> <span class='hs-varid'>shape__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-298"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"tensor"</span><span class='hs-layout'>,</span> <span class='hs-varid'>tensor__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-299"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"list"</span><span class='hs-layout'>,</span> <span class='hs-varid'>list__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-str'>"func"</span><span class='hs-layout'>,</span> <span class='hs-varid'>func__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-300"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"placeholder"</span><span class='hs-layout'>,</span> <span class='hs-varid'>placeholder__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-301"></a>
+<a name="line-302"></a><a name="AttrValue'ListValue"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AttrValue'ListValue</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'ListValue's</span>
+<a name="line-303"></a>                                               <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-304"></a>                                               <span class='hs-sel'>_AttrValue'ListValue'i</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-305"></a>                                               <span class='hs-sel'>_AttrValue'ListValue'f</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-306"></a>                                               <span class='hs-sel'>_AttrValue'ListValue'b</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-307"></a>                                               <span class='hs-sel'>_AttrValue'ListValue'type'</span> <span class='hs-keyglyph'>::</span>
+<a name="line-308"></a>                                               <span class='hs-keyglyph'>[</span><span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-309"></a>                                               <span class='hs-sel'>_AttrValue'ListValue'shape</span> <span class='hs-keyglyph'>::</span>
+<a name="line-310"></a>                                               <span class='hs-keyglyph'>[</span><span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-311"></a>                                               <span class='hs-sel'>_AttrValue'ListValue'tensor</span> <span class='hs-keyglyph'>::</span>
+<a name="line-312"></a>                                               <span class='hs-keyglyph'>[</span><span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorProto</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>}</span>
+<a name="line-313"></a>                         <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-314"></a>
+<a name="line-315"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"s"</span> <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-316"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-keyglyph'>]</span>
+<a name="line-317"></a>
+<a name="line-318"></a><a name="instance%20Data.ProtoLens.HasField%20%22s%22%20AttrValue'ListValue%20AttrValue'ListValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"s"</span> <span class='hs-conid'>AttrValue'ListValue</span>
+<a name="line-319"></a>         <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyword'>where</span>
+<a name="line-320"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-321"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue'ListValue's</span>
+<a name="line-322"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'ListValue's</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-323"></a>
+<a name="line-324"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"i"</span> <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-325"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span>
+<a name="line-326"></a>
+<a name="line-327"></a><a name="instance%20Data.ProtoLens.HasField%20%22i%22%20AttrValue'ListValue%20AttrValue'ListValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"i"</span> <span class='hs-conid'>AttrValue'ListValue</span>
+<a name="line-328"></a>         <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyword'>where</span>
+<a name="line-329"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-330"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue'ListValue'i</span>
+<a name="line-331"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'ListValue'i</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-332"></a>
+<a name="line-333"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"f"</span> <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-334"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span>
+<a name="line-335"></a>
+<a name="line-336"></a><a name="instance%20Data.ProtoLens.HasField%20%22f%22%20AttrValue'ListValue%20AttrValue'ListValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"f"</span> <span class='hs-conid'>AttrValue'ListValue</span>
+<a name="line-337"></a>         <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyword'>where</span>
+<a name="line-338"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-339"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue'ListValue'f</span>
+<a name="line-340"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'ListValue'f</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-341"></a>
+<a name="line-342"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"b"</span> <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-343"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-keyglyph'>]</span>
+<a name="line-344"></a>
+<a name="line-345"></a><a name="instance%20Data.ProtoLens.HasField%20%22b%22%20AttrValue'ListValue%20AttrValue'ListValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"b"</span> <span class='hs-conid'>AttrValue'ListValue</span>
+<a name="line-346"></a>         <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyword'>where</span>
+<a name="line-347"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-348"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue'ListValue'b</span>
+<a name="line-349"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'ListValue'b</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-350"></a>
+<a name="line-351"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"type'"</span> <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-352"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span><span class='hs-keyglyph'>]</span>
+<a name="line-353"></a>
+<a name="line-354"></a><a name="instance%20Data.ProtoLens.HasField%20%22type'%22%20AttrValue'ListValue%20AttrValue'ListValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"type'"</span> <span class='hs-conid'>AttrValue'ListValue</span>
+<a name="line-355"></a>         <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyword'>where</span>
+<a name="line-356"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-357"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue'ListValue'type'</span>
+<a name="line-358"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'ListValue'type'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-359"></a>
+<a name="line-360"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"shape"</span> <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-361"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span><span class='hs-keyglyph'>]</span>
+<a name="line-362"></a>
+<a name="line-363"></a><a name="instance%20Data.ProtoLens.HasField%20%22shape%22%20AttrValue'ListValue%20AttrValue'ListValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"shape"</span> <span class='hs-conid'>AttrValue'ListValue</span>
+<a name="line-364"></a>         <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyword'>where</span>
+<a name="line-365"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-366"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue'ListValue'shape</span>
+<a name="line-367"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'ListValue'shape</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-368"></a>
+<a name="line-369"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"tensor"</span> <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyglyph'>=</span>
+<a name="line-370"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorProto</span><span class='hs-keyglyph'>]</span>
+<a name="line-371"></a>
+<a name="line-372"></a><a name="instance%20Data.ProtoLens.HasField%20%22tensor%22%20AttrValue'ListValue%20AttrValue'ListValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"tensor"</span> <span class='hs-conid'>AttrValue'ListValue</span>
+<a name="line-373"></a>         <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyword'>where</span>
+<a name="line-374"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-375"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AttrValue'ListValue'tensor</span>
+<a name="line-376"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'ListValue'tensor</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-377"></a>
+<a name="line-378"></a><a name="instance%20Data.Default.Class.Default%20AttrValue'ListValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyword'>where</span>
+<a name="line-379"></a>        <span class='hs-varid'>def</span>
+<a name="line-380"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AttrValue'ListValue</span><span class='hs-layout'>{</span><span class='hs-sel'>_AttrValue'ListValue's</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-381"></a>                                <span class='hs-sel'>_AttrValue'ListValue'i</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span> <span class='hs-sel'>_AttrValue'ListValue'f</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-382"></a>                                <span class='hs-sel'>_AttrValue'ListValue'b</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span> <span class='hs-sel'>_AttrValue'ListValue'type'</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-383"></a>                                <span class='hs-sel'>_AttrValue'ListValue'shape</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span> <span class='hs-sel'>_AttrValue'ListValue'tensor</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>}</span>
+<a name="line-384"></a>
+<a name="line-385"></a><a name="instance%20Data.ProtoLens.Message%20AttrValue'ListValue"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>AttrValue'ListValue</span> <span class='hs-keyword'>where</span>
+<a name="line-386"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-387"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>s__field_descriptor</span>
+<a name="line-388"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"s"</span>
+<a name="line-389"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BytesField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-390"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>
+<a name="line-391"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>s</span><span class='hs-layout'>)</span>
+<a name="line-392"></a>                <span class='hs-varid'>i__field_descriptor</span>
+<a name="line-393"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"i"</span>
+<a name="line-394"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-395"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-396"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Packed</span> <span class='hs-varid'>i</span><span class='hs-layout'>)</span>
+<a name="line-397"></a>                <span class='hs-varid'>f__field_descriptor</span>
+<a name="line-398"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"f"</span>
+<a name="line-399"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FloatField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-400"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Float</span><span class='hs-layout'>)</span>
+<a name="line-401"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Packed</span> <span class='hs-varid'>f</span><span class='hs-layout'>)</span>
+<a name="line-402"></a>                <span class='hs-varid'>b__field_descriptor</span>
+<a name="line-403"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"b"</span>
+<a name="line-404"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-405"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-406"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Packed</span> <span class='hs-varid'>b</span><span class='hs-layout'>)</span>
+<a name="line-407"></a>                <span class='hs-varid'>type'__field_descriptor</span>
+<a name="line-408"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"type"</span>
+<a name="line-409"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>EnumField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-410"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-411"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span><span class='hs-layout'>)</span>
+<a name="line-412"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Packed</span> <span class='hs-varid'>type'</span><span class='hs-layout'>)</span>
+<a name="line-413"></a>                <span class='hs-varid'>shape__field_descriptor</span>
+<a name="line-414"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"shape"</span>
+<a name="line-415"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-416"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-417"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span><span class='hs-layout'>)</span>
+<a name="line-418"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>shape</span><span class='hs-layout'>)</span>
+<a name="line-419"></a>                <span class='hs-varid'>tensor__field_descriptor</span>
+<a name="line-420"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"tensor"</span>
+<a name="line-421"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-422"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-423"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorProto</span><span class='hs-layout'>)</span>
+<a name="line-424"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>tensor</span><span class='hs-layout'>)</span>
+<a name="line-425"></a>              <span class='hs-keyword'>in</span>
+<a name="line-426"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-427"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-428"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>s__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-429"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>i__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-430"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>f__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-431"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>5</span><span class='hs-layout'>,</span> <span class='hs-varid'>b__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-432"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>6</span><span class='hs-layout'>,</span> <span class='hs-varid'>type'__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-433"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>7</span><span class='hs-layout'>,</span> <span class='hs-varid'>shape__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-434"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>8</span><span class='hs-layout'>,</span> <span class='hs-varid'>tensor__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-435"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-436"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"s"</span><span class='hs-layout'>,</span> <span class='hs-varid'>s__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-str'>"i"</span><span class='hs-layout'>,</span> <span class='hs-varid'>i__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-437"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"f"</span><span class='hs-layout'>,</span> <span class='hs-varid'>f__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-str'>"b"</span><span class='hs-layout'>,</span> <span class='hs-varid'>b__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-438"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"type"</span><span class='hs-layout'>,</span> <span class='hs-varid'>type'__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-439"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"shape"</span><span class='hs-layout'>,</span> <span class='hs-varid'>shape__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-440"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"tensor"</span><span class='hs-layout'>,</span> <span class='hs-varid'>tensor__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-441"></a>
+<a name="line-442"></a><a name="NameAttrList"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>NameAttrList</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NameAttrList</span><span class='hs-layout'>{</span><span class='hs-sel'>_NameAttrList'name</span> <span class='hs-keyglyph'>::</span>
+<a name="line-443"></a>                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-444"></a>                                 <span class='hs-sel'>_NameAttrList'attr</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-conid'>AttrValue</span><span class='hs-layout'>}</span>
+<a name="line-445"></a>                  <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-446"></a>
+<a name="line-447"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-conid'>NameAttrList</span> <span class='hs-keyglyph'>=</span>
+<a name="line-448"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-449"></a>
+<a name="line-450"></a><a name="instance%20Data.ProtoLens.HasField%20%22name%22%20NameAttrList%20NameAttrList"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"name"</span> <span class='hs-conid'>NameAttrList</span> <span class='hs-conid'>NameAttrList</span>
+<a name="line-451"></a>         <span class='hs-keyword'>where</span>
+<a name="line-452"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-453"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NameAttrList'name</span>
+<a name="line-454"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NameAttrList'name</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-455"></a>
+<a name="line-456"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"attr"</span> <span class='hs-conid'>NameAttrList</span> <span class='hs-keyglyph'>=</span>
+<a name="line-457"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-conid'>AttrValue</span>
+<a name="line-458"></a>
+<a name="line-459"></a><a name="instance%20Data.ProtoLens.HasField%20%22attr%22%20NameAttrList%20NameAttrList"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"attr"</span> <span class='hs-conid'>NameAttrList</span> <span class='hs-conid'>NameAttrList</span>
+<a name="line-460"></a>         <span class='hs-keyword'>where</span>
+<a name="line-461"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-462"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NameAttrList'attr</span>
+<a name="line-463"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NameAttrList'attr</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-464"></a>
+<a name="line-465"></a><a name="instance%20Data.Default.Class.Default%20NameAttrList"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>NameAttrList</span> <span class='hs-keyword'>where</span>
+<a name="line-466"></a>        <span class='hs-varid'>def</span>
+<a name="line-467"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NameAttrList</span><span class='hs-layout'>{</span><span class='hs-sel'>_NameAttrList'name</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-468"></a>                         <span class='hs-sel'>_NameAttrList'attr</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>empty</span><span class='hs-layout'>}</span>
+<a name="line-469"></a>
+<a name="line-470"></a><a name="instance%20Data.ProtoLens.Message%20NameAttrList"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>NameAttrList</span> <span class='hs-keyword'>where</span>
+<a name="line-471"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-472"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>name__field_descriptor</span>
+<a name="line-473"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"name"</span>
+<a name="line-474"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-475"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-476"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span>
+<a name="line-477"></a>                <span class='hs-varid'>attr__field_descriptor</span>
+<a name="line-478"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"attr"</span>
+<a name="line-479"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-480"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>NameAttrList'AttrEntry</span><span class='hs-layout'>)</span>
+<a name="line-481"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MapField</span> <span class='hs-varid'>key</span> <span class='hs-varid'>value</span> <span class='hs-varid'>attr</span><span class='hs-layout'>)</span>
+<a name="line-482"></a>              <span class='hs-keyword'>in</span>
+<a name="line-483"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-484"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-485"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>name__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-486"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>attr__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-487"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-488"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"name"</span><span class='hs-layout'>,</span> <span class='hs-varid'>name__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-489"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"attr"</span><span class='hs-layout'>,</span> <span class='hs-varid'>attr__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-490"></a>
+<a name="line-491"></a><a name="NameAttrList'AttrEntry"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>NameAttrList'AttrEntry</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NameAttrList'AttrEntry</span><span class='hs-layout'>{</span><span class='hs-sel'>_NameAttrList'AttrEntry'key</span>
+<a name="line-492"></a>                                                     <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-493"></a>                                                     <span class='hs-sel'>_NameAttrList'AttrEntry'value</span> <span class='hs-keyglyph'>::</span>
+<a name="line-494"></a>                                                     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>AttrValue</span><span class='hs-layout'>}</span>
+<a name="line-495"></a>                            <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-496"></a>
+<a name="line-497"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"key"</span> <span class='hs-conid'>NameAttrList'AttrEntry</span> <span class='hs-keyglyph'>=</span>
+<a name="line-498"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-499"></a>
+<a name="line-500"></a><a name="instance%20Data.ProtoLens.HasField%20%22key%22%20NameAttrList'AttrEntry%20NameAttrList'AttrEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"key"</span> <span class='hs-conid'>NameAttrList'AttrEntry</span>
+<a name="line-501"></a>         <span class='hs-conid'>NameAttrList'AttrEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-502"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-503"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NameAttrList'AttrEntry'key</span>
+<a name="line-504"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NameAttrList'AttrEntry'key</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-505"></a>
+<a name="line-506"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"value"</span> <span class='hs-conid'>NameAttrList'AttrEntry</span> <span class='hs-keyglyph'>=</span>
+<a name="line-507"></a>     <span class='hs-conid'>AttrValue</span>
+<a name="line-508"></a>
+<a name="line-509"></a><a name="instance%20Data.ProtoLens.HasField%20%22value%22%20NameAttrList'AttrEntry%20NameAttrList'AttrEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"value"</span> <span class='hs-conid'>NameAttrList'AttrEntry</span>
+<a name="line-510"></a>         <span class='hs-conid'>NameAttrList'AttrEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-511"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-512"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'value</span>
+<a name="line-513"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-514"></a>
+<a name="line-515"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-516"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'value"</span> <span class='hs-conid'>NameAttrList'AttrEntry</span> <span class='hs-keyglyph'>=</span>
+<a name="line-517"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>AttrValue</span>
+<a name="line-518"></a>
+<a name="line-519"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'value%22%20NameAttrList'AttrEntry%20NameAttrList'AttrEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'value"</span>
+<a name="line-520"></a>         <span class='hs-conid'>NameAttrList'AttrEntry</span> <span class='hs-conid'>NameAttrList'AttrEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-521"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-522"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NameAttrList'AttrEntry'value</span>
+<a name="line-523"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NameAttrList'AttrEntry'value</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-524"></a>
+<a name="line-525"></a><a name="instance%20Data.Default.Class.Default%20NameAttrList'AttrEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>NameAttrList'AttrEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-526"></a>        <span class='hs-varid'>def</span>
+<a name="line-527"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NameAttrList'AttrEntry</span><span class='hs-layout'>{</span><span class='hs-sel'>_NameAttrList'AttrEntry'key</span> <span class='hs-keyglyph'>=</span>
+<a name="line-528"></a>                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-529"></a>                                   <span class='hs-sel'>_NameAttrList'AttrEntry'value</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>}</span>
+<a name="line-530"></a>
+<a name="line-531"></a><a name="instance%20Data.ProtoLens.Message%20NameAttrList'AttrEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>NameAttrList'AttrEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-532"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-533"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>key__field_descriptor</span>
+<a name="line-534"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"key"</span>
+<a name="line-535"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-536"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-537"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>key</span><span class='hs-layout'>)</span>
+<a name="line-538"></a>                <span class='hs-varid'>value__field_descriptor</span>
+<a name="line-539"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"value"</span>
+<a name="line-540"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-541"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>AttrValue</span><span class='hs-layout'>)</span>
+<a name="line-542"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'value</span><span class='hs-layout'>)</span>
+<a name="line-543"></a>              <span class='hs-keyword'>in</span>
+<a name="line-544"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-545"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-546"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>key__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-547"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>value__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-548"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-549"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"key"</span><span class='hs-layout'>,</span> <span class='hs-varid'>key__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-550"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"value"</span><span class='hs-layout'>,</span> <span class='hs-varid'>value__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-551"></a>
+<a name="line-552"></a><a name="attr"></a><span class='hs-definition'>attr</span> <span class='hs-keyglyph'>::</span>
+<a name="line-553"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"attr"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-554"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"attr"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-555"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"attr"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-556"></a><span class='hs-definition'>attr</span>
+<a name="line-557"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-558"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"attr"</span><span class='hs-layout'>)</span>
+<a name="line-559"></a>
+<a name="line-560"></a><a name="b"></a><span class='hs-definition'>b</span> <span class='hs-keyglyph'>::</span>
+<a name="line-561"></a>  <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"b"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-562"></a>    <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"b"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-563"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"b"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-564"></a><span class='hs-definition'>b</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-565"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"b"</span><span class='hs-layout'>)</span>
+<a name="line-566"></a>
+<a name="line-567"></a><a name="f"></a><span class='hs-definition'>f</span> <span class='hs-keyglyph'>::</span>
+<a name="line-568"></a>  <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"f"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-569"></a>    <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"f"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-570"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"f"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-571"></a><span class='hs-definition'>f</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-572"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"f"</span><span class='hs-layout'>)</span>
+<a name="line-573"></a>
+<a name="line-574"></a><a name="func"></a><span class='hs-definition'>func</span> <span class='hs-keyglyph'>::</span>
+<a name="line-575"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"func"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-576"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"func"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-577"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"func"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-578"></a><span class='hs-definition'>func</span>
+<a name="line-579"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-580"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"func"</span><span class='hs-layout'>)</span>
+<a name="line-581"></a>
+<a name="line-582"></a><a name="i"></a><span class='hs-definition'>i</span> <span class='hs-keyglyph'>::</span>
+<a name="line-583"></a>  <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"i"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-584"></a>    <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"i"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-585"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"i"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-586"></a><span class='hs-definition'>i</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-587"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"i"</span><span class='hs-layout'>)</span>
+<a name="line-588"></a>
+<a name="line-589"></a><a name="key"></a><span class='hs-definition'>key</span> <span class='hs-keyglyph'>::</span>
+<a name="line-590"></a>    <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"key"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-591"></a>      <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"key"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-592"></a>        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"key"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-593"></a><span class='hs-definition'>key</span>
+<a name="line-594"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-595"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"key"</span><span class='hs-layout'>)</span>
+<a name="line-596"></a>
+<a name="line-597"></a><a name="list"></a><span class='hs-definition'>list</span> <span class='hs-keyglyph'>::</span>
+<a name="line-598"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"list"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-599"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"list"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-600"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"list"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-601"></a><span class='hs-definition'>list</span>
+<a name="line-602"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-603"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"list"</span><span class='hs-layout'>)</span>
+<a name="line-604"></a>
+<a name="line-605"></a><a name="maybe'b"></a><span class='hs-definition'>maybe'b</span> <span class='hs-keyglyph'>::</span>
+<a name="line-606"></a>        <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'b"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-607"></a>          <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'b"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-608"></a>            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'b"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-609"></a><span class='hs-definition'>maybe'b</span>
+<a name="line-610"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-611"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'b"</span><span class='hs-layout'>)</span>
+<a name="line-612"></a>
+<a name="line-613"></a><a name="maybe'f"></a><span class='hs-definition'>maybe'f</span> <span class='hs-keyglyph'>::</span>
+<a name="line-614"></a>        <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'f"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-615"></a>          <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'f"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-616"></a>            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'f"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-617"></a><span class='hs-definition'>maybe'f</span>
+<a name="line-618"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-619"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'f"</span><span class='hs-layout'>)</span>
+<a name="line-620"></a>
+<a name="line-621"></a><a name="maybe'func"></a><span class='hs-definition'>maybe'func</span> <span class='hs-keyglyph'>::</span>
+<a name="line-622"></a>           <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'func"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-623"></a>             <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'func"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-624"></a>               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'func"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-625"></a><span class='hs-definition'>maybe'func</span>
+<a name="line-626"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-627"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'func"</span><span class='hs-layout'>)</span>
+<a name="line-628"></a>
+<a name="line-629"></a><a name="maybe'i"></a><span class='hs-definition'>maybe'i</span> <span class='hs-keyglyph'>::</span>
+<a name="line-630"></a>        <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'i"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-631"></a>          <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'i"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-632"></a>            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'i"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-633"></a><span class='hs-definition'>maybe'i</span>
+<a name="line-634"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-635"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'i"</span><span class='hs-layout'>)</span>
+<a name="line-636"></a>
+<a name="line-637"></a><a name="maybe'list"></a><span class='hs-definition'>maybe'list</span> <span class='hs-keyglyph'>::</span>
+<a name="line-638"></a>           <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'list"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-639"></a>             <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'list"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-640"></a>               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'list"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-641"></a><span class='hs-definition'>maybe'list</span>
+<a name="line-642"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-643"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'list"</span><span class='hs-layout'>)</span>
+<a name="line-644"></a>
+<a name="line-645"></a><a name="maybe'placeholder"></a><span class='hs-definition'>maybe'placeholder</span> <span class='hs-keyglyph'>::</span>
+<a name="line-646"></a>                  <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-647"></a>                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'placeholder"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-648"></a>                    <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-649"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'placeholder"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-650"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'placeholder"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-651"></a><span class='hs-definition'>maybe'placeholder</span>
+<a name="line-652"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-653"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-654"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'placeholder"</span><span class='hs-layout'>)</span>
+<a name="line-655"></a>
+<a name="line-656"></a><a name="maybe's"></a><span class='hs-definition'>maybe's</span> <span class='hs-keyglyph'>::</span>
+<a name="line-657"></a>        <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe's"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-658"></a>          <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe's"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-659"></a>            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe's"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-660"></a><span class='hs-definition'>maybe's</span>
+<a name="line-661"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-662"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe's"</span><span class='hs-layout'>)</span>
+<a name="line-663"></a>
+<a name="line-664"></a><a name="maybe'shape"></a><span class='hs-definition'>maybe'shape</span> <span class='hs-keyglyph'>::</span>
+<a name="line-665"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'shape"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-666"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'shape"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-667"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'shape"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-668"></a><span class='hs-definition'>maybe'shape</span>
+<a name="line-669"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-670"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'shape"</span><span class='hs-layout'>)</span>
+<a name="line-671"></a>
+<a name="line-672"></a><a name="maybe'tensor"></a><span class='hs-definition'>maybe'tensor</span> <span class='hs-keyglyph'>::</span>
+<a name="line-673"></a>             <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-674"></a>               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'tensor"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-675"></a>               <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-676"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'tensor"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-677"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'tensor"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-678"></a><span class='hs-definition'>maybe'tensor</span>
+<a name="line-679"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-680"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'tensor"</span><span class='hs-layout'>)</span>
+<a name="line-681"></a>
+<a name="line-682"></a><a name="maybe'type'"></a><span class='hs-definition'>maybe'type'</span> <span class='hs-keyglyph'>::</span>
+<a name="line-683"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'type'"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-684"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'type'"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-685"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'type'"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-686"></a><span class='hs-definition'>maybe'type'</span>
+<a name="line-687"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-688"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'type'"</span><span class='hs-layout'>)</span>
+<a name="line-689"></a>
+<a name="line-690"></a><a name="maybe'value"></a><span class='hs-definition'>maybe'value</span> <span class='hs-keyglyph'>::</span>
+<a name="line-691"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'value"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-692"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'value"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-693"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'value"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-694"></a><span class='hs-definition'>maybe'value</span>
+<a name="line-695"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-696"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'value"</span><span class='hs-layout'>)</span>
+<a name="line-697"></a>
+<a name="line-698"></a><a name="name"></a><span class='hs-definition'>name</span> <span class='hs-keyglyph'>::</span>
+<a name="line-699"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-700"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-701"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-702"></a><span class='hs-definition'>name</span>
+<a name="line-703"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-704"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"name"</span><span class='hs-layout'>)</span>
+<a name="line-705"></a>
+<a name="line-706"></a><a name="placeholder"></a><span class='hs-definition'>placeholder</span> <span class='hs-keyglyph'>::</span>
+<a name="line-707"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"placeholder"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-708"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"placeholder"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-709"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"placeholder"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-710"></a><span class='hs-definition'>placeholder</span>
+<a name="line-711"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-712"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"placeholder"</span><span class='hs-layout'>)</span>
+<a name="line-713"></a>
+<a name="line-714"></a><a name="s"></a><span class='hs-definition'>s</span> <span class='hs-keyglyph'>::</span>
+<a name="line-715"></a>  <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"s"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-716"></a>    <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"s"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-717"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"s"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-718"></a><span class='hs-definition'>s</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-719"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"s"</span><span class='hs-layout'>)</span>
+<a name="line-720"></a>
+<a name="line-721"></a><a name="shape"></a><span class='hs-definition'>shape</span> <span class='hs-keyglyph'>::</span>
+<a name="line-722"></a>      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"shape"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-723"></a>        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"shape"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-724"></a>          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"shape"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-725"></a><span class='hs-definition'>shape</span>
+<a name="line-726"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-727"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"shape"</span><span class='hs-layout'>)</span>
+<a name="line-728"></a>
+<a name="line-729"></a><a name="tensor"></a><span class='hs-definition'>tensor</span> <span class='hs-keyglyph'>::</span>
+<a name="line-730"></a>       <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"tensor"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-731"></a>         <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"tensor"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-732"></a>           <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"tensor"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-733"></a><span class='hs-definition'>tensor</span>
+<a name="line-734"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-735"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"tensor"</span><span class='hs-layout'>)</span>
+<a name="line-736"></a>
+<a name="line-737"></a><a name="type'"></a><span class='hs-definition'>type'</span> <span class='hs-keyglyph'>::</span>
+<a name="line-738"></a>      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"type'"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-739"></a>        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"type'"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-740"></a>          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"type'"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-741"></a><span class='hs-definition'>type'</span>
+<a name="line-742"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-743"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"type'"</span><span class='hs-layout'>)</span>
+<a name="line-744"></a>
+<a name="line-745"></a><a name="value"></a><span class='hs-definition'>value</span> <span class='hs-keyglyph'>::</span>
+<a name="line-746"></a>      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"value"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-747"></a>        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"value"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-748"></a>          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"value"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-749"></a><span class='hs-definition'>value</span>
+<a name="line-750"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-751"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"value"</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-CostGraph.html b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-CostGraph.html
new file mode 100644
index 0000000..4fd4bae
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-CostGraph.html
@@ -0,0 +1,570 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Proto/Tensorflow/Core/Framework/CostGraph.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>{- This file was auto-generated from tensorflow/core/framework/cost_graph.proto by the proto-lens-protoc program. -}</span>
+<a name="line-2"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables, DataKinds, TypeFamilies,
+<a name="line-3"></a>  MultiParamTypeClasses, FlexibleContexts, FlexibleInstances,
+<a name="line-4"></a>  PatternSynonyms #-}</span>
+<a name="line-5"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-unused-imports#-}</span>
+<a name="line-6"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>CostGraph</span> <span class='hs-keyword'>where</span>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Prelude</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-11"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-12"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-13"></a>       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-14"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-15"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-16"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-17"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-18"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-19"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-20"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-24"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-25"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span>
+<a name="line-26"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span>
+<a name="line-27"></a>
+<a name="line-28"></a><a name="CostGraphDef"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>CostGraphDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CostGraphDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'node</span> <span class='hs-keyglyph'>::</span>
+<a name="line-29"></a>                                 <span class='hs-keyglyph'>[</span><span class='hs-conid'>CostGraphDef'Node</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>}</span>
+<a name="line-30"></a>                  <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-31"></a>
+<a name="line-32"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"node"</span> <span class='hs-conid'>CostGraphDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-33"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>CostGraphDef'Node</span><span class='hs-keyglyph'>]</span>
+<a name="line-34"></a>
+<a name="line-35"></a><a name="instance%20Data.ProtoLens.HasField%20%22node%22%20CostGraphDef%20CostGraphDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"node"</span> <span class='hs-conid'>CostGraphDef</span> <span class='hs-conid'>CostGraphDef</span>
+<a name="line-36"></a>         <span class='hs-keyword'>where</span>
+<a name="line-37"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-38"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_CostGraphDef'node</span>
+<a name="line-39"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'node</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-40"></a>
+<a name="line-41"></a><a name="instance%20Data.Default.Class.Default%20CostGraphDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>CostGraphDef</span> <span class='hs-keyword'>where</span>
+<a name="line-42"></a>        <span class='hs-varid'>def</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CostGraphDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'node</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>}</span>
+<a name="line-43"></a>
+<a name="line-44"></a><a name="instance%20Data.ProtoLens.Message%20CostGraphDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>CostGraphDef</span> <span class='hs-keyword'>where</span>
+<a name="line-45"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-46"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>node__field_descriptor</span>
+<a name="line-47"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"node"</span>
+<a name="line-48"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-49"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>CostGraphDef'Node</span><span class='hs-layout'>)</span>
+<a name="line-50"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>node</span><span class='hs-layout'>)</span>
+<a name="line-51"></a>              <span class='hs-keyword'>in</span>
+<a name="line-52"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-53"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-54"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>node__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-55"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"node"</span><span class='hs-layout'>,</span> <span class='hs-varid'>node__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-56"></a>
+<a name="line-57"></a><a name="CostGraphDef'Node"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CostGraphDef'Node</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'name</span>
+<a name="line-58"></a>                                           <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-59"></a>                                           <span class='hs-sel'>_CostGraphDef'Node'device</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-60"></a>                                           <span class='hs-sel'>_CostGraphDef'Node'id</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-61"></a>                                           <span class='hs-sel'>_CostGraphDef'Node'inputInfo</span> <span class='hs-keyglyph'>::</span>
+<a name="line-62"></a>                                           <span class='hs-keyglyph'>[</span><span class='hs-conid'>CostGraphDef'Node'InputInfo</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-63"></a>                                           <span class='hs-sel'>_CostGraphDef'Node'outputInfo</span> <span class='hs-keyglyph'>::</span>
+<a name="line-64"></a>                                           <span class='hs-keyglyph'>[</span><span class='hs-conid'>CostGraphDef'Node'OutputInfo</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-65"></a>                                           <span class='hs-sel'>_CostGraphDef'Node'temporaryMemorySize</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-66"></a>                                           <span class='hs-sel'>_CostGraphDef'Node'computeCost</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-67"></a>                                           <span class='hs-sel'>_CostGraphDef'Node'isFinal</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-68"></a>                                           <span class='hs-sel'>_CostGraphDef'Node'controlInput</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>}</span>
+<a name="line-69"></a>                       <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-70"></a>
+<a name="line-71"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyglyph'>=</span>
+<a name="line-72"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-73"></a>
+<a name="line-74"></a><a name="instance%20Data.ProtoLens.HasField%20%22name%22%20CostGraphDef'Node%20CostGraphDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"name"</span> <span class='hs-conid'>CostGraphDef'Node</span>
+<a name="line-75"></a>         <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-76"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-77"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_CostGraphDef'Node'name</span>
+<a name="line-78"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'name</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-79"></a>
+<a name="line-80"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"device"</span> <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyglyph'>=</span>
+<a name="line-81"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-82"></a>
+<a name="line-83"></a><a name="instance%20Data.ProtoLens.HasField%20%22device%22%20CostGraphDef'Node%20CostGraphDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"device"</span> <span class='hs-conid'>CostGraphDef'Node</span>
+<a name="line-84"></a>         <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-85"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-86"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_CostGraphDef'Node'device</span>
+<a name="line-87"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'device</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-88"></a>
+<a name="line-89"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"id"</span> <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyglyph'>=</span>
+<a name="line-90"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-91"></a>
+<a name="line-92"></a><a name="instance%20Data.ProtoLens.HasField%20%22id%22%20CostGraphDef'Node%20CostGraphDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"id"</span> <span class='hs-conid'>CostGraphDef'Node</span>
+<a name="line-93"></a>         <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-94"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-95"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_CostGraphDef'Node'id</span>
+<a name="line-96"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'id</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-97"></a>
+<a name="line-98"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"inputInfo"</span> <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyglyph'>=</span>
+<a name="line-99"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>CostGraphDef'Node'InputInfo</span><span class='hs-keyglyph'>]</span>
+<a name="line-100"></a>
+<a name="line-101"></a><a name="instance%20Data.ProtoLens.HasField%20%22inputInfo%22%20CostGraphDef'Node%20CostGraphDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"inputInfo"</span> <span class='hs-conid'>CostGraphDef'Node</span>
+<a name="line-102"></a>         <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-103"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-104"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_CostGraphDef'Node'inputInfo</span>
+<a name="line-105"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'inputInfo</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-106"></a>
+<a name="line-107"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"outputInfo"</span> <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyglyph'>=</span>
+<a name="line-108"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>CostGraphDef'Node'OutputInfo</span><span class='hs-keyglyph'>]</span>
+<a name="line-109"></a>
+<a name="line-110"></a><a name="instance%20Data.ProtoLens.HasField%20%22outputInfo%22%20CostGraphDef'Node%20CostGraphDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"outputInfo"</span> <span class='hs-conid'>CostGraphDef'Node</span>
+<a name="line-111"></a>         <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-112"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-113"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_CostGraphDef'Node'outputInfo</span>
+<a name="line-114"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'outputInfo</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-115"></a>
+<a name="line-116"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-117"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"temporaryMemorySize"</span> <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyglyph'>=</span>
+<a name="line-118"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-119"></a>
+<a name="line-120"></a><a name="instance%20Data.ProtoLens.HasField%20%22temporaryMemorySize%22%20CostGraphDef'Node%20CostGraphDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"temporaryMemorySize"</span>
+<a name="line-121"></a>         <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-122"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-123"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span>
+<a name="line-124"></a>              <span class='hs-sel'>_CostGraphDef'Node'temporaryMemorySize</span>
+<a name="line-125"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'temporaryMemorySize</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-126"></a>
+<a name="line-127"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"computeCost"</span> <span class='hs-conid'>CostGraphDef'Node</span>
+<a name="line-128"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-129"></a>
+<a name="line-130"></a><a name="instance%20Data.ProtoLens.HasField%20%22computeCost%22%20CostGraphDef'Node%20CostGraphDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"computeCost"</span> <span class='hs-conid'>CostGraphDef'Node</span>
+<a name="line-131"></a>         <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-132"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-133"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_CostGraphDef'Node'computeCost</span>
+<a name="line-134"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'computeCost</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-135"></a>
+<a name="line-136"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"isFinal"</span> <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyglyph'>=</span>
+<a name="line-137"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-138"></a>
+<a name="line-139"></a><a name="instance%20Data.ProtoLens.HasField%20%22isFinal%22%20CostGraphDef'Node%20CostGraphDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"isFinal"</span> <span class='hs-conid'>CostGraphDef'Node</span>
+<a name="line-140"></a>         <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-141"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-142"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_CostGraphDef'Node'isFinal</span>
+<a name="line-143"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'isFinal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-144"></a>
+<a name="line-145"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"controlInput"</span> <span class='hs-conid'>CostGraphDef'Node</span>
+<a name="line-146"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-keyglyph'>]</span>
+<a name="line-147"></a>
+<a name="line-148"></a><a name="instance%20Data.ProtoLens.HasField%20%22controlInput%22%20CostGraphDef'Node%20CostGraphDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"controlInput"</span> <span class='hs-conid'>CostGraphDef'Node</span>
+<a name="line-149"></a>         <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-150"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-151"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_CostGraphDef'Node'controlInput</span>
+<a name="line-152"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'controlInput</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-153"></a>
+<a name="line-154"></a><a name="instance%20Data.Default.Class.Default%20CostGraphDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-155"></a>        <span class='hs-varid'>def</span>
+<a name="line-156"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CostGraphDef'Node</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'name</span> <span class='hs-keyglyph'>=</span>
+<a name="line-157"></a>                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-158"></a>                              <span class='hs-sel'>_CostGraphDef'Node'device</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-159"></a>                              <span class='hs-sel'>_CostGraphDef'Node'id</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-160"></a>                              <span class='hs-sel'>_CostGraphDef'Node'inputInfo</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-161"></a>                              <span class='hs-sel'>_CostGraphDef'Node'outputInfo</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-162"></a>                              <span class='hs-sel'>_CostGraphDef'Node'temporaryMemorySize</span> <span class='hs-keyglyph'>=</span>
+<a name="line-163"></a>                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-164"></a>                              <span class='hs-sel'>_CostGraphDef'Node'computeCost</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-165"></a>                              <span class='hs-sel'>_CostGraphDef'Node'isFinal</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-166"></a>                              <span class='hs-sel'>_CostGraphDef'Node'controlInput</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>}</span>
+<a name="line-167"></a>
+<a name="line-168"></a><a name="instance%20Data.ProtoLens.Message%20CostGraphDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>CostGraphDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-169"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-170"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>name__field_descriptor</span>
+<a name="line-171"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"name"</span>
+<a name="line-172"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-173"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-174"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span>
+<a name="line-175"></a>                <span class='hs-varid'>device__field_descriptor</span>
+<a name="line-176"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"device"</span>
+<a name="line-177"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-178"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-179"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>device</span><span class='hs-layout'>)</span>
+<a name="line-180"></a>                <span class='hs-varid'>id__field_descriptor</span>
+<a name="line-181"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"id"</span>
+<a name="line-182"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-183"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-184"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>id</span><span class='hs-layout'>)</span>
+<a name="line-185"></a>                <span class='hs-varid'>inputInfo__field_descriptor</span>
+<a name="line-186"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"input_info"</span>
+<a name="line-187"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-188"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>CostGraphDef'Node'InputInfo</span><span class='hs-layout'>)</span>
+<a name="line-189"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>inputInfo</span><span class='hs-layout'>)</span>
+<a name="line-190"></a>                <span class='hs-varid'>outputInfo__field_descriptor</span>
+<a name="line-191"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"output_info"</span>
+<a name="line-192"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-193"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span><span class='hs-layout'>)</span>
+<a name="line-194"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>outputInfo</span><span class='hs-layout'>)</span>
+<a name="line-195"></a>                <span class='hs-varid'>temporaryMemorySize__field_descriptor</span>
+<a name="line-196"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"temporary_memory_size"</span>
+<a name="line-197"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-198"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-199"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-200"></a>                         <span class='hs-varid'>temporaryMemorySize</span><span class='hs-layout'>)</span>
+<a name="line-201"></a>                <span class='hs-varid'>computeCost__field_descriptor</span>
+<a name="line-202"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"compute_cost"</span>
+<a name="line-203"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-204"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-205"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>computeCost</span><span class='hs-layout'>)</span>
+<a name="line-206"></a>                <span class='hs-varid'>isFinal__field_descriptor</span>
+<a name="line-207"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"is_final"</span>
+<a name="line-208"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-209"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-210"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>isFinal</span><span class='hs-layout'>)</span>
+<a name="line-211"></a>                <span class='hs-varid'>controlInput__field_descriptor</span>
+<a name="line-212"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"control_input"</span>
+<a name="line-213"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-214"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-215"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>controlInput</span><span class='hs-layout'>)</span>
+<a name="line-216"></a>              <span class='hs-keyword'>in</span>
+<a name="line-217"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-218"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-219"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>name__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-220"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>device__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-221"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>id__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-222"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>inputInfo__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-223"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>5</span><span class='hs-layout'>,</span> <span class='hs-varid'>outputInfo__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-224"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>6</span><span class='hs-layout'>,</span> <span class='hs-varid'>temporaryMemorySize__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-225"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>9</span><span class='hs-layout'>,</span> <span class='hs-varid'>computeCost__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-226"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>7</span><span class='hs-layout'>,</span> <span class='hs-varid'>isFinal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-227"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>8</span><span class='hs-layout'>,</span> <span class='hs-varid'>controlInput__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-228"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-229"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"name"</span><span class='hs-layout'>,</span> <span class='hs-varid'>name__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-230"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"device"</span><span class='hs-layout'>,</span> <span class='hs-varid'>device__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-str'>"id"</span><span class='hs-layout'>,</span> <span class='hs-varid'>id__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-231"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"input_info"</span><span class='hs-layout'>,</span> <span class='hs-varid'>inputInfo__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-232"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"output_info"</span><span class='hs-layout'>,</span> <span class='hs-varid'>outputInfo__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-233"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"temporary_memory_size"</span><span class='hs-layout'>,</span> <span class='hs-varid'>temporaryMemorySize__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-234"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"compute_cost"</span><span class='hs-layout'>,</span> <span class='hs-varid'>computeCost__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-235"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"is_final"</span><span class='hs-layout'>,</span> <span class='hs-varid'>isFinal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-236"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"control_input"</span><span class='hs-layout'>,</span> <span class='hs-varid'>controlInput__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-237"></a>
+<a name="line-238"></a><a name="CostGraphDef'Node'InputInfo"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>CostGraphDef'Node'InputInfo</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CostGraphDef'Node'InputInfo</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'InputInfo'precedingNode</span>
+<a name="line-239"></a>                                                               <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-240"></a>                                                               <span class='hs-sel'>_CostGraphDef'Node'InputInfo'precedingPort</span>
+<a name="line-241"></a>                                                               <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>}</span>
+<a name="line-242"></a>                                 <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-243"></a>
+<a name="line-244"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-245"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"precedingNode"</span> <span class='hs-conid'>CostGraphDef'Node'InputInfo</span> <span class='hs-keyglyph'>=</span>
+<a name="line-246"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-247"></a>
+<a name="line-248"></a><a name="instance%20Data.ProtoLens.HasField%20%22precedingNode%22%20CostGraphDef'Node'InputInfo%20CostGraphDef'Node'InputInfo"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"precedingNode"</span>
+<a name="line-249"></a>         <span class='hs-conid'>CostGraphDef'Node'InputInfo</span> <span class='hs-conid'>CostGraphDef'Node'InputInfo</span> <span class='hs-keyword'>where</span>
+<a name="line-250"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-251"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span>
+<a name="line-252"></a>              <span class='hs-sel'>_CostGraphDef'Node'InputInfo'precedingNode</span>
+<a name="line-253"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-254"></a>                 <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'InputInfo'precedingNode</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-255"></a>
+<a name="line-256"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-257"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"precedingPort"</span> <span class='hs-conid'>CostGraphDef'Node'InputInfo</span> <span class='hs-keyglyph'>=</span>
+<a name="line-258"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-259"></a>
+<a name="line-260"></a><a name="instance%20Data.ProtoLens.HasField%20%22precedingPort%22%20CostGraphDef'Node'InputInfo%20CostGraphDef'Node'InputInfo"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"precedingPort"</span>
+<a name="line-261"></a>         <span class='hs-conid'>CostGraphDef'Node'InputInfo</span> <span class='hs-conid'>CostGraphDef'Node'InputInfo</span> <span class='hs-keyword'>where</span>
+<a name="line-262"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-263"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span>
+<a name="line-264"></a>              <span class='hs-sel'>_CostGraphDef'Node'InputInfo'precedingPort</span>
+<a name="line-265"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-266"></a>                 <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'InputInfo'precedingPort</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-267"></a>
+<a name="line-268"></a><a name="instance%20Data.Default.Class.Default%20CostGraphDef'Node'InputInfo"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>CostGraphDef'Node'InputInfo</span>
+<a name="line-269"></a>         <span class='hs-keyword'>where</span>
+<a name="line-270"></a>        <span class='hs-varid'>def</span>
+<a name="line-271"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CostGraphDef'Node'InputInfo</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'InputInfo'precedingNode</span>
+<a name="line-272"></a>                                          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-273"></a>                                        <span class='hs-sel'>_CostGraphDef'Node'InputInfo'precedingPort</span> <span class='hs-keyglyph'>=</span>
+<a name="line-274"></a>                                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-275"></a>
+<a name="line-276"></a><a name="instance%20Data.ProtoLens.Message%20CostGraphDef'Node'InputInfo"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>CostGraphDef'Node'InputInfo</span> <span class='hs-keyword'>where</span>
+<a name="line-277"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-278"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>precedingNode__field_descriptor</span>
+<a name="line-279"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"preceding_node"</span>
+<a name="line-280"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-281"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-282"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>precedingNode</span><span class='hs-layout'>)</span>
+<a name="line-283"></a>                <span class='hs-varid'>precedingPort__field_descriptor</span>
+<a name="line-284"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"preceding_port"</span>
+<a name="line-285"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-286"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-287"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>precedingPort</span><span class='hs-layout'>)</span>
+<a name="line-288"></a>              <span class='hs-keyword'>in</span>
+<a name="line-289"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-290"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-291"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>precedingNode__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-292"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>precedingPort__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-293"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-294"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"preceding_node"</span><span class='hs-layout'>,</span> <span class='hs-varid'>precedingNode__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-295"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"preceding_port"</span><span class='hs-layout'>,</span> <span class='hs-varid'>precedingPort__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-296"></a>
+<a name="line-297"></a><a name="CostGraphDef'Node'OutputInfo"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'OutputInfo'size</span>
+<a name="line-298"></a>                                                                 <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-299"></a>                                                                 <span class='hs-sel'>_CostGraphDef'Node'OutputInfo'aliasInputPort</span>
+<a name="line-300"></a>                                                                 <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-301"></a>                                                                 <span class='hs-sel'>_CostGraphDef'Node'OutputInfo'shape</span>
+<a name="line-302"></a>                                                                 <span class='hs-keyglyph'>::</span>
+<a name="line-303"></a>                                                                 <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-304"></a>                                                                   <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span><span class='hs-layout'>,</span>
+<a name="line-305"></a>                                                                 <span class='hs-sel'>_CostGraphDef'Node'OutputInfo'dtype</span>
+<a name="line-306"></a>                                                                 <span class='hs-keyglyph'>::</span>
+<a name="line-307"></a>                                                                 <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span><span class='hs-layout'>}</span>
+<a name="line-308"></a>                                  <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-309"></a>
+<a name="line-310"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-311"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"size"</span> <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span> <span class='hs-keyglyph'>=</span>
+<a name="line-312"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-313"></a>
+<a name="line-314"></a><a name="instance%20Data.ProtoLens.HasField%20%22size%22%20CostGraphDef'Node'OutputInfo%20CostGraphDef'Node'OutputInfo"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"size"</span>
+<a name="line-315"></a>         <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span> <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span> <span class='hs-keyword'>where</span>
+<a name="line-316"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-317"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_CostGraphDef'Node'OutputInfo'size</span>
+<a name="line-318"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'OutputInfo'size</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-319"></a>
+<a name="line-320"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-321"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"aliasInputPort"</span> <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span>
+<a name="line-322"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-323"></a>
+<a name="line-324"></a><a name="instance%20Data.ProtoLens.HasField%20%22aliasInputPort%22%20CostGraphDef'Node'OutputInfo%20CostGraphDef'Node'OutputInfo"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"aliasInputPort"</span>
+<a name="line-325"></a>         <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span> <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span> <span class='hs-keyword'>where</span>
+<a name="line-326"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-327"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span>
+<a name="line-328"></a>              <span class='hs-sel'>_CostGraphDef'Node'OutputInfo'aliasInputPort</span>
+<a name="line-329"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-330"></a>                 <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'OutputInfo'aliasInputPort</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-331"></a>
+<a name="line-332"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-333"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"shape"</span> <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span> <span class='hs-keyglyph'>=</span>
+<a name="line-334"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span>
+<a name="line-335"></a>
+<a name="line-336"></a><a name="instance%20Data.ProtoLens.HasField%20%22shape%22%20CostGraphDef'Node'OutputInfo%20CostGraphDef'Node'OutputInfo"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"shape"</span>
+<a name="line-337"></a>         <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span> <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span> <span class='hs-keyword'>where</span>
+<a name="line-338"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-339"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'shape</span>
+<a name="line-340"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-341"></a>
+<a name="line-342"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-343"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'shape"</span> <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span> <span class='hs-keyglyph'>=</span>
+<a name="line-344"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-345"></a>       <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span>
+<a name="line-346"></a>
+<a name="line-347"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'shape%22%20CostGraphDef'Node'OutputInfo%20CostGraphDef'Node'OutputInfo"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'shape"</span>
+<a name="line-348"></a>         <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span> <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span> <span class='hs-keyword'>where</span>
+<a name="line-349"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-350"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_CostGraphDef'Node'OutputInfo'shape</span>
+<a name="line-351"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'OutputInfo'shape</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-352"></a>
+<a name="line-353"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-354"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dtype"</span> <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span> <span class='hs-keyglyph'>=</span>
+<a name="line-355"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span>
+<a name="line-356"></a>
+<a name="line-357"></a><a name="instance%20Data.ProtoLens.HasField%20%22dtype%22%20CostGraphDef'Node'OutputInfo%20CostGraphDef'Node'OutputInfo"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"dtype"</span>
+<a name="line-358"></a>         <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span> <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span> <span class='hs-keyword'>where</span>
+<a name="line-359"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-360"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_CostGraphDef'Node'OutputInfo'dtype</span>
+<a name="line-361"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'OutputInfo'dtype</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-362"></a>
+<a name="line-363"></a><a name="instance%20Data.Default.Class.Default%20CostGraphDef'Node'OutputInfo"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span>
+<a name="line-364"></a>         <span class='hs-keyword'>where</span>
+<a name="line-365"></a>        <span class='hs-varid'>def</span>
+<a name="line-366"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span><span class='hs-layout'>{</span><span class='hs-sel'>_CostGraphDef'Node'OutputInfo'size</span> <span class='hs-keyglyph'>=</span>
+<a name="line-367"></a>                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-368"></a>                                         <span class='hs-sel'>_CostGraphDef'Node'OutputInfo'aliasInputPort</span> <span class='hs-keyglyph'>=</span>
+<a name="line-369"></a>                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-370"></a>                                         <span class='hs-sel'>_CostGraphDef'Node'OutputInfo'shape</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-371"></a>                                         <span class='hs-sel'>_CostGraphDef'Node'OutputInfo'dtype</span> <span class='hs-keyglyph'>=</span>
+<a name="line-372"></a>                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>}</span>
+<a name="line-373"></a>
+<a name="line-374"></a><a name="instance%20Data.ProtoLens.Message%20CostGraphDef'Node'OutputInfo"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>CostGraphDef'Node'OutputInfo</span> <span class='hs-keyword'>where</span>
+<a name="line-375"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-376"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>size__field_descriptor</span>
+<a name="line-377"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"size"</span>
+<a name="line-378"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-379"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-380"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>size</span><span class='hs-layout'>)</span>
+<a name="line-381"></a>                <span class='hs-varid'>aliasInputPort__field_descriptor</span>
+<a name="line-382"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"alias_input_port"</span>
+<a name="line-383"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-384"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-385"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>aliasInputPort</span><span class='hs-layout'>)</span>
+<a name="line-386"></a>                <span class='hs-varid'>shape__field_descriptor</span>
+<a name="line-387"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"shape"</span>
+<a name="line-388"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-389"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-390"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span><span class='hs-layout'>)</span>
+<a name="line-391"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'shape</span><span class='hs-layout'>)</span>
+<a name="line-392"></a>                <span class='hs-varid'>dtype__field_descriptor</span>
+<a name="line-393"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"dtype"</span>
+<a name="line-394"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>EnumField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-395"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-396"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span><span class='hs-layout'>)</span>
+<a name="line-397"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span>
+<a name="line-398"></a>              <span class='hs-keyword'>in</span>
+<a name="line-399"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-400"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-401"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>size__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-402"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>aliasInputPort__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-403"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>shape__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-404"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>dtype__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-405"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-406"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"size"</span><span class='hs-layout'>,</span> <span class='hs-varid'>size__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-407"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"alias_input_port"</span><span class='hs-layout'>,</span> <span class='hs-varid'>aliasInputPort__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-408"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"shape"</span><span class='hs-layout'>,</span> <span class='hs-varid'>shape__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-409"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"dtype"</span><span class='hs-layout'>,</span> <span class='hs-varid'>dtype__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-410"></a>
+<a name="line-411"></a><a name="aliasInputPort"></a><span class='hs-definition'>aliasInputPort</span> <span class='hs-keyglyph'>::</span>
+<a name="line-412"></a>               <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-413"></a>                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"aliasInputPort"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-414"></a>                 <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-415"></a>                   <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"aliasInputPort"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-416"></a>                   <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"aliasInputPort"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-417"></a><span class='hs-definition'>aliasInputPort</span>
+<a name="line-418"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-419"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-420"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"aliasInputPort"</span><span class='hs-layout'>)</span>
+<a name="line-421"></a>
+<a name="line-422"></a><a name="computeCost"></a><span class='hs-definition'>computeCost</span> <span class='hs-keyglyph'>::</span>
+<a name="line-423"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"computeCost"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-424"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"computeCost"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-425"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"computeCost"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-426"></a><span class='hs-definition'>computeCost</span>
+<a name="line-427"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-428"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"computeCost"</span><span class='hs-layout'>)</span>
+<a name="line-429"></a>
+<a name="line-430"></a><a name="controlInput"></a><span class='hs-definition'>controlInput</span> <span class='hs-keyglyph'>::</span>
+<a name="line-431"></a>             <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-432"></a>               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"controlInput"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-433"></a>               <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-434"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"controlInput"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-435"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"controlInput"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-436"></a><span class='hs-definition'>controlInput</span>
+<a name="line-437"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-438"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"controlInput"</span><span class='hs-layout'>)</span>
+<a name="line-439"></a>
+<a name="line-440"></a><a name="device"></a><span class='hs-definition'>device</span> <span class='hs-keyglyph'>::</span>
+<a name="line-441"></a>       <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"device"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-442"></a>         <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"device"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-443"></a>           <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"device"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-444"></a><span class='hs-definition'>device</span>
+<a name="line-445"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-446"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"device"</span><span class='hs-layout'>)</span>
+<a name="line-447"></a>
+<a name="line-448"></a><a name="dtype"></a><span class='hs-definition'>dtype</span> <span class='hs-keyglyph'>::</span>
+<a name="line-449"></a>      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"dtype"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-450"></a>        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dtype"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-451"></a>          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dtype"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-452"></a><span class='hs-definition'>dtype</span>
+<a name="line-453"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-454"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"dtype"</span><span class='hs-layout'>)</span>
+<a name="line-455"></a>
+<a name="line-456"></a><a name="id"></a><span class='hs-definition'>id</span> <span class='hs-keyglyph'>::</span>
+<a name="line-457"></a>   <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"id"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-458"></a>     <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"id"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-459"></a>       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"id"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-460"></a><span class='hs-definition'>id</span>
+<a name="line-461"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-462"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"id"</span><span class='hs-layout'>)</span>
+<a name="line-463"></a>
+<a name="line-464"></a><a name="inputInfo"></a><span class='hs-definition'>inputInfo</span> <span class='hs-keyglyph'>::</span>
+<a name="line-465"></a>          <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"inputInfo"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-466"></a>            <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"inputInfo"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-467"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"inputInfo"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-468"></a><span class='hs-definition'>inputInfo</span>
+<a name="line-469"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-470"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"inputInfo"</span><span class='hs-layout'>)</span>
+<a name="line-471"></a>
+<a name="line-472"></a><a name="isFinal"></a><span class='hs-definition'>isFinal</span> <span class='hs-keyglyph'>::</span>
+<a name="line-473"></a>        <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"isFinal"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-474"></a>          <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"isFinal"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-475"></a>            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"isFinal"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-476"></a><span class='hs-definition'>isFinal</span>
+<a name="line-477"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-478"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"isFinal"</span><span class='hs-layout'>)</span>
+<a name="line-479"></a>
+<a name="line-480"></a><a name="maybe'shape"></a><span class='hs-definition'>maybe'shape</span> <span class='hs-keyglyph'>::</span>
+<a name="line-481"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'shape"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-482"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'shape"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-483"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'shape"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-484"></a><span class='hs-definition'>maybe'shape</span>
+<a name="line-485"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-486"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'shape"</span><span class='hs-layout'>)</span>
+<a name="line-487"></a>
+<a name="line-488"></a><a name="name"></a><span class='hs-definition'>name</span> <span class='hs-keyglyph'>::</span>
+<a name="line-489"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-490"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-491"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-492"></a><span class='hs-definition'>name</span>
+<a name="line-493"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-494"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"name"</span><span class='hs-layout'>)</span>
+<a name="line-495"></a>
+<a name="line-496"></a><a name="node"></a><span class='hs-definition'>node</span> <span class='hs-keyglyph'>::</span>
+<a name="line-497"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"node"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-498"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"node"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-499"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"node"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-500"></a><span class='hs-definition'>node</span>
+<a name="line-501"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-502"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"node"</span><span class='hs-layout'>)</span>
+<a name="line-503"></a>
+<a name="line-504"></a><a name="outputInfo"></a><span class='hs-definition'>outputInfo</span> <span class='hs-keyglyph'>::</span>
+<a name="line-505"></a>           <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"outputInfo"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-506"></a>             <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"outputInfo"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-507"></a>               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"outputInfo"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-508"></a><span class='hs-definition'>outputInfo</span>
+<a name="line-509"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-510"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"outputInfo"</span><span class='hs-layout'>)</span>
+<a name="line-511"></a>
+<a name="line-512"></a><a name="precedingNode"></a><span class='hs-definition'>precedingNode</span> <span class='hs-keyglyph'>::</span>
+<a name="line-513"></a>              <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-514"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"precedingNode"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-515"></a>                <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-516"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"precedingNode"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-517"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"precedingNode"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-518"></a><span class='hs-definition'>precedingNode</span>
+<a name="line-519"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-520"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-521"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"precedingNode"</span><span class='hs-layout'>)</span>
+<a name="line-522"></a>
+<a name="line-523"></a><a name="precedingPort"></a><span class='hs-definition'>precedingPort</span> <span class='hs-keyglyph'>::</span>
+<a name="line-524"></a>              <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-525"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"precedingPort"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-526"></a>                <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-527"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"precedingPort"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-528"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"precedingPort"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-529"></a><span class='hs-definition'>precedingPort</span>
+<a name="line-530"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-531"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-532"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"precedingPort"</span><span class='hs-layout'>)</span>
+<a name="line-533"></a>
+<a name="line-534"></a><a name="shape"></a><span class='hs-definition'>shape</span> <span class='hs-keyglyph'>::</span>
+<a name="line-535"></a>      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"shape"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-536"></a>        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"shape"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-537"></a>          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"shape"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-538"></a><span class='hs-definition'>shape</span>
+<a name="line-539"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-540"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"shape"</span><span class='hs-layout'>)</span>
+<a name="line-541"></a>
+<a name="line-542"></a><a name="size"></a><span class='hs-definition'>size</span> <span class='hs-keyglyph'>::</span>
+<a name="line-543"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"size"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-544"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"size"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-545"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"size"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-546"></a><span class='hs-definition'>size</span>
+<a name="line-547"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-548"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"size"</span><span class='hs-layout'>)</span>
+<a name="line-549"></a>
+<a name="line-550"></a><a name="temporaryMemorySize"></a><span class='hs-definition'>temporaryMemorySize</span> <span class='hs-keyglyph'>::</span>
+<a name="line-551"></a>                    <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-552"></a>                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"temporaryMemorySize"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-553"></a>                      <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-554"></a>                        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"temporaryMemorySize"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-555"></a>                        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"temporaryMemorySize"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-556"></a><span class='hs-definition'>temporaryMemorySize</span>
+<a name="line-557"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-558"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-559"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"temporaryMemorySize"</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Function.html b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Function.html
new file mode 100644
index 0000000..7a2dcbf
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Function.html
@@ -0,0 +1,586 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Proto/Tensorflow/Core/Framework/Function.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>{- This file was auto-generated from tensorflow/core/framework/function.proto by the proto-lens-protoc program. -}</span>
+<a name="line-2"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables, DataKinds, TypeFamilies,
+<a name="line-3"></a>  MultiParamTypeClasses, FlexibleContexts, FlexibleInstances,
+<a name="line-4"></a>  PatternSynonyms #-}</span>
+<a name="line-5"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-unused-imports#-}</span>
+<a name="line-6"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Function</span> <span class='hs-keyword'>where</span>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Prelude</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-11"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-12"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-13"></a>       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-14"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-15"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-16"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-17"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-18"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-19"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-20"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-24"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-25"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span>
+<a name="line-26"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span>
+<a name="line-27"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>OpDef</span>
+<a name="line-28"></a>
+<a name="line-29"></a><a name="FunctionDef"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>FunctionDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>FunctionDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'signature</span> <span class='hs-keyglyph'>::</span>
+<a name="line-30"></a>                               <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>OpDef</span><span class='hs-varop'>.</span><span class='hs-conid'>OpDef</span><span class='hs-layout'>,</span>
+<a name="line-31"></a>                               <span class='hs-sel'>_FunctionDef'node</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>FunctionDef'Node</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-32"></a>                               <span class='hs-sel'>_FunctionDef'nodeDef</span> <span class='hs-keyglyph'>::</span>
+<a name="line-33"></a>                               <span class='hs-keyglyph'>[</span><span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-34"></a>                               <span class='hs-sel'>_FunctionDef'ret</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>}</span>
+<a name="line-35"></a>                 <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-36"></a>
+<a name="line-37"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"signature"</span> <span class='hs-conid'>FunctionDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-38"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>OpDef</span><span class='hs-varop'>.</span><span class='hs-conid'>OpDef</span>
+<a name="line-39"></a>
+<a name="line-40"></a><a name="instance%20Data.ProtoLens.HasField%20%22signature%22%20FunctionDef%20FunctionDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"signature"</span> <span class='hs-conid'>FunctionDef</span>
+<a name="line-41"></a>         <span class='hs-conid'>FunctionDef</span> <span class='hs-keyword'>where</span>
+<a name="line-42"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-43"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'signature</span>
+<a name="line-44"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-45"></a>
+<a name="line-46"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'signature"</span> <span class='hs-conid'>FunctionDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-47"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>OpDef</span><span class='hs-varop'>.</span><span class='hs-conid'>OpDef</span>
+<a name="line-48"></a>
+<a name="line-49"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'signature%22%20FunctionDef%20FunctionDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'signature"</span> <span class='hs-conid'>FunctionDef</span>
+<a name="line-50"></a>         <span class='hs-conid'>FunctionDef</span> <span class='hs-keyword'>where</span>
+<a name="line-51"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-52"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_FunctionDef'signature</span>
+<a name="line-53"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'signature</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-54"></a>
+<a name="line-55"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"node"</span> <span class='hs-conid'>FunctionDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-56"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>FunctionDef'Node</span><span class='hs-keyglyph'>]</span>
+<a name="line-57"></a>
+<a name="line-58"></a><a name="instance%20Data.ProtoLens.HasField%20%22node%22%20FunctionDef%20FunctionDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"node"</span> <span class='hs-conid'>FunctionDef</span> <span class='hs-conid'>FunctionDef</span>
+<a name="line-59"></a>         <span class='hs-keyword'>where</span>
+<a name="line-60"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-61"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_FunctionDef'node</span>
+<a name="line-62"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'node</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-63"></a>
+<a name="line-64"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"nodeDef"</span> <span class='hs-conid'>FunctionDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-65"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span><span class='hs-keyglyph'>]</span>
+<a name="line-66"></a>
+<a name="line-67"></a><a name="instance%20Data.ProtoLens.HasField%20%22nodeDef%22%20FunctionDef%20FunctionDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"nodeDef"</span> <span class='hs-conid'>FunctionDef</span> <span class='hs-conid'>FunctionDef</span>
+<a name="line-68"></a>         <span class='hs-keyword'>where</span>
+<a name="line-69"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-70"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_FunctionDef'nodeDef</span>
+<a name="line-71"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'nodeDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-72"></a>
+<a name="line-73"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"ret"</span> <span class='hs-conid'>FunctionDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-74"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-75"></a>
+<a name="line-76"></a><a name="instance%20Data.ProtoLens.HasField%20%22ret%22%20FunctionDef%20FunctionDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"ret"</span> <span class='hs-conid'>FunctionDef</span> <span class='hs-conid'>FunctionDef</span>
+<a name="line-77"></a>         <span class='hs-keyword'>where</span>
+<a name="line-78"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-79"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_FunctionDef'ret</span>
+<a name="line-80"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'ret</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-81"></a>
+<a name="line-82"></a><a name="instance%20Data.Default.Class.Default%20FunctionDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>FunctionDef</span> <span class='hs-keyword'>where</span>
+<a name="line-83"></a>        <span class='hs-varid'>def</span>
+<a name="line-84"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>FunctionDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'signature</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-85"></a>                        <span class='hs-sel'>_FunctionDef'node</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span> <span class='hs-sel'>_FunctionDef'nodeDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-86"></a>                        <span class='hs-sel'>_FunctionDef'ret</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>empty</span><span class='hs-layout'>}</span>
+<a name="line-87"></a>
+<a name="line-88"></a><a name="instance%20Data.ProtoLens.Message%20FunctionDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>FunctionDef</span> <span class='hs-keyword'>where</span>
+<a name="line-89"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-90"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>signature__field_descriptor</span>
+<a name="line-91"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"signature"</span>
+<a name="line-92"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-93"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-94"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>OpDef</span><span class='hs-varop'>.</span><span class='hs-conid'>OpDef</span><span class='hs-layout'>)</span>
+<a name="line-95"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'signature</span><span class='hs-layout'>)</span>
+<a name="line-96"></a>                <span class='hs-varid'>node__field_descriptor</span>
+<a name="line-97"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"node"</span>
+<a name="line-98"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-99"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>FunctionDef'Node</span><span class='hs-layout'>)</span>
+<a name="line-100"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>node</span><span class='hs-layout'>)</span>
+<a name="line-101"></a>                <span class='hs-varid'>nodeDef__field_descriptor</span>
+<a name="line-102"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"node_def"</span>
+<a name="line-103"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-104"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-105"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span><span class='hs-layout'>)</span>
+<a name="line-106"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>nodeDef</span><span class='hs-layout'>)</span>
+<a name="line-107"></a>                <span class='hs-varid'>ret__field_descriptor</span>
+<a name="line-108"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"ret"</span>
+<a name="line-109"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-110"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>FunctionDef'RetEntry</span><span class='hs-layout'>)</span>
+<a name="line-111"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MapField</span> <span class='hs-varid'>key</span> <span class='hs-varid'>value</span> <span class='hs-varid'>ret</span><span class='hs-layout'>)</span>
+<a name="line-112"></a>              <span class='hs-keyword'>in</span>
+<a name="line-113"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-114"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-115"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>signature__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-116"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>node__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-117"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>nodeDef__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-118"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>ret__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-119"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-120"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"signature"</span><span class='hs-layout'>,</span> <span class='hs-varid'>signature__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-121"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"node"</span><span class='hs-layout'>,</span> <span class='hs-varid'>node__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-122"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"node_def"</span><span class='hs-layout'>,</span> <span class='hs-varid'>nodeDef__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-123"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"ret"</span><span class='hs-layout'>,</span> <span class='hs-varid'>ret__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-124"></a>
+<a name="line-125"></a><a name="FunctionDef'Node"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>FunctionDef'Node</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>FunctionDef'Node</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'Node'ret</span> <span class='hs-keyglyph'>::</span>
+<a name="line-126"></a>                                         <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-127"></a>                                         <span class='hs-sel'>_FunctionDef'Node'op</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-128"></a>                                         <span class='hs-sel'>_FunctionDef'Node'arg</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-129"></a>                                         <span class='hs-sel'>_FunctionDef'Node'dep</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-130"></a>                                         <span class='hs-sel'>_FunctionDef'Node'attr</span> <span class='hs-keyglyph'>::</span>
+<a name="line-131"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-132"></a>                                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-layout'>}</span>
+<a name="line-133"></a>                      <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-134"></a>
+<a name="line-135"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"ret"</span> <span class='hs-conid'>FunctionDef'Node</span> <span class='hs-keyglyph'>=</span>
+<a name="line-136"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-keyglyph'>]</span>
+<a name="line-137"></a>
+<a name="line-138"></a><a name="instance%20Data.ProtoLens.HasField%20%22ret%22%20FunctionDef'Node%20FunctionDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"ret"</span> <span class='hs-conid'>FunctionDef'Node</span>
+<a name="line-139"></a>         <span class='hs-conid'>FunctionDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-140"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-141"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_FunctionDef'Node'ret</span>
+<a name="line-142"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'Node'ret</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-143"></a>
+<a name="line-144"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"op"</span> <span class='hs-conid'>FunctionDef'Node</span> <span class='hs-keyglyph'>=</span>
+<a name="line-145"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-146"></a>
+<a name="line-147"></a><a name="instance%20Data.ProtoLens.HasField%20%22op%22%20FunctionDef'Node%20FunctionDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"op"</span> <span class='hs-conid'>FunctionDef'Node</span>
+<a name="line-148"></a>         <span class='hs-conid'>FunctionDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-149"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-150"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_FunctionDef'Node'op</span>
+<a name="line-151"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'Node'op</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-152"></a>
+<a name="line-153"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"arg"</span> <span class='hs-conid'>FunctionDef'Node</span> <span class='hs-keyglyph'>=</span>
+<a name="line-154"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-keyglyph'>]</span>
+<a name="line-155"></a>
+<a name="line-156"></a><a name="instance%20Data.ProtoLens.HasField%20%22arg%22%20FunctionDef'Node%20FunctionDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"arg"</span> <span class='hs-conid'>FunctionDef'Node</span>
+<a name="line-157"></a>         <span class='hs-conid'>FunctionDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-158"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-159"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_FunctionDef'Node'arg</span>
+<a name="line-160"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'Node'arg</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-161"></a>
+<a name="line-162"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dep"</span> <span class='hs-conid'>FunctionDef'Node</span> <span class='hs-keyglyph'>=</span>
+<a name="line-163"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-keyglyph'>]</span>
+<a name="line-164"></a>
+<a name="line-165"></a><a name="instance%20Data.ProtoLens.HasField%20%22dep%22%20FunctionDef'Node%20FunctionDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"dep"</span> <span class='hs-conid'>FunctionDef'Node</span>
+<a name="line-166"></a>         <span class='hs-conid'>FunctionDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-167"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-168"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_FunctionDef'Node'dep</span>
+<a name="line-169"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'Node'dep</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-170"></a>
+<a name="line-171"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"attr"</span> <span class='hs-conid'>FunctionDef'Node</span> <span class='hs-keyglyph'>=</span>
+<a name="line-172"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-173"></a>       <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span>
+<a name="line-174"></a>
+<a name="line-175"></a><a name="instance%20Data.ProtoLens.HasField%20%22attr%22%20FunctionDef'Node%20FunctionDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"attr"</span> <span class='hs-conid'>FunctionDef'Node</span>
+<a name="line-176"></a>         <span class='hs-conid'>FunctionDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-177"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-178"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_FunctionDef'Node'attr</span>
+<a name="line-179"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'Node'attr</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-180"></a>
+<a name="line-181"></a><a name="instance%20Data.Default.Class.Default%20FunctionDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>FunctionDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-182"></a>        <span class='hs-varid'>def</span>
+<a name="line-183"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>FunctionDef'Node</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'Node'ret</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-184"></a>                             <span class='hs-sel'>_FunctionDef'Node'op</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-185"></a>                             <span class='hs-sel'>_FunctionDef'Node'arg</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span> <span class='hs-sel'>_FunctionDef'Node'dep</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-186"></a>                             <span class='hs-sel'>_FunctionDef'Node'attr</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>empty</span><span class='hs-layout'>}</span>
+<a name="line-187"></a>
+<a name="line-188"></a><a name="instance%20Data.ProtoLens.Message%20FunctionDef'Node"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>FunctionDef'Node</span> <span class='hs-keyword'>where</span>
+<a name="line-189"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-190"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>ret__field_descriptor</span>
+<a name="line-191"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"ret"</span>
+<a name="line-192"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-193"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-194"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>ret</span><span class='hs-layout'>)</span>
+<a name="line-195"></a>                <span class='hs-varid'>op__field_descriptor</span>
+<a name="line-196"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"op"</span>
+<a name="line-197"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-198"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-199"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>op</span><span class='hs-layout'>)</span>
+<a name="line-200"></a>                <span class='hs-varid'>arg__field_descriptor</span>
+<a name="line-201"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"arg"</span>
+<a name="line-202"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-203"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-204"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>arg</span><span class='hs-layout'>)</span>
+<a name="line-205"></a>                <span class='hs-varid'>dep__field_descriptor</span>
+<a name="line-206"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"dep"</span>
+<a name="line-207"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-208"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-209"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>dep</span><span class='hs-layout'>)</span>
+<a name="line-210"></a>                <span class='hs-varid'>attr__field_descriptor</span>
+<a name="line-211"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"attr"</span>
+<a name="line-212"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-213"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>FunctionDef'Node'AttrEntry</span><span class='hs-layout'>)</span>
+<a name="line-214"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MapField</span> <span class='hs-varid'>key</span> <span class='hs-varid'>value</span> <span class='hs-varid'>attr</span><span class='hs-layout'>)</span>
+<a name="line-215"></a>              <span class='hs-keyword'>in</span>
+<a name="line-216"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-217"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-218"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>ret__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-219"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>op__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-220"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>arg__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-221"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>dep__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-222"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>5</span><span class='hs-layout'>,</span> <span class='hs-varid'>attr__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-223"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-224"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"ret"</span><span class='hs-layout'>,</span> <span class='hs-varid'>ret__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-str'>"op"</span><span class='hs-layout'>,</span> <span class='hs-varid'>op__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-225"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"arg"</span><span class='hs-layout'>,</span> <span class='hs-varid'>arg__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-str'>"dep"</span><span class='hs-layout'>,</span> <span class='hs-varid'>dep__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-226"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"attr"</span><span class='hs-layout'>,</span> <span class='hs-varid'>attr__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-227"></a>
+<a name="line-228"></a><a name="FunctionDef'Node'AttrEntry"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>FunctionDef'Node'AttrEntry</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>FunctionDef'Node'AttrEntry</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'Node'AttrEntry'key</span>
+<a name="line-229"></a>                                                             <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-230"></a>                                                             <span class='hs-sel'>_FunctionDef'Node'AttrEntry'value</span> <span class='hs-keyglyph'>::</span>
+<a name="line-231"></a>                                                             <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-232"></a>                                                               <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-layout'>}</span>
+<a name="line-233"></a>                                <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-234"></a>
+<a name="line-235"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"key"</span> <span class='hs-conid'>FunctionDef'Node'AttrEntry</span>
+<a name="line-236"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-237"></a>
+<a name="line-238"></a><a name="instance%20Data.ProtoLens.HasField%20%22key%22%20FunctionDef'Node'AttrEntry%20FunctionDef'Node'AttrEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"key"</span> <span class='hs-conid'>FunctionDef'Node'AttrEntry</span>
+<a name="line-239"></a>         <span class='hs-conid'>FunctionDef'Node'AttrEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-240"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-241"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_FunctionDef'Node'AttrEntry'key</span>
+<a name="line-242"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'Node'AttrEntry'key</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-243"></a>
+<a name="line-244"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-245"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"value"</span> <span class='hs-conid'>FunctionDef'Node'AttrEntry</span> <span class='hs-keyglyph'>=</span>
+<a name="line-246"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span>
+<a name="line-247"></a>
+<a name="line-248"></a><a name="instance%20Data.ProtoLens.HasField%20%22value%22%20FunctionDef'Node'AttrEntry%20FunctionDef'Node'AttrEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"value"</span> <span class='hs-conid'>FunctionDef'Node'AttrEntry</span>
+<a name="line-249"></a>         <span class='hs-conid'>FunctionDef'Node'AttrEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-250"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-251"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'value</span>
+<a name="line-252"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-253"></a>
+<a name="line-254"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-255"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'value"</span> <span class='hs-conid'>FunctionDef'Node'AttrEntry</span> <span class='hs-keyglyph'>=</span>
+<a name="line-256"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span>
+<a name="line-257"></a>
+<a name="line-258"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'value%22%20FunctionDef'Node'AttrEntry%20FunctionDef'Node'AttrEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'value"</span>
+<a name="line-259"></a>         <span class='hs-conid'>FunctionDef'Node'AttrEntry</span> <span class='hs-conid'>FunctionDef'Node'AttrEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-260"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-261"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_FunctionDef'Node'AttrEntry'value</span>
+<a name="line-262"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'Node'AttrEntry'value</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-263"></a>
+<a name="line-264"></a><a name="instance%20Data.Default.Class.Default%20FunctionDef'Node'AttrEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>FunctionDef'Node'AttrEntry</span>
+<a name="line-265"></a>         <span class='hs-keyword'>where</span>
+<a name="line-266"></a>        <span class='hs-varid'>def</span>
+<a name="line-267"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>FunctionDef'Node'AttrEntry</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'Node'AttrEntry'key</span> <span class='hs-keyglyph'>=</span>
+<a name="line-268"></a>                                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-269"></a>                                       <span class='hs-sel'>_FunctionDef'Node'AttrEntry'value</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>}</span>
+<a name="line-270"></a>
+<a name="line-271"></a><a name="instance%20Data.ProtoLens.Message%20FunctionDef'Node'AttrEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>FunctionDef'Node'AttrEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-272"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-273"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>key__field_descriptor</span>
+<a name="line-274"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"key"</span>
+<a name="line-275"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-276"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-277"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>key</span><span class='hs-layout'>)</span>
+<a name="line-278"></a>                <span class='hs-varid'>value__field_descriptor</span>
+<a name="line-279"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"value"</span>
+<a name="line-280"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-281"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-282"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-layout'>)</span>
+<a name="line-283"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'value</span><span class='hs-layout'>)</span>
+<a name="line-284"></a>              <span class='hs-keyword'>in</span>
+<a name="line-285"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-286"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-287"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>key__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-288"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>value__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-289"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-290"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"key"</span><span class='hs-layout'>,</span> <span class='hs-varid'>key__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-291"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"value"</span><span class='hs-layout'>,</span> <span class='hs-varid'>value__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-292"></a>
+<a name="line-293"></a><a name="FunctionDef'RetEntry"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>FunctionDef'RetEntry</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>FunctionDef'RetEntry</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'RetEntry'key</span>
+<a name="line-294"></a>                                                 <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-295"></a>                                                 <span class='hs-sel'>_FunctionDef'RetEntry'value</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>}</span>
+<a name="line-296"></a>                          <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-297"></a>
+<a name="line-298"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"key"</span> <span class='hs-conid'>FunctionDef'RetEntry</span> <span class='hs-keyglyph'>=</span>
+<a name="line-299"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-300"></a>
+<a name="line-301"></a><a name="instance%20Data.ProtoLens.HasField%20%22key%22%20FunctionDef'RetEntry%20FunctionDef'RetEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"key"</span> <span class='hs-conid'>FunctionDef'RetEntry</span>
+<a name="line-302"></a>         <span class='hs-conid'>FunctionDef'RetEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-303"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-304"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_FunctionDef'RetEntry'key</span>
+<a name="line-305"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'RetEntry'key</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-306"></a>
+<a name="line-307"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"value"</span> <span class='hs-conid'>FunctionDef'RetEntry</span> <span class='hs-keyglyph'>=</span>
+<a name="line-308"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-309"></a>
+<a name="line-310"></a><a name="instance%20Data.ProtoLens.HasField%20%22value%22%20FunctionDef'RetEntry%20FunctionDef'RetEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"value"</span> <span class='hs-conid'>FunctionDef'RetEntry</span>
+<a name="line-311"></a>         <span class='hs-conid'>FunctionDef'RetEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-312"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-313"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_FunctionDef'RetEntry'value</span>
+<a name="line-314"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'RetEntry'value</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-315"></a>
+<a name="line-316"></a><a name="instance%20Data.Default.Class.Default%20FunctionDef'RetEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>FunctionDef'RetEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-317"></a>        <span class='hs-varid'>def</span>
+<a name="line-318"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>FunctionDef'RetEntry</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDef'RetEntry'key</span> <span class='hs-keyglyph'>=</span>
+<a name="line-319"></a>                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-320"></a>                                 <span class='hs-sel'>_FunctionDef'RetEntry'value</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-321"></a>
+<a name="line-322"></a><a name="instance%20Data.ProtoLens.Message%20FunctionDef'RetEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>FunctionDef'RetEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-323"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-324"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>key__field_descriptor</span>
+<a name="line-325"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"key"</span>
+<a name="line-326"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-327"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-328"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>key</span><span class='hs-layout'>)</span>
+<a name="line-329"></a>                <span class='hs-varid'>value__field_descriptor</span>
+<a name="line-330"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"value"</span>
+<a name="line-331"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-332"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-333"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>value</span><span class='hs-layout'>)</span>
+<a name="line-334"></a>              <span class='hs-keyword'>in</span>
+<a name="line-335"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-336"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-337"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>key__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-338"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>value__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-339"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-340"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"key"</span><span class='hs-layout'>,</span> <span class='hs-varid'>key__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-341"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"value"</span><span class='hs-layout'>,</span> <span class='hs-varid'>value__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-342"></a>
+<a name="line-343"></a><a name="FunctionDefLibrary"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>FunctionDefLibrary</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>FunctionDefLibrary</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDefLibrary'function</span>
+<a name="line-344"></a>                                             <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>FunctionDef</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-345"></a>                                             <span class='hs-sel'>_FunctionDefLibrary'gradient</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>GradientDef</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>}</span>
+<a name="line-346"></a>                        <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-347"></a>
+<a name="line-348"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"function"</span> <span class='hs-conid'>FunctionDefLibrary</span> <span class='hs-keyglyph'>=</span>
+<a name="line-349"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>FunctionDef</span><span class='hs-keyglyph'>]</span>
+<a name="line-350"></a>
+<a name="line-351"></a><a name="instance%20Data.ProtoLens.HasField%20%22function%22%20FunctionDefLibrary%20FunctionDefLibrary"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"function"</span> <span class='hs-conid'>FunctionDefLibrary</span>
+<a name="line-352"></a>         <span class='hs-conid'>FunctionDefLibrary</span> <span class='hs-keyword'>where</span>
+<a name="line-353"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-354"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_FunctionDefLibrary'function</span>
+<a name="line-355"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDefLibrary'function</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-356"></a>
+<a name="line-357"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"gradient"</span> <span class='hs-conid'>FunctionDefLibrary</span> <span class='hs-keyglyph'>=</span>
+<a name="line-358"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>GradientDef</span><span class='hs-keyglyph'>]</span>
+<a name="line-359"></a>
+<a name="line-360"></a><a name="instance%20Data.ProtoLens.HasField%20%22gradient%22%20FunctionDefLibrary%20FunctionDefLibrary"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"gradient"</span> <span class='hs-conid'>FunctionDefLibrary</span>
+<a name="line-361"></a>         <span class='hs-conid'>FunctionDefLibrary</span> <span class='hs-keyword'>where</span>
+<a name="line-362"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-363"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_FunctionDefLibrary'gradient</span>
+<a name="line-364"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDefLibrary'gradient</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-365"></a>
+<a name="line-366"></a><a name="instance%20Data.Default.Class.Default%20FunctionDefLibrary"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>FunctionDefLibrary</span> <span class='hs-keyword'>where</span>
+<a name="line-367"></a>        <span class='hs-varid'>def</span>
+<a name="line-368"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>FunctionDefLibrary</span><span class='hs-layout'>{</span><span class='hs-sel'>_FunctionDefLibrary'function</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-369"></a>                               <span class='hs-sel'>_FunctionDefLibrary'gradient</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>}</span>
+<a name="line-370"></a>
+<a name="line-371"></a><a name="instance%20Data.ProtoLens.Message%20FunctionDefLibrary"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>FunctionDefLibrary</span> <span class='hs-keyword'>where</span>
+<a name="line-372"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-373"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>function__field_descriptor</span>
+<a name="line-374"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"function"</span>
+<a name="line-375"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-376"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>FunctionDef</span><span class='hs-layout'>)</span>
+<a name="line-377"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>function</span><span class='hs-layout'>)</span>
+<a name="line-378"></a>                <span class='hs-varid'>gradient__field_descriptor</span>
+<a name="line-379"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"gradient"</span>
+<a name="line-380"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-381"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>GradientDef</span><span class='hs-layout'>)</span>
+<a name="line-382"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>gradient</span><span class='hs-layout'>)</span>
+<a name="line-383"></a>              <span class='hs-keyword'>in</span>
+<a name="line-384"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-385"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-386"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>function__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-387"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>gradient__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-388"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-389"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"function"</span><span class='hs-layout'>,</span> <span class='hs-varid'>function__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-390"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"gradient"</span><span class='hs-layout'>,</span> <span class='hs-varid'>gradient__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-391"></a>
+<a name="line-392"></a><a name="GradientDef"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>GradientDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>GradientDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_GradientDef'functionName</span> <span class='hs-keyglyph'>::</span>
+<a name="line-393"></a>                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-394"></a>                               <span class='hs-sel'>_GradientDef'gradientFunc</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>}</span>
+<a name="line-395"></a>                 <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-396"></a>
+<a name="line-397"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"functionName"</span> <span class='hs-conid'>GradientDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-398"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-399"></a>
+<a name="line-400"></a><a name="instance%20Data.ProtoLens.HasField%20%22functionName%22%20GradientDef%20GradientDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"functionName"</span> <span class='hs-conid'>GradientDef</span>
+<a name="line-401"></a>         <span class='hs-conid'>GradientDef</span> <span class='hs-keyword'>where</span>
+<a name="line-402"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-403"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GradientDef'functionName</span>
+<a name="line-404"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GradientDef'functionName</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-405"></a>
+<a name="line-406"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"gradientFunc"</span> <span class='hs-conid'>GradientDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-407"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-408"></a>
+<a name="line-409"></a><a name="instance%20Data.ProtoLens.HasField%20%22gradientFunc%22%20GradientDef%20GradientDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"gradientFunc"</span> <span class='hs-conid'>GradientDef</span>
+<a name="line-410"></a>         <span class='hs-conid'>GradientDef</span> <span class='hs-keyword'>where</span>
+<a name="line-411"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-412"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GradientDef'gradientFunc</span>
+<a name="line-413"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GradientDef'gradientFunc</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-414"></a>
+<a name="line-415"></a><a name="instance%20Data.Default.Class.Default%20GradientDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>GradientDef</span> <span class='hs-keyword'>where</span>
+<a name="line-416"></a>        <span class='hs-varid'>def</span>
+<a name="line-417"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>GradientDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_GradientDef'functionName</span> <span class='hs-keyglyph'>=</span>
+<a name="line-418"></a>                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-419"></a>                        <span class='hs-sel'>_GradientDef'gradientFunc</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-420"></a>
+<a name="line-421"></a><a name="instance%20Data.ProtoLens.Message%20GradientDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>GradientDef</span> <span class='hs-keyword'>where</span>
+<a name="line-422"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-423"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>functionName__field_descriptor</span>
+<a name="line-424"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"function_name"</span>
+<a name="line-425"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-426"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-427"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>functionName</span><span class='hs-layout'>)</span>
+<a name="line-428"></a>                <span class='hs-varid'>gradientFunc__field_descriptor</span>
+<a name="line-429"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"gradient_func"</span>
+<a name="line-430"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-431"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-432"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>gradientFunc</span><span class='hs-layout'>)</span>
+<a name="line-433"></a>              <span class='hs-keyword'>in</span>
+<a name="line-434"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-435"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-436"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>functionName__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-437"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>gradientFunc__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-438"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-439"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"function_name"</span><span class='hs-layout'>,</span> <span class='hs-varid'>functionName__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-440"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"gradient_func"</span><span class='hs-layout'>,</span> <span class='hs-varid'>gradientFunc__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-441"></a>
+<a name="line-442"></a><a name="arg"></a><span class='hs-definition'>arg</span> <span class='hs-keyglyph'>::</span>
+<a name="line-443"></a>    <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"arg"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-444"></a>      <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"arg"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-445"></a>        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"arg"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-446"></a><span class='hs-definition'>arg</span>
+<a name="line-447"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-448"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"arg"</span><span class='hs-layout'>)</span>
+<a name="line-449"></a>
+<a name="line-450"></a><a name="attr"></a><span class='hs-definition'>attr</span> <span class='hs-keyglyph'>::</span>
+<a name="line-451"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"attr"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-452"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"attr"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-453"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"attr"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-454"></a><span class='hs-definition'>attr</span>
+<a name="line-455"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-456"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"attr"</span><span class='hs-layout'>)</span>
+<a name="line-457"></a>
+<a name="line-458"></a><a name="dep"></a><span class='hs-definition'>dep</span> <span class='hs-keyglyph'>::</span>
+<a name="line-459"></a>    <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"dep"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-460"></a>      <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dep"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-461"></a>        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dep"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-462"></a><span class='hs-definition'>dep</span>
+<a name="line-463"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-464"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"dep"</span><span class='hs-layout'>)</span>
+<a name="line-465"></a>
+<a name="line-466"></a><a name="function"></a><span class='hs-definition'>function</span> <span class='hs-keyglyph'>::</span>
+<a name="line-467"></a>         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"function"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-468"></a>           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"function"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-469"></a>             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"function"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-470"></a><span class='hs-definition'>function</span>
+<a name="line-471"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-472"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"function"</span><span class='hs-layout'>)</span>
+<a name="line-473"></a>
+<a name="line-474"></a><a name="functionName"></a><span class='hs-definition'>functionName</span> <span class='hs-keyglyph'>::</span>
+<a name="line-475"></a>             <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-476"></a>               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"functionName"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-477"></a>               <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-478"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"functionName"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-479"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"functionName"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-480"></a><span class='hs-definition'>functionName</span>
+<a name="line-481"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-482"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"functionName"</span><span class='hs-layout'>)</span>
+<a name="line-483"></a>
+<a name="line-484"></a><a name="gradient"></a><span class='hs-definition'>gradient</span> <span class='hs-keyglyph'>::</span>
+<a name="line-485"></a>         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"gradient"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-486"></a>           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"gradient"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-487"></a>             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"gradient"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-488"></a><span class='hs-definition'>gradient</span>
+<a name="line-489"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-490"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"gradient"</span><span class='hs-layout'>)</span>
+<a name="line-491"></a>
+<a name="line-492"></a><a name="gradientFunc"></a><span class='hs-definition'>gradientFunc</span> <span class='hs-keyglyph'>::</span>
+<a name="line-493"></a>             <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-494"></a>               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"gradientFunc"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-495"></a>               <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-496"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"gradientFunc"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-497"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"gradientFunc"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-498"></a><span class='hs-definition'>gradientFunc</span>
+<a name="line-499"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-500"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"gradientFunc"</span><span class='hs-layout'>)</span>
+<a name="line-501"></a>
+<a name="line-502"></a><a name="key"></a><span class='hs-definition'>key</span> <span class='hs-keyglyph'>::</span>
+<a name="line-503"></a>    <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"key"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-504"></a>      <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"key"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-505"></a>        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"key"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-506"></a><span class='hs-definition'>key</span>
+<a name="line-507"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-508"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"key"</span><span class='hs-layout'>)</span>
+<a name="line-509"></a>
+<a name="line-510"></a><a name="maybe'signature"></a><span class='hs-definition'>maybe'signature</span> <span class='hs-keyglyph'>::</span>
+<a name="line-511"></a>                <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-512"></a>                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'signature"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-513"></a>                  <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-514"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'signature"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-515"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'signature"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-516"></a><span class='hs-definition'>maybe'signature</span>
+<a name="line-517"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-518"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-519"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'signature"</span><span class='hs-layout'>)</span>
+<a name="line-520"></a>
+<a name="line-521"></a><a name="maybe'value"></a><span class='hs-definition'>maybe'value</span> <span class='hs-keyglyph'>::</span>
+<a name="line-522"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'value"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-523"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'value"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-524"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'value"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-525"></a><span class='hs-definition'>maybe'value</span>
+<a name="line-526"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-527"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'value"</span><span class='hs-layout'>)</span>
+<a name="line-528"></a>
+<a name="line-529"></a><a name="node"></a><span class='hs-definition'>node</span> <span class='hs-keyglyph'>::</span>
+<a name="line-530"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"node"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-531"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"node"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-532"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"node"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-533"></a><span class='hs-definition'>node</span>
+<a name="line-534"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-535"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"node"</span><span class='hs-layout'>)</span>
+<a name="line-536"></a>
+<a name="line-537"></a><a name="nodeDef"></a><span class='hs-definition'>nodeDef</span> <span class='hs-keyglyph'>::</span>
+<a name="line-538"></a>        <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"nodeDef"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-539"></a>          <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"nodeDef"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-540"></a>            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"nodeDef"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-541"></a><span class='hs-definition'>nodeDef</span>
+<a name="line-542"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-543"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"nodeDef"</span><span class='hs-layout'>)</span>
+<a name="line-544"></a>
+<a name="line-545"></a><a name="op"></a><span class='hs-definition'>op</span> <span class='hs-keyglyph'>::</span>
+<a name="line-546"></a>   <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"op"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-547"></a>     <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"op"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-548"></a>       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"op"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-549"></a><span class='hs-definition'>op</span>
+<a name="line-550"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-551"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"op"</span><span class='hs-layout'>)</span>
+<a name="line-552"></a>
+<a name="line-553"></a><a name="ret"></a><span class='hs-definition'>ret</span> <span class='hs-keyglyph'>::</span>
+<a name="line-554"></a>    <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"ret"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-555"></a>      <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"ret"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-556"></a>        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"ret"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-557"></a><span class='hs-definition'>ret</span>
+<a name="line-558"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-559"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"ret"</span><span class='hs-layout'>)</span>
+<a name="line-560"></a>
+<a name="line-561"></a><a name="signature"></a><span class='hs-definition'>signature</span> <span class='hs-keyglyph'>::</span>
+<a name="line-562"></a>          <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"signature"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-563"></a>            <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"signature"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-564"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"signature"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-565"></a><span class='hs-definition'>signature</span>
+<a name="line-566"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-567"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"signature"</span><span class='hs-layout'>)</span>
+<a name="line-568"></a>
+<a name="line-569"></a><a name="value"></a><span class='hs-definition'>value</span> <span class='hs-keyglyph'>::</span>
+<a name="line-570"></a>      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"value"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-571"></a>        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"value"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-572"></a>          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"value"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-573"></a><span class='hs-definition'>value</span>
+<a name="line-574"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-575"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"value"</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Graph.html b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Graph.html
new file mode 100644
index 0000000..b6ab688
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Graph.html
@@ -0,0 +1,198 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Proto/Tensorflow/Core/Framework/Graph.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>{- This file was auto-generated from tensorflow/core/framework/graph.proto by the proto-lens-protoc program. -}</span>
+<a name="line-2"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables, DataKinds, TypeFamilies,
+<a name="line-3"></a>  MultiParamTypeClasses, FlexibleContexts, FlexibleInstances,
+<a name="line-4"></a>  PatternSynonyms #-}</span>
+<a name="line-5"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-unused-imports#-}</span>
+<a name="line-6"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Graph</span> <span class='hs-keyword'>where</span>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Prelude</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-11"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-12"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-13"></a>       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-14"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-15"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-16"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-17"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-18"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-19"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-20"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-24"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-25"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Function</span>
+<a name="line-26"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span>
+<a name="line-27"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Versions</span>
+<a name="line-28"></a>
+<a name="line-29"></a><a name="GraphDef"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>GraphDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>GraphDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_GraphDef'node</span> <span class='hs-keyglyph'>::</span>
+<a name="line-30"></a>                         <span class='hs-keyglyph'>[</span><span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-31"></a>                         <span class='hs-sel'>_GraphDef'versions</span> <span class='hs-keyglyph'>::</span>
+<a name="line-32"></a>                         <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Versions</span><span class='hs-varop'>.</span><span class='hs-conid'>VersionDef</span><span class='hs-layout'>,</span>
+<a name="line-33"></a>                         <span class='hs-sel'>_GraphDef'version</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-34"></a>                         <span class='hs-sel'>_GraphDef'library</span> <span class='hs-keyglyph'>::</span>
+<a name="line-35"></a>                         <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-36"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Function</span><span class='hs-varop'>.</span><span class='hs-conid'>FunctionDefLibrary</span><span class='hs-layout'>}</span>
+<a name="line-37"></a>              <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-38"></a>
+<a name="line-39"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"node"</span> <span class='hs-conid'>GraphDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-40"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span><span class='hs-keyglyph'>]</span>
+<a name="line-41"></a>
+<a name="line-42"></a><a name="instance%20Data.ProtoLens.HasField%20%22node%22%20GraphDef%20GraphDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"node"</span> <span class='hs-conid'>GraphDef</span> <span class='hs-conid'>GraphDef</span> <span class='hs-keyword'>where</span>
+<a name="line-43"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-44"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GraphDef'node</span>
+<a name="line-45"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GraphDef'node</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-46"></a>
+<a name="line-47"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"versions"</span> <span class='hs-conid'>GraphDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-48"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Versions</span><span class='hs-varop'>.</span><span class='hs-conid'>VersionDef</span>
+<a name="line-49"></a>
+<a name="line-50"></a><a name="instance%20Data.ProtoLens.HasField%20%22versions%22%20GraphDef%20GraphDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"versions"</span> <span class='hs-conid'>GraphDef</span> <span class='hs-conid'>GraphDef</span> <span class='hs-keyword'>where</span>
+<a name="line-51"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-52"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'versions</span>
+<a name="line-53"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-54"></a>
+<a name="line-55"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'versions"</span> <span class='hs-conid'>GraphDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-56"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Versions</span><span class='hs-varop'>.</span><span class='hs-conid'>VersionDef</span>
+<a name="line-57"></a>
+<a name="line-58"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'versions%22%20GraphDef%20GraphDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'versions"</span> <span class='hs-conid'>GraphDef</span> <span class='hs-conid'>GraphDef</span>
+<a name="line-59"></a>         <span class='hs-keyword'>where</span>
+<a name="line-60"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-61"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GraphDef'versions</span>
+<a name="line-62"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GraphDef'versions</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-63"></a>
+<a name="line-64"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"version"</span> <span class='hs-conid'>GraphDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-65"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-66"></a>
+<a name="line-67"></a><a name="instance%20Data.ProtoLens.HasField%20%22version%22%20GraphDef%20GraphDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"version"</span> <span class='hs-conid'>GraphDef</span> <span class='hs-conid'>GraphDef</span> <span class='hs-keyword'>where</span>
+<a name="line-68"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-69"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GraphDef'version</span>
+<a name="line-70"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GraphDef'version</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-71"></a>
+<a name="line-72"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"library"</span> <span class='hs-conid'>GraphDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-73"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Function</span><span class='hs-varop'>.</span><span class='hs-conid'>FunctionDefLibrary</span>
+<a name="line-74"></a>
+<a name="line-75"></a><a name="instance%20Data.ProtoLens.HasField%20%22library%22%20GraphDef%20GraphDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"library"</span> <span class='hs-conid'>GraphDef</span> <span class='hs-conid'>GraphDef</span> <span class='hs-keyword'>where</span>
+<a name="line-76"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-77"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'library</span>
+<a name="line-78"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-79"></a>
+<a name="line-80"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'library"</span> <span class='hs-conid'>GraphDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-81"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-82"></a>       <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Function</span><span class='hs-varop'>.</span><span class='hs-conid'>FunctionDefLibrary</span>
+<a name="line-83"></a>
+<a name="line-84"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'library%22%20GraphDef%20GraphDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'library"</span> <span class='hs-conid'>GraphDef</span> <span class='hs-conid'>GraphDef</span>
+<a name="line-85"></a>         <span class='hs-keyword'>where</span>
+<a name="line-86"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-87"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GraphDef'library</span>
+<a name="line-88"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GraphDef'library</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-89"></a>
+<a name="line-90"></a><a name="instance%20Data.Default.Class.Default%20GraphDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>GraphDef</span> <span class='hs-keyword'>where</span>
+<a name="line-91"></a>        <span class='hs-varid'>def</span>
+<a name="line-92"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>GraphDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_GraphDef'node</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-93"></a>                     <span class='hs-sel'>_GraphDef'versions</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-94"></a>                     <span class='hs-sel'>_GraphDef'version</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-95"></a>                     <span class='hs-sel'>_GraphDef'library</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>}</span>
+<a name="line-96"></a>
+<a name="line-97"></a><a name="instance%20Data.ProtoLens.Message%20GraphDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>GraphDef</span> <span class='hs-keyword'>where</span>
+<a name="line-98"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-99"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>node__field_descriptor</span>
+<a name="line-100"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"node"</span>
+<a name="line-101"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-102"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-103"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span><span class='hs-layout'>)</span>
+<a name="line-104"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>node</span><span class='hs-layout'>)</span>
+<a name="line-105"></a>                <span class='hs-varid'>versions__field_descriptor</span>
+<a name="line-106"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"versions"</span>
+<a name="line-107"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-108"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-109"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Versions</span><span class='hs-varop'>.</span><span class='hs-conid'>VersionDef</span><span class='hs-layout'>)</span>
+<a name="line-110"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'versions</span><span class='hs-layout'>)</span>
+<a name="line-111"></a>                <span class='hs-varid'>version__field_descriptor</span>
+<a name="line-112"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"version"</span>
+<a name="line-113"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-114"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-115"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>version</span><span class='hs-layout'>)</span>
+<a name="line-116"></a>                <span class='hs-varid'>library__field_descriptor</span>
+<a name="line-117"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"library"</span>
+<a name="line-118"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-119"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-120"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Function</span><span class='hs-varop'>.</span><span class='hs-conid'>FunctionDefLibrary</span><span class='hs-layout'>)</span>
+<a name="line-121"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'library</span><span class='hs-layout'>)</span>
+<a name="line-122"></a>              <span class='hs-keyword'>in</span>
+<a name="line-123"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-124"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-125"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>node__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-126"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>versions__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-127"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>version__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-128"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>library__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-129"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-130"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"node"</span><span class='hs-layout'>,</span> <span class='hs-varid'>node__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-131"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"versions"</span><span class='hs-layout'>,</span> <span class='hs-varid'>versions__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-132"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"version"</span><span class='hs-layout'>,</span> <span class='hs-varid'>version__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-133"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"library"</span><span class='hs-layout'>,</span> <span class='hs-varid'>library__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-134"></a>
+<a name="line-135"></a><a name="library"></a><span class='hs-definition'>library</span> <span class='hs-keyglyph'>::</span>
+<a name="line-136"></a>        <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"library"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-137"></a>          <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"library"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-138"></a>            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"library"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-139"></a><span class='hs-definition'>library</span>
+<a name="line-140"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-141"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"library"</span><span class='hs-layout'>)</span>
+<a name="line-142"></a>
+<a name="line-143"></a><a name="maybe'library"></a><span class='hs-definition'>maybe'library</span> <span class='hs-keyglyph'>::</span>
+<a name="line-144"></a>              <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-145"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'library"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-146"></a>                <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-147"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'library"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-148"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'library"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-149"></a><span class='hs-definition'>maybe'library</span>
+<a name="line-150"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-151"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-152"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'library"</span><span class='hs-layout'>)</span>
+<a name="line-153"></a>
+<a name="line-154"></a><a name="maybe'versions"></a><span class='hs-definition'>maybe'versions</span> <span class='hs-keyglyph'>::</span>
+<a name="line-155"></a>               <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-156"></a>                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'versions"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-157"></a>                 <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-158"></a>                   <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'versions"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-159"></a>                   <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'versions"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-160"></a><span class='hs-definition'>maybe'versions</span>
+<a name="line-161"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-162"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-163"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'versions"</span><span class='hs-layout'>)</span>
+<a name="line-164"></a>
+<a name="line-165"></a><a name="node"></a><span class='hs-definition'>node</span> <span class='hs-keyglyph'>::</span>
+<a name="line-166"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"node"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-167"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"node"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-168"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"node"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-169"></a><span class='hs-definition'>node</span>
+<a name="line-170"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-171"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"node"</span><span class='hs-layout'>)</span>
+<a name="line-172"></a>
+<a name="line-173"></a><a name="version"></a><span class='hs-definition'>version</span> <span class='hs-keyglyph'>::</span>
+<a name="line-174"></a>        <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"version"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-175"></a>          <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"version"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-176"></a>            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"version"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-177"></a><span class='hs-definition'>version</span>
+<a name="line-178"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-179"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"version"</span><span class='hs-layout'>)</span>
+<a name="line-180"></a>
+<a name="line-181"></a><a name="versions"></a><span class='hs-definition'>versions</span> <span class='hs-keyglyph'>::</span>
+<a name="line-182"></a>         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"versions"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-183"></a>           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"versions"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-184"></a>             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"versions"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-185"></a><span class='hs-definition'>versions</span>
+<a name="line-186"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-187"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"versions"</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-NodeDef.html b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-NodeDef.html
new file mode 100644
index 0000000..a91bb4a
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-NodeDef.html
@@ -0,0 +1,257 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Proto/Tensorflow/Core/Framework/NodeDef.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>{- This file was auto-generated from tensorflow/core/framework/node_def.proto by the proto-lens-protoc program. -}</span>
+<a name="line-2"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables, DataKinds, TypeFamilies,
+<a name="line-3"></a>  MultiParamTypeClasses, FlexibleContexts, FlexibleInstances,
+<a name="line-4"></a>  PatternSynonyms #-}</span>
+<a name="line-5"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-unused-imports#-}</span>
+<a name="line-6"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>NodeDef</span> <span class='hs-keyword'>where</span>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Prelude</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-11"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-12"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-13"></a>       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-14"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-15"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-16"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-17"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-18"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-19"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-20"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-24"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-25"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span>
+<a name="line-26"></a>
+<a name="line-27"></a><a name="NodeDef"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>NodeDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NodeDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeDef'name</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-28"></a>                       <span class='hs-sel'>_NodeDef'op</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span> <span class='hs-sel'>_NodeDef'input</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-29"></a>                       <span class='hs-sel'>_NodeDef'device</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-30"></a>                       <span class='hs-sel'>_NodeDef'attr</span> <span class='hs-keyglyph'>::</span>
+<a name="line-31"></a>                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-32"></a>                         <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-layout'>}</span>
+<a name="line-33"></a>             <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-34"></a>
+<a name="line-35"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-conid'>NodeDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-36"></a>
+<a name="line-37"></a><a name="instance%20Data.ProtoLens.HasField%20%22name%22%20NodeDef%20NodeDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"name"</span> <span class='hs-conid'>NodeDef</span> <span class='hs-conid'>NodeDef</span> <span class='hs-keyword'>where</span>
+<a name="line-38"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-39"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeDef'name</span>
+<a name="line-40"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeDef'name</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-41"></a>
+<a name="line-42"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"op"</span> <span class='hs-conid'>NodeDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-43"></a>
+<a name="line-44"></a><a name="instance%20Data.ProtoLens.HasField%20%22op%22%20NodeDef%20NodeDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"op"</span> <span class='hs-conid'>NodeDef</span> <span class='hs-conid'>NodeDef</span> <span class='hs-keyword'>where</span>
+<a name="line-45"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-46"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeDef'op</span>
+<a name="line-47"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeDef'op</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-48"></a>
+<a name="line-49"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"input"</span> <span class='hs-conid'>NodeDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-50"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-keyglyph'>]</span>
+<a name="line-51"></a>
+<a name="line-52"></a><a name="instance%20Data.ProtoLens.HasField%20%22input%22%20NodeDef%20NodeDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"input"</span> <span class='hs-conid'>NodeDef</span> <span class='hs-conid'>NodeDef</span> <span class='hs-keyword'>where</span>
+<a name="line-53"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-54"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeDef'input</span>
+<a name="line-55"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeDef'input</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-56"></a>
+<a name="line-57"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"device"</span> <span class='hs-conid'>NodeDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-58"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-59"></a>
+<a name="line-60"></a><a name="instance%20Data.ProtoLens.HasField%20%22device%22%20NodeDef%20NodeDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"device"</span> <span class='hs-conid'>NodeDef</span> <span class='hs-conid'>NodeDef</span> <span class='hs-keyword'>where</span>
+<a name="line-61"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-62"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeDef'device</span>
+<a name="line-63"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeDef'device</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-64"></a>
+<a name="line-65"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"attr"</span> <span class='hs-conid'>NodeDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-66"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-67"></a>       <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span>
+<a name="line-68"></a>
+<a name="line-69"></a><a name="instance%20Data.ProtoLens.HasField%20%22attr%22%20NodeDef%20NodeDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"attr"</span> <span class='hs-conid'>NodeDef</span> <span class='hs-conid'>NodeDef</span> <span class='hs-keyword'>where</span>
+<a name="line-70"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-71"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeDef'attr</span>
+<a name="line-72"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeDef'attr</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-73"></a>
+<a name="line-74"></a><a name="instance%20Data.Default.Class.Default%20NodeDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>NodeDef</span> <span class='hs-keyword'>where</span>
+<a name="line-75"></a>        <span class='hs-varid'>def</span>
+<a name="line-76"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NodeDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeDef'name</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-77"></a>                    <span class='hs-sel'>_NodeDef'op</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span> <span class='hs-sel'>_NodeDef'input</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-78"></a>                    <span class='hs-sel'>_NodeDef'device</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-79"></a>                    <span class='hs-sel'>_NodeDef'attr</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>empty</span><span class='hs-layout'>}</span>
+<a name="line-80"></a>
+<a name="line-81"></a><a name="instance%20Data.ProtoLens.Message%20NodeDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>NodeDef</span> <span class='hs-keyword'>where</span>
+<a name="line-82"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-83"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>name__field_descriptor</span>
+<a name="line-84"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"name"</span>
+<a name="line-85"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-86"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-87"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span>
+<a name="line-88"></a>                <span class='hs-varid'>op__field_descriptor</span>
+<a name="line-89"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"op"</span>
+<a name="line-90"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-91"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-92"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>op</span><span class='hs-layout'>)</span>
+<a name="line-93"></a>                <span class='hs-varid'>input__field_descriptor</span>
+<a name="line-94"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"input"</span>
+<a name="line-95"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-96"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-97"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>input</span><span class='hs-layout'>)</span>
+<a name="line-98"></a>                <span class='hs-varid'>device__field_descriptor</span>
+<a name="line-99"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"device"</span>
+<a name="line-100"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-101"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-102"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>device</span><span class='hs-layout'>)</span>
+<a name="line-103"></a>                <span class='hs-varid'>attr__field_descriptor</span>
+<a name="line-104"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"attr"</span>
+<a name="line-105"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-106"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>NodeDef'AttrEntry</span><span class='hs-layout'>)</span>
+<a name="line-107"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MapField</span> <span class='hs-varid'>key</span> <span class='hs-varid'>value</span> <span class='hs-varid'>attr</span><span class='hs-layout'>)</span>
+<a name="line-108"></a>              <span class='hs-keyword'>in</span>
+<a name="line-109"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-110"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-111"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>name__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-112"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>op__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-113"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>input__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-114"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>device__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-115"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>5</span><span class='hs-layout'>,</span> <span class='hs-varid'>attr__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-116"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-117"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"name"</span><span class='hs-layout'>,</span> <span class='hs-varid'>name__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-str'>"op"</span><span class='hs-layout'>,</span> <span class='hs-varid'>op__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-118"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"input"</span><span class='hs-layout'>,</span> <span class='hs-varid'>input__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-119"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"device"</span><span class='hs-layout'>,</span> <span class='hs-varid'>device__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-120"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"attr"</span><span class='hs-layout'>,</span> <span class='hs-varid'>attr__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-121"></a>
+<a name="line-122"></a><a name="NodeDef'AttrEntry"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>NodeDef'AttrEntry</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NodeDef'AttrEntry</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeDef'AttrEntry'key</span>
+<a name="line-123"></a>                                           <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-124"></a>                                           <span class='hs-sel'>_NodeDef'AttrEntry'value</span> <span class='hs-keyglyph'>::</span>
+<a name="line-125"></a>                                           <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-126"></a>                                             <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-layout'>}</span>
+<a name="line-127"></a>                       <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-128"></a>
+<a name="line-129"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"key"</span> <span class='hs-conid'>NodeDef'AttrEntry</span> <span class='hs-keyglyph'>=</span>
+<a name="line-130"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-131"></a>
+<a name="line-132"></a><a name="instance%20Data.ProtoLens.HasField%20%22key%22%20NodeDef'AttrEntry%20NodeDef'AttrEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"key"</span> <span class='hs-conid'>NodeDef'AttrEntry</span>
+<a name="line-133"></a>         <span class='hs-conid'>NodeDef'AttrEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-134"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-135"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeDef'AttrEntry'key</span>
+<a name="line-136"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeDef'AttrEntry'key</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-137"></a>
+<a name="line-138"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"value"</span> <span class='hs-conid'>NodeDef'AttrEntry</span> <span class='hs-keyglyph'>=</span>
+<a name="line-139"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span>
+<a name="line-140"></a>
+<a name="line-141"></a><a name="instance%20Data.ProtoLens.HasField%20%22value%22%20NodeDef'AttrEntry%20NodeDef'AttrEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"value"</span> <span class='hs-conid'>NodeDef'AttrEntry</span>
+<a name="line-142"></a>         <span class='hs-conid'>NodeDef'AttrEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-143"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-144"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'value</span>
+<a name="line-145"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-146"></a>
+<a name="line-147"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'value"</span> <span class='hs-conid'>NodeDef'AttrEntry</span>
+<a name="line-148"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span>
+<a name="line-149"></a>
+<a name="line-150"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'value%22%20NodeDef'AttrEntry%20NodeDef'AttrEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'value"</span> <span class='hs-conid'>NodeDef'AttrEntry</span>
+<a name="line-151"></a>         <span class='hs-conid'>NodeDef'AttrEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-152"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-153"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeDef'AttrEntry'value</span>
+<a name="line-154"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeDef'AttrEntry'value</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-155"></a>
+<a name="line-156"></a><a name="instance%20Data.Default.Class.Default%20NodeDef'AttrEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>NodeDef'AttrEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-157"></a>        <span class='hs-varid'>def</span>
+<a name="line-158"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NodeDef'AttrEntry</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeDef'AttrEntry'key</span> <span class='hs-keyglyph'>=</span>
+<a name="line-159"></a>                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-160"></a>                              <span class='hs-sel'>_NodeDef'AttrEntry'value</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>}</span>
+<a name="line-161"></a>
+<a name="line-162"></a><a name="instance%20Data.ProtoLens.Message%20NodeDef'AttrEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>NodeDef'AttrEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-163"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-164"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>key__field_descriptor</span>
+<a name="line-165"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"key"</span>
+<a name="line-166"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-167"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-168"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>key</span><span class='hs-layout'>)</span>
+<a name="line-169"></a>                <span class='hs-varid'>value__field_descriptor</span>
+<a name="line-170"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"value"</span>
+<a name="line-171"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-172"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-173"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-layout'>)</span>
+<a name="line-174"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'value</span><span class='hs-layout'>)</span>
+<a name="line-175"></a>              <span class='hs-keyword'>in</span>
+<a name="line-176"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-177"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-178"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>key__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-179"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>value__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-180"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-181"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"key"</span><span class='hs-layout'>,</span> <span class='hs-varid'>key__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-182"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"value"</span><span class='hs-layout'>,</span> <span class='hs-varid'>value__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-183"></a>
+<a name="line-184"></a><a name="attr"></a><span class='hs-definition'>attr</span> <span class='hs-keyglyph'>::</span>
+<a name="line-185"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"attr"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-186"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"attr"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-187"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"attr"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-188"></a><span class='hs-definition'>attr</span>
+<a name="line-189"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-190"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"attr"</span><span class='hs-layout'>)</span>
+<a name="line-191"></a>
+<a name="line-192"></a><a name="device"></a><span class='hs-definition'>device</span> <span class='hs-keyglyph'>::</span>
+<a name="line-193"></a>       <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"device"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-194"></a>         <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"device"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-195"></a>           <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"device"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-196"></a><span class='hs-definition'>device</span>
+<a name="line-197"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-198"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"device"</span><span class='hs-layout'>)</span>
+<a name="line-199"></a>
+<a name="line-200"></a><a name="input"></a><span class='hs-definition'>input</span> <span class='hs-keyglyph'>::</span>
+<a name="line-201"></a>      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"input"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-202"></a>        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"input"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-203"></a>          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"input"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-204"></a><span class='hs-definition'>input</span>
+<a name="line-205"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-206"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"input"</span><span class='hs-layout'>)</span>
+<a name="line-207"></a>
+<a name="line-208"></a><a name="key"></a><span class='hs-definition'>key</span> <span class='hs-keyglyph'>::</span>
+<a name="line-209"></a>    <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"key"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-210"></a>      <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"key"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-211"></a>        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"key"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-212"></a><span class='hs-definition'>key</span>
+<a name="line-213"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-214"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"key"</span><span class='hs-layout'>)</span>
+<a name="line-215"></a>
+<a name="line-216"></a><a name="maybe'value"></a><span class='hs-definition'>maybe'value</span> <span class='hs-keyglyph'>::</span>
+<a name="line-217"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'value"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-218"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'value"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-219"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'value"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-220"></a><span class='hs-definition'>maybe'value</span>
+<a name="line-221"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-222"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'value"</span><span class='hs-layout'>)</span>
+<a name="line-223"></a>
+<a name="line-224"></a><a name="name"></a><span class='hs-definition'>name</span> <span class='hs-keyglyph'>::</span>
+<a name="line-225"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-226"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-227"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-228"></a><span class='hs-definition'>name</span>
+<a name="line-229"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-230"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"name"</span><span class='hs-layout'>)</span>
+<a name="line-231"></a>
+<a name="line-232"></a><a name="op"></a><span class='hs-definition'>op</span> <span class='hs-keyglyph'>::</span>
+<a name="line-233"></a>   <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"op"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-234"></a>     <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"op"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-235"></a>       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"op"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-236"></a><span class='hs-definition'>op</span>
+<a name="line-237"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-238"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"op"</span><span class='hs-layout'>)</span>
+<a name="line-239"></a>
+<a name="line-240"></a><a name="value"></a><span class='hs-definition'>value</span> <span class='hs-keyglyph'>::</span>
+<a name="line-241"></a>      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"value"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-242"></a>        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"value"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-243"></a>          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"value"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-244"></a><span class='hs-definition'>value</span>
+<a name="line-245"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-246"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"value"</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-OpDef.html b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-OpDef.html
new file mode 100644
index 0000000..fe7a2c4
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-OpDef.html
@@ -0,0 +1,854 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Proto/Tensorflow/Core/Framework/OpDef.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>{- This file was auto-generated from tensorflow/core/framework/op_def.proto by the proto-lens-protoc program. -}</span>
+<a name="line-2"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables, DataKinds, TypeFamilies,
+<a name="line-3"></a>  MultiParamTypeClasses, FlexibleContexts, FlexibleInstances,
+<a name="line-4"></a>  PatternSynonyms #-}</span>
+<a name="line-5"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-unused-imports#-}</span>
+<a name="line-6"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>OpDef</span> <span class='hs-keyword'>where</span>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Prelude</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-11"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-12"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-13"></a>       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-14"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-15"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-16"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-17"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-18"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-19"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-20"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-24"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-25"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span>
+<a name="line-26"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span>
+<a name="line-27"></a>
+<a name="line-28"></a><a name="OpDef"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OpDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'name</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-29"></a>                   <span class='hs-sel'>_OpDef'inputArg</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>OpDef'ArgDef</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-30"></a>                   <span class='hs-sel'>_OpDef'outputArg</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>OpDef'ArgDef</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span> <span class='hs-sel'>_OpDef'attr</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>OpDef'AttrDef</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-31"></a>                   <span class='hs-sel'>_OpDef'deprecation</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>OpDeprecation</span><span class='hs-layout'>,</span>
+<a name="line-32"></a>                   <span class='hs-sel'>_OpDef'summary</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-33"></a>                   <span class='hs-sel'>_OpDef'description</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-34"></a>                   <span class='hs-sel'>_OpDef'isCommutative</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-35"></a>                   <span class='hs-sel'>_OpDef'isAggregate</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-36"></a>                   <span class='hs-sel'>_OpDef'isStateful</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-37"></a>                   <span class='hs-sel'>_OpDef'allowsUninitializedInput</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>}</span>
+<a name="line-38"></a>           <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-39"></a>
+<a name="line-40"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-41"></a>
+<a name="line-42"></a><a name="instance%20Data.ProtoLens.HasField%20%22name%22%20OpDef%20OpDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"name"</span> <span class='hs-conid'>OpDef</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyword'>where</span>
+<a name="line-43"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-44"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'name</span>
+<a name="line-45"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'name</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-46"></a>
+<a name="line-47"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"inputArg"</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-48"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>OpDef'ArgDef</span><span class='hs-keyglyph'>]</span>
+<a name="line-49"></a>
+<a name="line-50"></a><a name="instance%20Data.ProtoLens.HasField%20%22inputArg%22%20OpDef%20OpDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"inputArg"</span> <span class='hs-conid'>OpDef</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyword'>where</span>
+<a name="line-51"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-52"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'inputArg</span>
+<a name="line-53"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'inputArg</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-54"></a>
+<a name="line-55"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"outputArg"</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-56"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>OpDef'ArgDef</span><span class='hs-keyglyph'>]</span>
+<a name="line-57"></a>
+<a name="line-58"></a><a name="instance%20Data.ProtoLens.HasField%20%22outputArg%22%20OpDef%20OpDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"outputArg"</span> <span class='hs-conid'>OpDef</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyword'>where</span>
+<a name="line-59"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-60"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'outputArg</span>
+<a name="line-61"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'outputArg</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-62"></a>
+<a name="line-63"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"attr"</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>OpDef'AttrDef</span><span class='hs-keyglyph'>]</span>
+<a name="line-64"></a>
+<a name="line-65"></a><a name="instance%20Data.ProtoLens.HasField%20%22attr%22%20OpDef%20OpDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"attr"</span> <span class='hs-conid'>OpDef</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyword'>where</span>
+<a name="line-66"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-67"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'attr</span>
+<a name="line-68"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'attr</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-69"></a>
+<a name="line-70"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"deprecation"</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-71"></a>     <span class='hs-conid'>OpDeprecation</span>
+<a name="line-72"></a>
+<a name="line-73"></a><a name="instance%20Data.ProtoLens.HasField%20%22deprecation%22%20OpDef%20OpDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"deprecation"</span> <span class='hs-conid'>OpDef</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyword'>where</span>
+<a name="line-74"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-75"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'deprecation</span>
+<a name="line-76"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-77"></a>
+<a name="line-78"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'deprecation"</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-79"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>OpDeprecation</span>
+<a name="line-80"></a>
+<a name="line-81"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'deprecation%22%20OpDef%20OpDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'deprecation"</span> <span class='hs-conid'>OpDef</span> <span class='hs-conid'>OpDef</span>
+<a name="line-82"></a>         <span class='hs-keyword'>where</span>
+<a name="line-83"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-84"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'deprecation</span>
+<a name="line-85"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'deprecation</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-86"></a>
+<a name="line-87"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"summary"</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-88"></a>
+<a name="line-89"></a><a name="instance%20Data.ProtoLens.HasField%20%22summary%22%20OpDef%20OpDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"summary"</span> <span class='hs-conid'>OpDef</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyword'>where</span>
+<a name="line-90"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-91"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'summary</span>
+<a name="line-92"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'summary</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-93"></a>
+<a name="line-94"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"description"</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-95"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-96"></a>
+<a name="line-97"></a><a name="instance%20Data.ProtoLens.HasField%20%22description%22%20OpDef%20OpDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"description"</span> <span class='hs-conid'>OpDef</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyword'>where</span>
+<a name="line-98"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-99"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'description</span>
+<a name="line-100"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'description</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-101"></a>
+<a name="line-102"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"isCommutative"</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-103"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-104"></a>
+<a name="line-105"></a><a name="instance%20Data.ProtoLens.HasField%20%22isCommutative%22%20OpDef%20OpDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"isCommutative"</span> <span class='hs-conid'>OpDef</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyword'>where</span>
+<a name="line-106"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-107"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'isCommutative</span>
+<a name="line-108"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'isCommutative</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-109"></a>
+<a name="line-110"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"isAggregate"</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-111"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-112"></a>
+<a name="line-113"></a><a name="instance%20Data.ProtoLens.HasField%20%22isAggregate%22%20OpDef%20OpDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"isAggregate"</span> <span class='hs-conid'>OpDef</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyword'>where</span>
+<a name="line-114"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-115"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'isAggregate</span>
+<a name="line-116"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'isAggregate</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-117"></a>
+<a name="line-118"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"isStateful"</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-119"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-120"></a>
+<a name="line-121"></a><a name="instance%20Data.ProtoLens.HasField%20%22isStateful%22%20OpDef%20OpDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"isStateful"</span> <span class='hs-conid'>OpDef</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyword'>where</span>
+<a name="line-122"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-123"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'isStateful</span>
+<a name="line-124"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'isStateful</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-125"></a>
+<a name="line-126"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allowsUninitializedInput"</span> <span class='hs-conid'>OpDef</span>
+<a name="line-127"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-128"></a>
+<a name="line-129"></a><a name="instance%20Data.ProtoLens.HasField%20%22allowsUninitializedInput%22%20OpDef%20OpDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allowsUninitializedInput"</span> <span class='hs-conid'>OpDef</span>
+<a name="line-130"></a>         <span class='hs-conid'>OpDef</span> <span class='hs-keyword'>where</span>
+<a name="line-131"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-132"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'allowsUninitializedInput</span>
+<a name="line-133"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'allowsUninitializedInput</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-134"></a>
+<a name="line-135"></a><a name="instance%20Data.Default.Class.Default%20OpDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyword'>where</span>
+<a name="line-136"></a>        <span class='hs-varid'>def</span>
+<a name="line-137"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OpDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'name</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-138"></a>                  <span class='hs-sel'>_OpDef'inputArg</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span> <span class='hs-sel'>_OpDef'outputArg</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span> <span class='hs-sel'>_OpDef'attr</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-139"></a>                  <span class='hs-sel'>_OpDef'deprecation</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-140"></a>                  <span class='hs-sel'>_OpDef'summary</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-141"></a>                  <span class='hs-sel'>_OpDef'description</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-142"></a>                  <span class='hs-sel'>_OpDef'isCommutative</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-143"></a>                  <span class='hs-sel'>_OpDef'isAggregate</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-144"></a>                  <span class='hs-sel'>_OpDef'isStateful</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-145"></a>                  <span class='hs-sel'>_OpDef'allowsUninitializedInput</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-146"></a>
+<a name="line-147"></a><a name="instance%20Data.ProtoLens.Message%20OpDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>OpDef</span> <span class='hs-keyword'>where</span>
+<a name="line-148"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-149"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>name__field_descriptor</span>
+<a name="line-150"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"name"</span>
+<a name="line-151"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-152"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-153"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span>
+<a name="line-154"></a>                <span class='hs-varid'>inputArg__field_descriptor</span>
+<a name="line-155"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"input_arg"</span>
+<a name="line-156"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-157"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>OpDef'ArgDef</span><span class='hs-layout'>)</span>
+<a name="line-158"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>inputArg</span><span class='hs-layout'>)</span>
+<a name="line-159"></a>                <span class='hs-varid'>outputArg__field_descriptor</span>
+<a name="line-160"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"output_arg"</span>
+<a name="line-161"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-162"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>OpDef'ArgDef</span><span class='hs-layout'>)</span>
+<a name="line-163"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>outputArg</span><span class='hs-layout'>)</span>
+<a name="line-164"></a>                <span class='hs-varid'>attr__field_descriptor</span>
+<a name="line-165"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"attr"</span>
+<a name="line-166"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-167"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>OpDef'AttrDef</span><span class='hs-layout'>)</span>
+<a name="line-168"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>attr</span><span class='hs-layout'>)</span>
+<a name="line-169"></a>                <span class='hs-varid'>deprecation__field_descriptor</span>
+<a name="line-170"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"deprecation"</span>
+<a name="line-171"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-172"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>OpDeprecation</span><span class='hs-layout'>)</span>
+<a name="line-173"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'deprecation</span><span class='hs-layout'>)</span>
+<a name="line-174"></a>                <span class='hs-varid'>summary__field_descriptor</span>
+<a name="line-175"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"summary"</span>
+<a name="line-176"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-177"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-178"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>summary</span><span class='hs-layout'>)</span>
+<a name="line-179"></a>                <span class='hs-varid'>description__field_descriptor</span>
+<a name="line-180"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"description"</span>
+<a name="line-181"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-182"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-183"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>description</span><span class='hs-layout'>)</span>
+<a name="line-184"></a>                <span class='hs-varid'>isCommutative__field_descriptor</span>
+<a name="line-185"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"is_commutative"</span>
+<a name="line-186"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-187"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-188"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>isCommutative</span><span class='hs-layout'>)</span>
+<a name="line-189"></a>                <span class='hs-varid'>isAggregate__field_descriptor</span>
+<a name="line-190"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"is_aggregate"</span>
+<a name="line-191"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-192"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-193"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>isAggregate</span><span class='hs-layout'>)</span>
+<a name="line-194"></a>                <span class='hs-varid'>isStateful__field_descriptor</span>
+<a name="line-195"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"is_stateful"</span>
+<a name="line-196"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-197"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-198"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>isStateful</span><span class='hs-layout'>)</span>
+<a name="line-199"></a>                <span class='hs-varid'>allowsUninitializedInput__field_descriptor</span>
+<a name="line-200"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"allows_uninitialized_input"</span>
+<a name="line-201"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-202"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-203"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-204"></a>                         <span class='hs-varid'>allowsUninitializedInput</span><span class='hs-layout'>)</span>
+<a name="line-205"></a>              <span class='hs-keyword'>in</span>
+<a name="line-206"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-207"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-208"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>name__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-209"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>inputArg__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-210"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>outputArg__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-211"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>attr__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-212"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>8</span><span class='hs-layout'>,</span> <span class='hs-varid'>deprecation__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-213"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>5</span><span class='hs-layout'>,</span> <span class='hs-varid'>summary__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-214"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>6</span><span class='hs-layout'>,</span> <span class='hs-varid'>description__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-215"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>18</span><span class='hs-layout'>,</span> <span class='hs-varid'>isCommutative__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-216"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>16</span><span class='hs-layout'>,</span> <span class='hs-varid'>isAggregate__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-217"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>17</span><span class='hs-layout'>,</span> <span class='hs-varid'>isStateful__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-218"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>19</span><span class='hs-layout'>,</span>
+<a name="line-219"></a>                     <span class='hs-varid'>allowsUninitializedInput__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-220"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-221"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"name"</span><span class='hs-layout'>,</span> <span class='hs-varid'>name__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-222"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"input_arg"</span><span class='hs-layout'>,</span> <span class='hs-varid'>inputArg__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-223"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"output_arg"</span><span class='hs-layout'>,</span> <span class='hs-varid'>outputArg__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-224"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"attr"</span><span class='hs-layout'>,</span> <span class='hs-varid'>attr__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-225"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"deprecation"</span><span class='hs-layout'>,</span> <span class='hs-varid'>deprecation__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-226"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"summary"</span><span class='hs-layout'>,</span> <span class='hs-varid'>summary__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-227"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"description"</span><span class='hs-layout'>,</span> <span class='hs-varid'>description__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-228"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"is_commutative"</span><span class='hs-layout'>,</span> <span class='hs-varid'>isCommutative__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-229"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"is_aggregate"</span><span class='hs-layout'>,</span> <span class='hs-varid'>isAggregate__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-230"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"is_stateful"</span><span class='hs-layout'>,</span> <span class='hs-varid'>isStateful__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-231"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"allows_uninitialized_input"</span><span class='hs-layout'>,</span>
+<a name="line-232"></a>                     <span class='hs-varid'>allowsUninitializedInput__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-233"></a>
+<a name="line-234"></a><a name="OpDef'ArgDef"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OpDef'ArgDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'ArgDef'name</span> <span class='hs-keyglyph'>::</span>
+<a name="line-235"></a>                                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-236"></a>                                 <span class='hs-sel'>_OpDef'ArgDef'description</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-237"></a>                                 <span class='hs-sel'>_OpDef'ArgDef'type'</span> <span class='hs-keyglyph'>::</span>
+<a name="line-238"></a>                                 <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span><span class='hs-layout'>,</span>
+<a name="line-239"></a>                                 <span class='hs-sel'>_OpDef'ArgDef'typeAttr</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-240"></a>                                 <span class='hs-sel'>_OpDef'ArgDef'numberAttr</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-241"></a>                                 <span class='hs-sel'>_OpDef'ArgDef'typeListAttr</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-242"></a>                                 <span class='hs-sel'>_OpDef'ArgDef'isRef</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>}</span>
+<a name="line-243"></a>                  <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-244"></a>
+<a name="line-245"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-246"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-247"></a>
+<a name="line-248"></a><a name="instance%20Data.ProtoLens.HasField%20%22name%22%20OpDef'ArgDef%20OpDef'ArgDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"name"</span> <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-conid'>OpDef'ArgDef</span>
+<a name="line-249"></a>         <span class='hs-keyword'>where</span>
+<a name="line-250"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-251"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'ArgDef'name</span>
+<a name="line-252"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'ArgDef'name</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-253"></a>
+<a name="line-254"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"description"</span> <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-255"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-256"></a>
+<a name="line-257"></a><a name="instance%20Data.ProtoLens.HasField%20%22description%22%20OpDef'ArgDef%20OpDef'ArgDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"description"</span> <span class='hs-conid'>OpDef'ArgDef</span>
+<a name="line-258"></a>         <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-keyword'>where</span>
+<a name="line-259"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-260"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'ArgDef'description</span>
+<a name="line-261"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'ArgDef'description</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-262"></a>
+<a name="line-263"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"type'"</span> <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-264"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span>
+<a name="line-265"></a>
+<a name="line-266"></a><a name="instance%20Data.ProtoLens.HasField%20%22type'%22%20OpDef'ArgDef%20OpDef'ArgDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"type'"</span> <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-conid'>OpDef'ArgDef</span>
+<a name="line-267"></a>         <span class='hs-keyword'>where</span>
+<a name="line-268"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-269"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'ArgDef'type'</span>
+<a name="line-270"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'ArgDef'type'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-271"></a>
+<a name="line-272"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"typeAttr"</span> <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-273"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-274"></a>
+<a name="line-275"></a><a name="instance%20Data.ProtoLens.HasField%20%22typeAttr%22%20OpDef'ArgDef%20OpDef'ArgDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"typeAttr"</span> <span class='hs-conid'>OpDef'ArgDef</span>
+<a name="line-276"></a>         <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-keyword'>where</span>
+<a name="line-277"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-278"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'ArgDef'typeAttr</span>
+<a name="line-279"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'ArgDef'typeAttr</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-280"></a>
+<a name="line-281"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"numberAttr"</span> <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-282"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-283"></a>
+<a name="line-284"></a><a name="instance%20Data.ProtoLens.HasField%20%22numberAttr%22%20OpDef'ArgDef%20OpDef'ArgDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"numberAttr"</span> <span class='hs-conid'>OpDef'ArgDef</span>
+<a name="line-285"></a>         <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-keyword'>where</span>
+<a name="line-286"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-287"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'ArgDef'numberAttr</span>
+<a name="line-288"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'ArgDef'numberAttr</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-289"></a>
+<a name="line-290"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"typeListAttr"</span> <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-291"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-292"></a>
+<a name="line-293"></a><a name="instance%20Data.ProtoLens.HasField%20%22typeListAttr%22%20OpDef'ArgDef%20OpDef'ArgDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"typeListAttr"</span> <span class='hs-conid'>OpDef'ArgDef</span>
+<a name="line-294"></a>         <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-keyword'>where</span>
+<a name="line-295"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-296"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'ArgDef'typeListAttr</span>
+<a name="line-297"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'ArgDef'typeListAttr</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-298"></a>
+<a name="line-299"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"isRef"</span> <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-300"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-301"></a>
+<a name="line-302"></a><a name="instance%20Data.ProtoLens.HasField%20%22isRef%22%20OpDef'ArgDef%20OpDef'ArgDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"isRef"</span> <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-conid'>OpDef'ArgDef</span>
+<a name="line-303"></a>         <span class='hs-keyword'>where</span>
+<a name="line-304"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-305"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'ArgDef'isRef</span>
+<a name="line-306"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'ArgDef'isRef</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-307"></a>
+<a name="line-308"></a><a name="instance%20Data.Default.Class.Default%20OpDef'ArgDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-keyword'>where</span>
+<a name="line-309"></a>        <span class='hs-varid'>def</span>
+<a name="line-310"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OpDef'ArgDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'ArgDef'name</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-311"></a>                         <span class='hs-sel'>_OpDef'ArgDef'description</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-312"></a>                         <span class='hs-sel'>_OpDef'ArgDef'type'</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>,</span>
+<a name="line-313"></a>                         <span class='hs-sel'>_OpDef'ArgDef'typeAttr</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-314"></a>                         <span class='hs-sel'>_OpDef'ArgDef'numberAttr</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-315"></a>                         <span class='hs-sel'>_OpDef'ArgDef'typeListAttr</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-316"></a>                         <span class='hs-sel'>_OpDef'ArgDef'isRef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-317"></a>
+<a name="line-318"></a><a name="instance%20Data.ProtoLens.Message%20OpDef'ArgDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>OpDef'ArgDef</span> <span class='hs-keyword'>where</span>
+<a name="line-319"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-320"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>name__field_descriptor</span>
+<a name="line-321"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"name"</span>
+<a name="line-322"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-323"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-324"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span>
+<a name="line-325"></a>                <span class='hs-varid'>description__field_descriptor</span>
+<a name="line-326"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"description"</span>
+<a name="line-327"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-328"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-329"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>description</span><span class='hs-layout'>)</span>
+<a name="line-330"></a>                <span class='hs-varid'>type'__field_descriptor</span>
+<a name="line-331"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"type"</span>
+<a name="line-332"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>EnumField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-333"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-334"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span><span class='hs-layout'>)</span>
+<a name="line-335"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>type'</span><span class='hs-layout'>)</span>
+<a name="line-336"></a>                <span class='hs-varid'>typeAttr__field_descriptor</span>
+<a name="line-337"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"type_attr"</span>
+<a name="line-338"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-339"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-340"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>typeAttr</span><span class='hs-layout'>)</span>
+<a name="line-341"></a>                <span class='hs-varid'>numberAttr__field_descriptor</span>
+<a name="line-342"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"number_attr"</span>
+<a name="line-343"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-344"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-345"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>numberAttr</span><span class='hs-layout'>)</span>
+<a name="line-346"></a>                <span class='hs-varid'>typeListAttr__field_descriptor</span>
+<a name="line-347"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"type_list_attr"</span>
+<a name="line-348"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-349"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-350"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>typeListAttr</span><span class='hs-layout'>)</span>
+<a name="line-351"></a>                <span class='hs-varid'>isRef__field_descriptor</span>
+<a name="line-352"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"is_ref"</span>
+<a name="line-353"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-354"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-355"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>isRef</span><span class='hs-layout'>)</span>
+<a name="line-356"></a>              <span class='hs-keyword'>in</span>
+<a name="line-357"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-358"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-359"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>name__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-360"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>description__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-361"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>type'__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-362"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>typeAttr__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-363"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>5</span><span class='hs-layout'>,</span> <span class='hs-varid'>numberAttr__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-364"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>6</span><span class='hs-layout'>,</span> <span class='hs-varid'>typeListAttr__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-365"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>16</span><span class='hs-layout'>,</span> <span class='hs-varid'>isRef__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-366"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-367"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"name"</span><span class='hs-layout'>,</span> <span class='hs-varid'>name__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-368"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"description"</span><span class='hs-layout'>,</span> <span class='hs-varid'>description__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-369"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"type"</span><span class='hs-layout'>,</span> <span class='hs-varid'>type'__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-370"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"type_attr"</span><span class='hs-layout'>,</span> <span class='hs-varid'>typeAttr__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-371"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"number_attr"</span><span class='hs-layout'>,</span> <span class='hs-varid'>numberAttr__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-372"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"type_list_attr"</span><span class='hs-layout'>,</span> <span class='hs-varid'>typeListAttr__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-373"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"is_ref"</span><span class='hs-layout'>,</span> <span class='hs-varid'>isRef__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-374"></a>
+<a name="line-375"></a><a name="OpDef'AttrDef"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OpDef'AttrDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'AttrDef'name</span> <span class='hs-keyglyph'>::</span>
+<a name="line-376"></a>                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-377"></a>                                   <span class='hs-sel'>_OpDef'AttrDef'type'</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-378"></a>                                   <span class='hs-sel'>_OpDef'AttrDef'defaultValue</span> <span class='hs-keyglyph'>::</span>
+<a name="line-379"></a>                                   <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-380"></a>                                     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-layout'>,</span>
+<a name="line-381"></a>                                   <span class='hs-sel'>_OpDef'AttrDef'description</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-382"></a>                                   <span class='hs-sel'>_OpDef'AttrDef'hasMinimum</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-383"></a>                                   <span class='hs-sel'>_OpDef'AttrDef'minimum</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-384"></a>                                   <span class='hs-sel'>_OpDef'AttrDef'allowedValues</span> <span class='hs-keyglyph'>::</span>
+<a name="line-385"></a>                                   <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-386"></a>                                     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-layout'>}</span>
+<a name="line-387"></a>                   <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-388"></a>
+<a name="line-389"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-390"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-391"></a>
+<a name="line-392"></a><a name="instance%20Data.ProtoLens.HasField%20%22name%22%20OpDef'AttrDef%20OpDef'AttrDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"name"</span> <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-conid'>OpDef'AttrDef</span>
+<a name="line-393"></a>         <span class='hs-keyword'>where</span>
+<a name="line-394"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-395"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'AttrDef'name</span>
+<a name="line-396"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'AttrDef'name</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-397"></a>
+<a name="line-398"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"type'"</span> <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-399"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-400"></a>
+<a name="line-401"></a><a name="instance%20Data.ProtoLens.HasField%20%22type'%22%20OpDef'AttrDef%20OpDef'AttrDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"type'"</span> <span class='hs-conid'>OpDef'AttrDef</span>
+<a name="line-402"></a>         <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyword'>where</span>
+<a name="line-403"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-404"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'AttrDef'type'</span>
+<a name="line-405"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'AttrDef'type'</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-406"></a>
+<a name="line-407"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"defaultValue"</span> <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-408"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span>
+<a name="line-409"></a>
+<a name="line-410"></a><a name="instance%20Data.ProtoLens.HasField%20%22defaultValue%22%20OpDef'AttrDef%20OpDef'AttrDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"defaultValue"</span> <span class='hs-conid'>OpDef'AttrDef</span>
+<a name="line-411"></a>         <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyword'>where</span>
+<a name="line-412"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-413"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'defaultValue</span>
+<a name="line-414"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-415"></a>
+<a name="line-416"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-417"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'defaultValue"</span> <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-418"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span>
+<a name="line-419"></a>
+<a name="line-420"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'defaultValue%22%20OpDef'AttrDef%20OpDef'AttrDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'defaultValue"</span> <span class='hs-conid'>OpDef'AttrDef</span>
+<a name="line-421"></a>         <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyword'>where</span>
+<a name="line-422"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-423"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'AttrDef'defaultValue</span>
+<a name="line-424"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'AttrDef'defaultValue</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-425"></a>
+<a name="line-426"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"description"</span> <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-427"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-428"></a>
+<a name="line-429"></a><a name="instance%20Data.ProtoLens.HasField%20%22description%22%20OpDef'AttrDef%20OpDef'AttrDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"description"</span> <span class='hs-conid'>OpDef'AttrDef</span>
+<a name="line-430"></a>         <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyword'>where</span>
+<a name="line-431"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-432"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'AttrDef'description</span>
+<a name="line-433"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'AttrDef'description</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-434"></a>
+<a name="line-435"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"hasMinimum"</span> <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-436"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-437"></a>
+<a name="line-438"></a><a name="instance%20Data.ProtoLens.HasField%20%22hasMinimum%22%20OpDef'AttrDef%20OpDef'AttrDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"hasMinimum"</span> <span class='hs-conid'>OpDef'AttrDef</span>
+<a name="line-439"></a>         <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyword'>where</span>
+<a name="line-440"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-441"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'AttrDef'hasMinimum</span>
+<a name="line-442"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'AttrDef'hasMinimum</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-443"></a>
+<a name="line-444"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"minimum"</span> <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-445"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-446"></a>
+<a name="line-447"></a><a name="instance%20Data.ProtoLens.HasField%20%22minimum%22%20OpDef'AttrDef%20OpDef'AttrDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"minimum"</span> <span class='hs-conid'>OpDef'AttrDef</span>
+<a name="line-448"></a>         <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyword'>where</span>
+<a name="line-449"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-450"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'AttrDef'minimum</span>
+<a name="line-451"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'AttrDef'minimum</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-452"></a>
+<a name="line-453"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allowedValues"</span> <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-454"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span>
+<a name="line-455"></a>
+<a name="line-456"></a><a name="instance%20Data.ProtoLens.HasField%20%22allowedValues%22%20OpDef'AttrDef%20OpDef'AttrDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allowedValues"</span> <span class='hs-conid'>OpDef'AttrDef</span>
+<a name="line-457"></a>         <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyword'>where</span>
+<a name="line-458"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-459"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'allowedValues</span>
+<a name="line-460"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-461"></a>
+<a name="line-462"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-463"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'allowedValues"</span> <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-464"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span>
+<a name="line-465"></a>
+<a name="line-466"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'allowedValues%22%20OpDef'AttrDef%20OpDef'AttrDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'allowedValues"</span>
+<a name="line-467"></a>         <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyword'>where</span>
+<a name="line-468"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-469"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDef'AttrDef'allowedValues</span>
+<a name="line-470"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'AttrDef'allowedValues</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-471"></a>
+<a name="line-472"></a><a name="instance%20Data.Default.Class.Default%20OpDef'AttrDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyword'>where</span>
+<a name="line-473"></a>        <span class='hs-varid'>def</span>
+<a name="line-474"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OpDef'AttrDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDef'AttrDef'name</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-475"></a>                          <span class='hs-sel'>_OpDef'AttrDef'type'</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-476"></a>                          <span class='hs-sel'>_OpDef'AttrDef'defaultValue</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-477"></a>                          <span class='hs-sel'>_OpDef'AttrDef'description</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-478"></a>                          <span class='hs-sel'>_OpDef'AttrDef'hasMinimum</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-479"></a>                          <span class='hs-sel'>_OpDef'AttrDef'minimum</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-480"></a>                          <span class='hs-sel'>_OpDef'AttrDef'allowedValues</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>}</span>
+<a name="line-481"></a>
+<a name="line-482"></a><a name="instance%20Data.ProtoLens.Message%20OpDef'AttrDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>OpDef'AttrDef</span> <span class='hs-keyword'>where</span>
+<a name="line-483"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-484"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>name__field_descriptor</span>
+<a name="line-485"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"name"</span>
+<a name="line-486"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-487"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-488"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span>
+<a name="line-489"></a>                <span class='hs-varid'>type'__field_descriptor</span>
+<a name="line-490"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"type"</span>
+<a name="line-491"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-492"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-493"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>type'</span><span class='hs-layout'>)</span>
+<a name="line-494"></a>                <span class='hs-varid'>defaultValue__field_descriptor</span>
+<a name="line-495"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"default_value"</span>
+<a name="line-496"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-497"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-498"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-layout'>)</span>
+<a name="line-499"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'defaultValue</span><span class='hs-layout'>)</span>
+<a name="line-500"></a>                <span class='hs-varid'>description__field_descriptor</span>
+<a name="line-501"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"description"</span>
+<a name="line-502"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-503"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-504"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>description</span><span class='hs-layout'>)</span>
+<a name="line-505"></a>                <span class='hs-varid'>hasMinimum__field_descriptor</span>
+<a name="line-506"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"has_minimum"</span>
+<a name="line-507"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-508"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-509"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>hasMinimum</span><span class='hs-layout'>)</span>
+<a name="line-510"></a>                <span class='hs-varid'>minimum__field_descriptor</span>
+<a name="line-511"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"minimum"</span>
+<a name="line-512"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-513"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-514"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>minimum</span><span class='hs-layout'>)</span>
+<a name="line-515"></a>                <span class='hs-varid'>allowedValues__field_descriptor</span>
+<a name="line-516"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"allowed_values"</span>
+<a name="line-517"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-518"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-519"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-varop'>.</span><span class='hs-conid'>AttrValue</span><span class='hs-layout'>)</span>
+<a name="line-520"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'allowedValues</span><span class='hs-layout'>)</span>
+<a name="line-521"></a>              <span class='hs-keyword'>in</span>
+<a name="line-522"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-523"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-524"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>name__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-525"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>type'__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-526"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>defaultValue__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-527"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>description__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-528"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>5</span><span class='hs-layout'>,</span> <span class='hs-varid'>hasMinimum__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-529"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>6</span><span class='hs-layout'>,</span> <span class='hs-varid'>minimum__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-530"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>7</span><span class='hs-layout'>,</span> <span class='hs-varid'>allowedValues__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-531"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-532"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"name"</span><span class='hs-layout'>,</span> <span class='hs-varid'>name__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-533"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"type"</span><span class='hs-layout'>,</span> <span class='hs-varid'>type'__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-534"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"default_value"</span><span class='hs-layout'>,</span> <span class='hs-varid'>defaultValue__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-535"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"description"</span><span class='hs-layout'>,</span> <span class='hs-varid'>description__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-536"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"has_minimum"</span><span class='hs-layout'>,</span> <span class='hs-varid'>hasMinimum__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-537"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"minimum"</span><span class='hs-layout'>,</span> <span class='hs-varid'>minimum__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-538"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"allowed_values"</span><span class='hs-layout'>,</span> <span class='hs-varid'>allowedValues__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-539"></a>
+<a name="line-540"></a><a name="OpDeprecation"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>OpDeprecation</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OpDeprecation</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDeprecation'version</span> <span class='hs-keyglyph'>::</span>
+<a name="line-541"></a>                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-542"></a>                                   <span class='hs-sel'>_OpDeprecation'explanation</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>}</span>
+<a name="line-543"></a>                   <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-544"></a>
+<a name="line-545"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"version"</span> <span class='hs-conid'>OpDeprecation</span> <span class='hs-keyglyph'>=</span>
+<a name="line-546"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-547"></a>
+<a name="line-548"></a><a name="instance%20Data.ProtoLens.HasField%20%22version%22%20OpDeprecation%20OpDeprecation"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"version"</span> <span class='hs-conid'>OpDeprecation</span>
+<a name="line-549"></a>         <span class='hs-conid'>OpDeprecation</span> <span class='hs-keyword'>where</span>
+<a name="line-550"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-551"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDeprecation'version</span>
+<a name="line-552"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDeprecation'version</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-553"></a>
+<a name="line-554"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"explanation"</span> <span class='hs-conid'>OpDeprecation</span> <span class='hs-keyglyph'>=</span>
+<a name="line-555"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-556"></a>
+<a name="line-557"></a><a name="instance%20Data.ProtoLens.HasField%20%22explanation%22%20OpDeprecation%20OpDeprecation"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"explanation"</span> <span class='hs-conid'>OpDeprecation</span>
+<a name="line-558"></a>         <span class='hs-conid'>OpDeprecation</span> <span class='hs-keyword'>where</span>
+<a name="line-559"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-560"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpDeprecation'explanation</span>
+<a name="line-561"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDeprecation'explanation</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-562"></a>
+<a name="line-563"></a><a name="instance%20Data.Default.Class.Default%20OpDeprecation"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>OpDeprecation</span> <span class='hs-keyword'>where</span>
+<a name="line-564"></a>        <span class='hs-varid'>def</span>
+<a name="line-565"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OpDeprecation</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpDeprecation'version</span> <span class='hs-keyglyph'>=</span>
+<a name="line-566"></a>                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-567"></a>                          <span class='hs-sel'>_OpDeprecation'explanation</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-568"></a>
+<a name="line-569"></a><a name="instance%20Data.ProtoLens.Message%20OpDeprecation"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>OpDeprecation</span> <span class='hs-keyword'>where</span>
+<a name="line-570"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-571"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>version__field_descriptor</span>
+<a name="line-572"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"version"</span>
+<a name="line-573"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-574"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-575"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>version</span><span class='hs-layout'>)</span>
+<a name="line-576"></a>                <span class='hs-varid'>explanation__field_descriptor</span>
+<a name="line-577"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"explanation"</span>
+<a name="line-578"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-579"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-580"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>explanation</span><span class='hs-layout'>)</span>
+<a name="line-581"></a>              <span class='hs-keyword'>in</span>
+<a name="line-582"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-583"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-584"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>version__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-585"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>explanation__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-586"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-587"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"version"</span><span class='hs-layout'>,</span> <span class='hs-varid'>version__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-588"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"explanation"</span><span class='hs-layout'>,</span> <span class='hs-varid'>explanation__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-589"></a>
+<a name="line-590"></a><a name="OpList"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>OpList</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OpList</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpList'op</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>OpDef</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>}</span>
+<a name="line-591"></a>            <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-592"></a>
+<a name="line-593"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"op"</span> <span class='hs-conid'>OpList</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>OpDef</span><span class='hs-keyglyph'>]</span>
+<a name="line-594"></a>
+<a name="line-595"></a><a name="instance%20Data.ProtoLens.HasField%20%22op%22%20OpList%20OpList"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"op"</span> <span class='hs-conid'>OpList</span> <span class='hs-conid'>OpList</span> <span class='hs-keyword'>where</span>
+<a name="line-596"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-597"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OpList'op</span>
+<a name="line-598"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpList'op</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-599"></a>
+<a name="line-600"></a><a name="instance%20Data.Default.Class.Default%20OpList"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>OpList</span> <span class='hs-keyword'>where</span>
+<a name="line-601"></a>        <span class='hs-varid'>def</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OpList</span><span class='hs-layout'>{</span><span class='hs-sel'>_OpList'op</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>}</span>
+<a name="line-602"></a>
+<a name="line-603"></a><a name="instance%20Data.ProtoLens.Message%20OpList"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>OpList</span> <span class='hs-keyword'>where</span>
+<a name="line-604"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-605"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>op__field_descriptor</span>
+<a name="line-606"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"op"</span>
+<a name="line-607"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-608"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>OpDef</span><span class='hs-layout'>)</span>
+<a name="line-609"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>op</span><span class='hs-layout'>)</span>
+<a name="line-610"></a>              <span class='hs-keyword'>in</span>
+<a name="line-611"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-612"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>op__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-613"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"op"</span><span class='hs-layout'>,</span> <span class='hs-varid'>op__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-614"></a>
+<a name="line-615"></a><a name="allowedValues"></a><span class='hs-definition'>allowedValues</span> <span class='hs-keyglyph'>::</span>
+<a name="line-616"></a>              <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-617"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allowedValues"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-618"></a>                <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-619"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allowedValues"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-620"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allowedValues"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-621"></a><span class='hs-definition'>allowedValues</span>
+<a name="line-622"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-623"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-624"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"allowedValues"</span><span class='hs-layout'>)</span>
+<a name="line-625"></a>
+<a name="line-626"></a><a name="allowsUninitializedInput"></a><span class='hs-definition'>allowsUninitializedInput</span> <span class='hs-keyglyph'>::</span>
+<a name="line-627"></a>                         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-628"></a>                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allowsUninitializedInput"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-629"></a>                           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-630"></a>                             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allowsUninitializedInput"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-631"></a>                             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allowsUninitializedInput"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-632"></a><span class='hs-definition'>allowsUninitializedInput</span>
+<a name="line-633"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-634"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-635"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"allowsUninitializedInput"</span><span class='hs-layout'>)</span>
+<a name="line-636"></a>
+<a name="line-637"></a><a name="attr"></a><span class='hs-definition'>attr</span> <span class='hs-keyglyph'>::</span>
+<a name="line-638"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"attr"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-639"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"attr"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-640"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"attr"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-641"></a><span class='hs-definition'>attr</span>
+<a name="line-642"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-643"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"attr"</span><span class='hs-layout'>)</span>
+<a name="line-644"></a>
+<a name="line-645"></a><a name="defaultValue"></a><span class='hs-definition'>defaultValue</span> <span class='hs-keyglyph'>::</span>
+<a name="line-646"></a>             <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-647"></a>               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"defaultValue"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-648"></a>               <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-649"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"defaultValue"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-650"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"defaultValue"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-651"></a><span class='hs-definition'>defaultValue</span>
+<a name="line-652"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-653"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"defaultValue"</span><span class='hs-layout'>)</span>
+<a name="line-654"></a>
+<a name="line-655"></a><a name="deprecation"></a><span class='hs-definition'>deprecation</span> <span class='hs-keyglyph'>::</span>
+<a name="line-656"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"deprecation"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-657"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"deprecation"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-658"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"deprecation"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-659"></a><span class='hs-definition'>deprecation</span>
+<a name="line-660"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-661"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"deprecation"</span><span class='hs-layout'>)</span>
+<a name="line-662"></a>
+<a name="line-663"></a><a name="description"></a><span class='hs-definition'>description</span> <span class='hs-keyglyph'>::</span>
+<a name="line-664"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"description"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-665"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"description"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-666"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"description"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-667"></a><span class='hs-definition'>description</span>
+<a name="line-668"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-669"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"description"</span><span class='hs-layout'>)</span>
+<a name="line-670"></a>
+<a name="line-671"></a><a name="explanation"></a><span class='hs-definition'>explanation</span> <span class='hs-keyglyph'>::</span>
+<a name="line-672"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"explanation"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-673"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"explanation"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-674"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"explanation"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-675"></a><span class='hs-definition'>explanation</span>
+<a name="line-676"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-677"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"explanation"</span><span class='hs-layout'>)</span>
+<a name="line-678"></a>
+<a name="line-679"></a><a name="hasMinimum"></a><span class='hs-definition'>hasMinimum</span> <span class='hs-keyglyph'>::</span>
+<a name="line-680"></a>           <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"hasMinimum"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-681"></a>             <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"hasMinimum"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-682"></a>               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"hasMinimum"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-683"></a><span class='hs-definition'>hasMinimum</span>
+<a name="line-684"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-685"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"hasMinimum"</span><span class='hs-layout'>)</span>
+<a name="line-686"></a>
+<a name="line-687"></a><a name="inputArg"></a><span class='hs-definition'>inputArg</span> <span class='hs-keyglyph'>::</span>
+<a name="line-688"></a>         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"inputArg"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-689"></a>           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"inputArg"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-690"></a>             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"inputArg"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-691"></a><span class='hs-definition'>inputArg</span>
+<a name="line-692"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-693"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"inputArg"</span><span class='hs-layout'>)</span>
+<a name="line-694"></a>
+<a name="line-695"></a><a name="isAggregate"></a><span class='hs-definition'>isAggregate</span> <span class='hs-keyglyph'>::</span>
+<a name="line-696"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"isAggregate"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-697"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"isAggregate"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-698"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"isAggregate"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-699"></a><span class='hs-definition'>isAggregate</span>
+<a name="line-700"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-701"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"isAggregate"</span><span class='hs-layout'>)</span>
+<a name="line-702"></a>
+<a name="line-703"></a><a name="isCommutative"></a><span class='hs-definition'>isCommutative</span> <span class='hs-keyglyph'>::</span>
+<a name="line-704"></a>              <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-705"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"isCommutative"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-706"></a>                <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-707"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"isCommutative"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-708"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"isCommutative"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-709"></a><span class='hs-definition'>isCommutative</span>
+<a name="line-710"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-711"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-712"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"isCommutative"</span><span class='hs-layout'>)</span>
+<a name="line-713"></a>
+<a name="line-714"></a><a name="isRef"></a><span class='hs-definition'>isRef</span> <span class='hs-keyglyph'>::</span>
+<a name="line-715"></a>      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"isRef"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-716"></a>        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"isRef"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-717"></a>          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"isRef"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-718"></a><span class='hs-definition'>isRef</span>
+<a name="line-719"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-720"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"isRef"</span><span class='hs-layout'>)</span>
+<a name="line-721"></a>
+<a name="line-722"></a><a name="isStateful"></a><span class='hs-definition'>isStateful</span> <span class='hs-keyglyph'>::</span>
+<a name="line-723"></a>           <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"isStateful"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-724"></a>             <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"isStateful"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-725"></a>               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"isStateful"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-726"></a><span class='hs-definition'>isStateful</span>
+<a name="line-727"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-728"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"isStateful"</span><span class='hs-layout'>)</span>
+<a name="line-729"></a>
+<a name="line-730"></a><a name="maybe'allowedValues"></a><span class='hs-definition'>maybe'allowedValues</span> <span class='hs-keyglyph'>::</span>
+<a name="line-731"></a>                    <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-732"></a>                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'allowedValues"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-733"></a>                      <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-734"></a>                        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'allowedValues"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-735"></a>                        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'allowedValues"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-736"></a><span class='hs-definition'>maybe'allowedValues</span>
+<a name="line-737"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-738"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-739"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'allowedValues"</span><span class='hs-layout'>)</span>
+<a name="line-740"></a>
+<a name="line-741"></a><a name="maybe'defaultValue"></a><span class='hs-definition'>maybe'defaultValue</span> <span class='hs-keyglyph'>::</span>
+<a name="line-742"></a>                   <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-743"></a>                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'defaultValue"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-744"></a>                     <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-745"></a>                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'defaultValue"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-746"></a>                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'defaultValue"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-747"></a><span class='hs-definition'>maybe'defaultValue</span>
+<a name="line-748"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-749"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-750"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'defaultValue"</span><span class='hs-layout'>)</span>
+<a name="line-751"></a>
+<a name="line-752"></a><a name="maybe'deprecation"></a><span class='hs-definition'>maybe'deprecation</span> <span class='hs-keyglyph'>::</span>
+<a name="line-753"></a>                  <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-754"></a>                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'deprecation"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-755"></a>                    <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-756"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'deprecation"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-757"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'deprecation"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-758"></a><span class='hs-definition'>maybe'deprecation</span>
+<a name="line-759"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-760"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-761"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'deprecation"</span><span class='hs-layout'>)</span>
+<a name="line-762"></a>
+<a name="line-763"></a><a name="minimum"></a><span class='hs-definition'>minimum</span> <span class='hs-keyglyph'>::</span>
+<a name="line-764"></a>        <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"minimum"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-765"></a>          <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"minimum"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-766"></a>            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"minimum"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-767"></a><span class='hs-definition'>minimum</span>
+<a name="line-768"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-769"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"minimum"</span><span class='hs-layout'>)</span>
+<a name="line-770"></a>
+<a name="line-771"></a><a name="name"></a><span class='hs-definition'>name</span> <span class='hs-keyglyph'>::</span>
+<a name="line-772"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-773"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-774"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-775"></a><span class='hs-definition'>name</span>
+<a name="line-776"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-777"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"name"</span><span class='hs-layout'>)</span>
+<a name="line-778"></a>
+<a name="line-779"></a><a name="numberAttr"></a><span class='hs-definition'>numberAttr</span> <span class='hs-keyglyph'>::</span>
+<a name="line-780"></a>           <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"numberAttr"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-781"></a>             <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"numberAttr"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-782"></a>               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"numberAttr"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-783"></a><span class='hs-definition'>numberAttr</span>
+<a name="line-784"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-785"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"numberAttr"</span><span class='hs-layout'>)</span>
+<a name="line-786"></a>
+<a name="line-787"></a><a name="op"></a><span class='hs-definition'>op</span> <span class='hs-keyglyph'>::</span>
+<a name="line-788"></a>   <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"op"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-789"></a>     <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"op"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-790"></a>       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"op"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-791"></a><span class='hs-definition'>op</span>
+<a name="line-792"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-793"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"op"</span><span class='hs-layout'>)</span>
+<a name="line-794"></a>
+<a name="line-795"></a><a name="outputArg"></a><span class='hs-definition'>outputArg</span> <span class='hs-keyglyph'>::</span>
+<a name="line-796"></a>          <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"outputArg"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-797"></a>            <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"outputArg"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-798"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"outputArg"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-799"></a><span class='hs-definition'>outputArg</span>
+<a name="line-800"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-801"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"outputArg"</span><span class='hs-layout'>)</span>
+<a name="line-802"></a>
+<a name="line-803"></a><a name="summary"></a><span class='hs-definition'>summary</span> <span class='hs-keyglyph'>::</span>
+<a name="line-804"></a>        <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"summary"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-805"></a>          <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"summary"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-806"></a>            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"summary"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-807"></a><span class='hs-definition'>summary</span>
+<a name="line-808"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-809"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"summary"</span><span class='hs-layout'>)</span>
+<a name="line-810"></a>
+<a name="line-811"></a><a name="type'"></a><span class='hs-definition'>type'</span> <span class='hs-keyglyph'>::</span>
+<a name="line-812"></a>      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"type'"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-813"></a>        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"type'"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-814"></a>          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"type'"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-815"></a><span class='hs-definition'>type'</span>
+<a name="line-816"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-817"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"type'"</span><span class='hs-layout'>)</span>
+<a name="line-818"></a>
+<a name="line-819"></a><a name="typeAttr"></a><span class='hs-definition'>typeAttr</span> <span class='hs-keyglyph'>::</span>
+<a name="line-820"></a>         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"typeAttr"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-821"></a>           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"typeAttr"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-822"></a>             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"typeAttr"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-823"></a><span class='hs-definition'>typeAttr</span>
+<a name="line-824"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-825"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"typeAttr"</span><span class='hs-layout'>)</span>
+<a name="line-826"></a>
+<a name="line-827"></a><a name="typeListAttr"></a><span class='hs-definition'>typeListAttr</span> <span class='hs-keyglyph'>::</span>
+<a name="line-828"></a>             <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-829"></a>               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"typeListAttr"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-830"></a>               <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-831"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"typeListAttr"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-832"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"typeListAttr"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-833"></a><span class='hs-definition'>typeListAttr</span>
+<a name="line-834"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-835"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"typeListAttr"</span><span class='hs-layout'>)</span>
+<a name="line-836"></a>
+<a name="line-837"></a><a name="version"></a><span class='hs-definition'>version</span> <span class='hs-keyglyph'>::</span>
+<a name="line-838"></a>        <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"version"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-839"></a>          <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"version"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-840"></a>            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"version"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-841"></a><span class='hs-definition'>version</span>
+<a name="line-842"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-843"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"version"</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-ResourceHandle.html b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-ResourceHandle.html
new file mode 100644
index 0000000..8d2cde7
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-ResourceHandle.html
@@ -0,0 +1,182 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Proto/Tensorflow/Core/Framework/ResourceHandle.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>{- This file was auto-generated from tensorflow/core/framework/resource_handle.proto by the proto-lens-protoc program. -}</span>
+<a name="line-2"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables, DataKinds, TypeFamilies,
+<a name="line-3"></a>  MultiParamTypeClasses, FlexibleContexts, FlexibleInstances,
+<a name="line-4"></a>  PatternSynonyms #-}</span>
+<a name="line-5"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-unused-imports#-}</span>
+<a name="line-6"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>ResourceHandle</span> <span class='hs-keyword'>where</span>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Prelude</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-11"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-12"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-13"></a>       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-14"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-15"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-16"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-17"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-18"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-19"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-20"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-24"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-25"></a>
+<a name="line-26"></a><a name="ResourceHandle"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>ResourceHandle</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>ResourceHandle</span><span class='hs-layout'>{</span><span class='hs-sel'>_ResourceHandle'device</span> <span class='hs-keyglyph'>::</span>
+<a name="line-27"></a>                                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-28"></a>                                     <span class='hs-sel'>_ResourceHandle'container</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-29"></a>                                     <span class='hs-sel'>_ResourceHandle'name</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-30"></a>                                     <span class='hs-sel'>_ResourceHandle'hashCode</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word64</span><span class='hs-layout'>,</span>
+<a name="line-31"></a>                                     <span class='hs-sel'>_ResourceHandle'maybeTypeName</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>}</span>
+<a name="line-32"></a>                    <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-33"></a>
+<a name="line-34"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"device"</span> <span class='hs-conid'>ResourceHandle</span> <span class='hs-keyglyph'>=</span>
+<a name="line-35"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-36"></a>
+<a name="line-37"></a><a name="instance%20Data.ProtoLens.HasField%20%22device%22%20ResourceHandle%20ResourceHandle"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"device"</span> <span class='hs-conid'>ResourceHandle</span>
+<a name="line-38"></a>         <span class='hs-conid'>ResourceHandle</span> <span class='hs-keyword'>where</span>
+<a name="line-39"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-40"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ResourceHandle'device</span>
+<a name="line-41"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ResourceHandle'device</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-42"></a>
+<a name="line-43"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"container"</span> <span class='hs-conid'>ResourceHandle</span> <span class='hs-keyglyph'>=</span>
+<a name="line-44"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-45"></a>
+<a name="line-46"></a><a name="instance%20Data.ProtoLens.HasField%20%22container%22%20ResourceHandle%20ResourceHandle"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"container"</span> <span class='hs-conid'>ResourceHandle</span>
+<a name="line-47"></a>         <span class='hs-conid'>ResourceHandle</span> <span class='hs-keyword'>where</span>
+<a name="line-48"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-49"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ResourceHandle'container</span>
+<a name="line-50"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ResourceHandle'container</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-51"></a>
+<a name="line-52"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-conid'>ResourceHandle</span> <span class='hs-keyglyph'>=</span>
+<a name="line-53"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-54"></a>
+<a name="line-55"></a><a name="instance%20Data.ProtoLens.HasField%20%22name%22%20ResourceHandle%20ResourceHandle"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"name"</span> <span class='hs-conid'>ResourceHandle</span>
+<a name="line-56"></a>         <span class='hs-conid'>ResourceHandle</span> <span class='hs-keyword'>where</span>
+<a name="line-57"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-58"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ResourceHandle'name</span>
+<a name="line-59"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ResourceHandle'name</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-60"></a>
+<a name="line-61"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"hashCode"</span> <span class='hs-conid'>ResourceHandle</span> <span class='hs-keyglyph'>=</span>
+<a name="line-62"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word64</span>
+<a name="line-63"></a>
+<a name="line-64"></a><a name="instance%20Data.ProtoLens.HasField%20%22hashCode%22%20ResourceHandle%20ResourceHandle"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"hashCode"</span> <span class='hs-conid'>ResourceHandle</span>
+<a name="line-65"></a>         <span class='hs-conid'>ResourceHandle</span> <span class='hs-keyword'>where</span>
+<a name="line-66"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-67"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ResourceHandle'hashCode</span>
+<a name="line-68"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ResourceHandle'hashCode</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-69"></a>
+<a name="line-70"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybeTypeName"</span> <span class='hs-conid'>ResourceHandle</span> <span class='hs-keyglyph'>=</span>
+<a name="line-71"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-72"></a>
+<a name="line-73"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybeTypeName%22%20ResourceHandle%20ResourceHandle"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybeTypeName"</span> <span class='hs-conid'>ResourceHandle</span>
+<a name="line-74"></a>         <span class='hs-conid'>ResourceHandle</span> <span class='hs-keyword'>where</span>
+<a name="line-75"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-76"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ResourceHandle'maybeTypeName</span>
+<a name="line-77"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ResourceHandle'maybeTypeName</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-78"></a>
+<a name="line-79"></a><a name="instance%20Data.Default.Class.Default%20ResourceHandle"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>ResourceHandle</span> <span class='hs-keyword'>where</span>
+<a name="line-80"></a>        <span class='hs-varid'>def</span>
+<a name="line-81"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>ResourceHandle</span><span class='hs-layout'>{</span><span class='hs-sel'>_ResourceHandle'device</span> <span class='hs-keyglyph'>=</span>
+<a name="line-82"></a>                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-83"></a>                           <span class='hs-sel'>_ResourceHandle'container</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-84"></a>                           <span class='hs-sel'>_ResourceHandle'name</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-85"></a>                           <span class='hs-sel'>_ResourceHandle'hashCode</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-86"></a>                           <span class='hs-sel'>_ResourceHandle'maybeTypeName</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-87"></a>
+<a name="line-88"></a><a name="instance%20Data.ProtoLens.Message%20ResourceHandle"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>ResourceHandle</span> <span class='hs-keyword'>where</span>
+<a name="line-89"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-90"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>device__field_descriptor</span>
+<a name="line-91"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"device"</span>
+<a name="line-92"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-93"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-94"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>device</span><span class='hs-layout'>)</span>
+<a name="line-95"></a>                <span class='hs-varid'>container__field_descriptor</span>
+<a name="line-96"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"container"</span>
+<a name="line-97"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-98"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-99"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>container</span><span class='hs-layout'>)</span>
+<a name="line-100"></a>                <span class='hs-varid'>name__field_descriptor</span>
+<a name="line-101"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"name"</span>
+<a name="line-102"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-103"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-104"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span>
+<a name="line-105"></a>                <span class='hs-varid'>hashCode__field_descriptor</span>
+<a name="line-106"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"hash_code"</span>
+<a name="line-107"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>UInt64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-108"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word64</span><span class='hs-layout'>)</span>
+<a name="line-109"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>hashCode</span><span class='hs-layout'>)</span>
+<a name="line-110"></a>                <span class='hs-varid'>maybeTypeName__field_descriptor</span>
+<a name="line-111"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"maybe_type_name"</span>
+<a name="line-112"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-113"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-114"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>maybeTypeName</span><span class='hs-layout'>)</span>
+<a name="line-115"></a>              <span class='hs-keyword'>in</span>
+<a name="line-116"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-117"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-118"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>device__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-119"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>container__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-120"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>name__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-121"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>hashCode__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-122"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>5</span><span class='hs-layout'>,</span> <span class='hs-varid'>maybeTypeName__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-123"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-124"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"device"</span><span class='hs-layout'>,</span> <span class='hs-varid'>device__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-125"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"container"</span><span class='hs-layout'>,</span> <span class='hs-varid'>container__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-126"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"name"</span><span class='hs-layout'>,</span> <span class='hs-varid'>name__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-127"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"hash_code"</span><span class='hs-layout'>,</span> <span class='hs-varid'>hashCode__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-128"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"maybe_type_name"</span><span class='hs-layout'>,</span> <span class='hs-varid'>maybeTypeName__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-129"></a>
+<a name="line-130"></a><a name="container"></a><span class='hs-definition'>container</span> <span class='hs-keyglyph'>::</span>
+<a name="line-131"></a>          <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"container"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-132"></a>            <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"container"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-133"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"container"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-134"></a><span class='hs-definition'>container</span>
+<a name="line-135"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-136"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"container"</span><span class='hs-layout'>)</span>
+<a name="line-137"></a>
+<a name="line-138"></a><a name="device"></a><span class='hs-definition'>device</span> <span class='hs-keyglyph'>::</span>
+<a name="line-139"></a>       <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"device"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-140"></a>         <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"device"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-141"></a>           <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"device"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-142"></a><span class='hs-definition'>device</span>
+<a name="line-143"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-144"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"device"</span><span class='hs-layout'>)</span>
+<a name="line-145"></a>
+<a name="line-146"></a><a name="hashCode"></a><span class='hs-definition'>hashCode</span> <span class='hs-keyglyph'>::</span>
+<a name="line-147"></a>         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"hashCode"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-148"></a>           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"hashCode"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-149"></a>             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"hashCode"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-150"></a><span class='hs-definition'>hashCode</span>
+<a name="line-151"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-152"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"hashCode"</span><span class='hs-layout'>)</span>
+<a name="line-153"></a>
+<a name="line-154"></a><a name="maybeTypeName"></a><span class='hs-definition'>maybeTypeName</span> <span class='hs-keyglyph'>::</span>
+<a name="line-155"></a>              <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-156"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybeTypeName"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-157"></a>                <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-158"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybeTypeName"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-159"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybeTypeName"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-160"></a><span class='hs-definition'>maybeTypeName</span>
+<a name="line-161"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-162"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-163"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybeTypeName"</span><span class='hs-layout'>)</span>
+<a name="line-164"></a>
+<a name="line-165"></a><a name="name"></a><span class='hs-definition'>name</span> <span class='hs-keyglyph'>::</span>
+<a name="line-166"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-167"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-168"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-169"></a><span class='hs-definition'>name</span>
+<a name="line-170"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-171"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"name"</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-StepStats.html b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-StepStats.html
new file mode 100644
index 0000000..5fc4c9a
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-StepStats.html
@@ -0,0 +1,653 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Proto/Tensorflow/Core/Framework/StepStats.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>{- This file was auto-generated from tensorflow/core/framework/step_stats.proto by the proto-lens-protoc program. -}</span>
+<a name="line-2"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables, DataKinds, TypeFamilies,
+<a name="line-3"></a>  MultiParamTypeClasses, FlexibleContexts, FlexibleInstances,
+<a name="line-4"></a>  PatternSynonyms #-}</span>
+<a name="line-5"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-unused-imports#-}</span>
+<a name="line-6"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>StepStats</span> <span class='hs-keyword'>where</span>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Prelude</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-11"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-12"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-13"></a>       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-14"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-15"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-16"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-17"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-18"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-19"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-20"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-24"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-25"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-26"></a>       <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span>
+<a name="line-27"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorDescription</span>
+<a name="line-28"></a>
+<a name="line-29"></a><a name="AllocatorMemoryUsed"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>AllocatorMemoryUsed</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AllocatorMemoryUsed</span><span class='hs-layout'>{</span><span class='hs-sel'>_AllocatorMemoryUsed'allocatorName</span>
+<a name="line-30"></a>                                               <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-31"></a>                                               <span class='hs-sel'>_AllocatorMemoryUsed'totalBytes</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-32"></a>                                               <span class='hs-sel'>_AllocatorMemoryUsed'peakBytes</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>}</span>
+<a name="line-33"></a>                         <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-34"></a>
+<a name="line-35"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-36"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocatorName"</span> <span class='hs-conid'>AllocatorMemoryUsed</span> <span class='hs-keyglyph'>=</span>
+<a name="line-37"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-38"></a>
+<a name="line-39"></a><a name="instance%20Data.ProtoLens.HasField%20%22allocatorName%22%20AllocatorMemoryUsed%20AllocatorMemoryUsed"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allocatorName"</span>
+<a name="line-40"></a>         <span class='hs-conid'>AllocatorMemoryUsed</span> <span class='hs-conid'>AllocatorMemoryUsed</span> <span class='hs-keyword'>where</span>
+<a name="line-41"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-42"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AllocatorMemoryUsed'allocatorName</span>
+<a name="line-43"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AllocatorMemoryUsed'allocatorName</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-44"></a>
+<a name="line-45"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"totalBytes"</span> <span class='hs-conid'>AllocatorMemoryUsed</span>
+<a name="line-46"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-47"></a>
+<a name="line-48"></a><a name="instance%20Data.ProtoLens.HasField%20%22totalBytes%22%20AllocatorMemoryUsed%20AllocatorMemoryUsed"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"totalBytes"</span> <span class='hs-conid'>AllocatorMemoryUsed</span>
+<a name="line-49"></a>         <span class='hs-conid'>AllocatorMemoryUsed</span> <span class='hs-keyword'>where</span>
+<a name="line-50"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-51"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AllocatorMemoryUsed'totalBytes</span>
+<a name="line-52"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AllocatorMemoryUsed'totalBytes</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-53"></a>
+<a name="line-54"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"peakBytes"</span> <span class='hs-conid'>AllocatorMemoryUsed</span>
+<a name="line-55"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-56"></a>
+<a name="line-57"></a><a name="instance%20Data.ProtoLens.HasField%20%22peakBytes%22%20AllocatorMemoryUsed%20AllocatorMemoryUsed"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"peakBytes"</span> <span class='hs-conid'>AllocatorMemoryUsed</span>
+<a name="line-58"></a>         <span class='hs-conid'>AllocatorMemoryUsed</span> <span class='hs-keyword'>where</span>
+<a name="line-59"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-60"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_AllocatorMemoryUsed'peakBytes</span>
+<a name="line-61"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_AllocatorMemoryUsed'peakBytes</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-62"></a>
+<a name="line-63"></a><a name="instance%20Data.Default.Class.Default%20AllocatorMemoryUsed"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>AllocatorMemoryUsed</span> <span class='hs-keyword'>where</span>
+<a name="line-64"></a>        <span class='hs-varid'>def</span>
+<a name="line-65"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>AllocatorMemoryUsed</span><span class='hs-layout'>{</span><span class='hs-sel'>_AllocatorMemoryUsed'allocatorName</span> <span class='hs-keyglyph'>=</span>
+<a name="line-66"></a>                                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-67"></a>                                <span class='hs-sel'>_AllocatorMemoryUsed'totalBytes</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-68"></a>                                <span class='hs-sel'>_AllocatorMemoryUsed'peakBytes</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-69"></a>
+<a name="line-70"></a><a name="instance%20Data.ProtoLens.Message%20AllocatorMemoryUsed"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>AllocatorMemoryUsed</span> <span class='hs-keyword'>where</span>
+<a name="line-71"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-72"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>allocatorName__field_descriptor</span>
+<a name="line-73"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"allocator_name"</span>
+<a name="line-74"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-75"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-76"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>allocatorName</span><span class='hs-layout'>)</span>
+<a name="line-77"></a>                <span class='hs-varid'>totalBytes__field_descriptor</span>
+<a name="line-78"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"total_bytes"</span>
+<a name="line-79"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-80"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-81"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>totalBytes</span><span class='hs-layout'>)</span>
+<a name="line-82"></a>                <span class='hs-varid'>peakBytes__field_descriptor</span>
+<a name="line-83"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"peak_bytes"</span>
+<a name="line-84"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-85"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-86"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>peakBytes</span><span class='hs-layout'>)</span>
+<a name="line-87"></a>              <span class='hs-keyword'>in</span>
+<a name="line-88"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-89"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-90"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>allocatorName__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-91"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>totalBytes__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-92"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>peakBytes__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-93"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-94"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"allocator_name"</span><span class='hs-layout'>,</span> <span class='hs-varid'>allocatorName__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-95"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"total_bytes"</span><span class='hs-layout'>,</span> <span class='hs-varid'>totalBytes__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-96"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"peak_bytes"</span><span class='hs-layout'>,</span> <span class='hs-varid'>peakBytes__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-97"></a>
+<a name="line-98"></a><a name="DeviceStepStats"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>DeviceStepStats</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DeviceStepStats</span><span class='hs-layout'>{</span><span class='hs-sel'>_DeviceStepStats'device</span> <span class='hs-keyglyph'>::</span>
+<a name="line-99"></a>                                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-100"></a>                                       <span class='hs-sel'>_DeviceStepStats'nodeStats</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>NodeExecStats</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>}</span>
+<a name="line-101"></a>                     <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-102"></a>
+<a name="line-103"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"device"</span> <span class='hs-conid'>DeviceStepStats</span> <span class='hs-keyglyph'>=</span>
+<a name="line-104"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-105"></a>
+<a name="line-106"></a><a name="instance%20Data.ProtoLens.HasField%20%22device%22%20DeviceStepStats%20DeviceStepStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"device"</span> <span class='hs-conid'>DeviceStepStats</span>
+<a name="line-107"></a>         <span class='hs-conid'>DeviceStepStats</span> <span class='hs-keyword'>where</span>
+<a name="line-108"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-109"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_DeviceStepStats'device</span>
+<a name="line-110"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_DeviceStepStats'device</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-111"></a>
+<a name="line-112"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"nodeStats"</span> <span class='hs-conid'>DeviceStepStats</span> <span class='hs-keyglyph'>=</span>
+<a name="line-113"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>NodeExecStats</span><span class='hs-keyglyph'>]</span>
+<a name="line-114"></a>
+<a name="line-115"></a><a name="instance%20Data.ProtoLens.HasField%20%22nodeStats%22%20DeviceStepStats%20DeviceStepStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"nodeStats"</span> <span class='hs-conid'>DeviceStepStats</span>
+<a name="line-116"></a>         <span class='hs-conid'>DeviceStepStats</span> <span class='hs-keyword'>where</span>
+<a name="line-117"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-118"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_DeviceStepStats'nodeStats</span>
+<a name="line-119"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_DeviceStepStats'nodeStats</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-120"></a>
+<a name="line-121"></a><a name="instance%20Data.Default.Class.Default%20DeviceStepStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>DeviceStepStats</span> <span class='hs-keyword'>where</span>
+<a name="line-122"></a>        <span class='hs-varid'>def</span>
+<a name="line-123"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DeviceStepStats</span><span class='hs-layout'>{</span><span class='hs-sel'>_DeviceStepStats'device</span> <span class='hs-keyglyph'>=</span>
+<a name="line-124"></a>                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-125"></a>                            <span class='hs-sel'>_DeviceStepStats'nodeStats</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>}</span>
+<a name="line-126"></a>
+<a name="line-127"></a><a name="instance%20Data.ProtoLens.Message%20DeviceStepStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>DeviceStepStats</span> <span class='hs-keyword'>where</span>
+<a name="line-128"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-129"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>device__field_descriptor</span>
+<a name="line-130"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"device"</span>
+<a name="line-131"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-132"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-133"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>device</span><span class='hs-layout'>)</span>
+<a name="line-134"></a>                <span class='hs-varid'>nodeStats__field_descriptor</span>
+<a name="line-135"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"node_stats"</span>
+<a name="line-136"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-137"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>NodeExecStats</span><span class='hs-layout'>)</span>
+<a name="line-138"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>nodeStats</span><span class='hs-layout'>)</span>
+<a name="line-139"></a>              <span class='hs-keyword'>in</span>
+<a name="line-140"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-141"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-142"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>device__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-143"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>nodeStats__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-144"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-145"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"device"</span><span class='hs-layout'>,</span> <span class='hs-varid'>device__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-146"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"node_stats"</span><span class='hs-layout'>,</span> <span class='hs-varid'>nodeStats__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-147"></a>
+<a name="line-148"></a><a name="NodeExecStats"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NodeExecStats</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeExecStats'nodeName</span> <span class='hs-keyglyph'>::</span>
+<a name="line-149"></a>                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-150"></a>                                   <span class='hs-sel'>_NodeExecStats'allStartMicros</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-151"></a>                                   <span class='hs-sel'>_NodeExecStats'opStartRelMicros</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-152"></a>                                   <span class='hs-sel'>_NodeExecStats'opEndRelMicros</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-153"></a>                                   <span class='hs-sel'>_NodeExecStats'allEndRelMicros</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-154"></a>                                   <span class='hs-sel'>_NodeExecStats'memory</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>AllocatorMemoryUsed</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-155"></a>                                   <span class='hs-sel'>_NodeExecStats'output</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>NodeOutput</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-156"></a>                                   <span class='hs-sel'>_NodeExecStats'timelineLabel</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-157"></a>                                   <span class='hs-sel'>_NodeExecStats'scheduledMicros</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-158"></a>                                   <span class='hs-sel'>_NodeExecStats'threadId</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word32</span><span class='hs-layout'>,</span>
+<a name="line-159"></a>                                   <span class='hs-sel'>_NodeExecStats'referencedTensor</span> <span class='hs-keyglyph'>::</span>
+<a name="line-160"></a>                                   <span class='hs-keyglyph'>[</span><span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>}</span>
+<a name="line-161"></a>                   <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-162"></a>
+<a name="line-163"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"nodeName"</span> <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyglyph'>=</span>
+<a name="line-164"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-165"></a>
+<a name="line-166"></a><a name="instance%20Data.ProtoLens.HasField%20%22nodeName%22%20NodeExecStats%20NodeExecStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"nodeName"</span> <span class='hs-conid'>NodeExecStats</span>
+<a name="line-167"></a>         <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyword'>where</span>
+<a name="line-168"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-169"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeExecStats'nodeName</span>
+<a name="line-170"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeExecStats'nodeName</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-171"></a>
+<a name="line-172"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allStartMicros"</span> <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyglyph'>=</span>
+<a name="line-173"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-174"></a>
+<a name="line-175"></a><a name="instance%20Data.ProtoLens.HasField%20%22allStartMicros%22%20NodeExecStats%20NodeExecStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allStartMicros"</span> <span class='hs-conid'>NodeExecStats</span>
+<a name="line-176"></a>         <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyword'>where</span>
+<a name="line-177"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-178"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeExecStats'allStartMicros</span>
+<a name="line-179"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeExecStats'allStartMicros</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-180"></a>
+<a name="line-181"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"opStartRelMicros"</span> <span class='hs-conid'>NodeExecStats</span>
+<a name="line-182"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-183"></a>
+<a name="line-184"></a><a name="instance%20Data.ProtoLens.HasField%20%22opStartRelMicros%22%20NodeExecStats%20NodeExecStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"opStartRelMicros"</span> <span class='hs-conid'>NodeExecStats</span>
+<a name="line-185"></a>         <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyword'>where</span>
+<a name="line-186"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-187"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeExecStats'opStartRelMicros</span>
+<a name="line-188"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeExecStats'opStartRelMicros</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-189"></a>
+<a name="line-190"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"opEndRelMicros"</span> <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyglyph'>=</span>
+<a name="line-191"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-192"></a>
+<a name="line-193"></a><a name="instance%20Data.ProtoLens.HasField%20%22opEndRelMicros%22%20NodeExecStats%20NodeExecStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"opEndRelMicros"</span> <span class='hs-conid'>NodeExecStats</span>
+<a name="line-194"></a>         <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyword'>where</span>
+<a name="line-195"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-196"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeExecStats'opEndRelMicros</span>
+<a name="line-197"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeExecStats'opEndRelMicros</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-198"></a>
+<a name="line-199"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allEndRelMicros"</span> <span class='hs-conid'>NodeExecStats</span>
+<a name="line-200"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-201"></a>
+<a name="line-202"></a><a name="instance%20Data.ProtoLens.HasField%20%22allEndRelMicros%22%20NodeExecStats%20NodeExecStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allEndRelMicros"</span> <span class='hs-conid'>NodeExecStats</span>
+<a name="line-203"></a>         <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyword'>where</span>
+<a name="line-204"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-205"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeExecStats'allEndRelMicros</span>
+<a name="line-206"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeExecStats'allEndRelMicros</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-207"></a>
+<a name="line-208"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"memory"</span> <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyglyph'>=</span>
+<a name="line-209"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>AllocatorMemoryUsed</span><span class='hs-keyglyph'>]</span>
+<a name="line-210"></a>
+<a name="line-211"></a><a name="instance%20Data.ProtoLens.HasField%20%22memory%22%20NodeExecStats%20NodeExecStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"memory"</span> <span class='hs-conid'>NodeExecStats</span>
+<a name="line-212"></a>         <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyword'>where</span>
+<a name="line-213"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-214"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeExecStats'memory</span>
+<a name="line-215"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeExecStats'memory</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-216"></a>
+<a name="line-217"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"output"</span> <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyglyph'>=</span>
+<a name="line-218"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>NodeOutput</span><span class='hs-keyglyph'>]</span>
+<a name="line-219"></a>
+<a name="line-220"></a><a name="instance%20Data.ProtoLens.HasField%20%22output%22%20NodeExecStats%20NodeExecStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"output"</span> <span class='hs-conid'>NodeExecStats</span>
+<a name="line-221"></a>         <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyword'>where</span>
+<a name="line-222"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-223"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeExecStats'output</span>
+<a name="line-224"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeExecStats'output</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-225"></a>
+<a name="line-226"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"timelineLabel"</span> <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyglyph'>=</span>
+<a name="line-227"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-228"></a>
+<a name="line-229"></a><a name="instance%20Data.ProtoLens.HasField%20%22timelineLabel%22%20NodeExecStats%20NodeExecStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"timelineLabel"</span> <span class='hs-conid'>NodeExecStats</span>
+<a name="line-230"></a>         <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyword'>where</span>
+<a name="line-231"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-232"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeExecStats'timelineLabel</span>
+<a name="line-233"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeExecStats'timelineLabel</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-234"></a>
+<a name="line-235"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"scheduledMicros"</span> <span class='hs-conid'>NodeExecStats</span>
+<a name="line-236"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-237"></a>
+<a name="line-238"></a><a name="instance%20Data.ProtoLens.HasField%20%22scheduledMicros%22%20NodeExecStats%20NodeExecStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"scheduledMicros"</span> <span class='hs-conid'>NodeExecStats</span>
+<a name="line-239"></a>         <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyword'>where</span>
+<a name="line-240"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-241"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeExecStats'scheduledMicros</span>
+<a name="line-242"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeExecStats'scheduledMicros</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-243"></a>
+<a name="line-244"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"threadId"</span> <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyglyph'>=</span>
+<a name="line-245"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word32</span>
+<a name="line-246"></a>
+<a name="line-247"></a><a name="instance%20Data.ProtoLens.HasField%20%22threadId%22%20NodeExecStats%20NodeExecStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"threadId"</span> <span class='hs-conid'>NodeExecStats</span>
+<a name="line-248"></a>         <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyword'>where</span>
+<a name="line-249"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-250"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeExecStats'threadId</span>
+<a name="line-251"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeExecStats'threadId</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-252"></a>
+<a name="line-253"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"referencedTensor"</span> <span class='hs-conid'>NodeExecStats</span>
+<a name="line-254"></a>     <span class='hs-keyglyph'>=</span>
+<a name="line-255"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span><span class='hs-keyglyph'>]</span>
+<a name="line-256"></a>
+<a name="line-257"></a><a name="instance%20Data.ProtoLens.HasField%20%22referencedTensor%22%20NodeExecStats%20NodeExecStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"referencedTensor"</span> <span class='hs-conid'>NodeExecStats</span>
+<a name="line-258"></a>         <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyword'>where</span>
+<a name="line-259"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-260"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeExecStats'referencedTensor</span>
+<a name="line-261"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeExecStats'referencedTensor</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-262"></a>
+<a name="line-263"></a><a name="instance%20Data.Default.Class.Default%20NodeExecStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyword'>where</span>
+<a name="line-264"></a>        <span class='hs-varid'>def</span>
+<a name="line-265"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NodeExecStats</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeExecStats'nodeName</span> <span class='hs-keyglyph'>=</span>
+<a name="line-266"></a>                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-267"></a>                          <span class='hs-sel'>_NodeExecStats'allStartMicros</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-268"></a>                          <span class='hs-sel'>_NodeExecStats'opStartRelMicros</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-269"></a>                          <span class='hs-sel'>_NodeExecStats'opEndRelMicros</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-270"></a>                          <span class='hs-sel'>_NodeExecStats'allEndRelMicros</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-271"></a>                          <span class='hs-sel'>_NodeExecStats'memory</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span> <span class='hs-sel'>_NodeExecStats'output</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-272"></a>                          <span class='hs-sel'>_NodeExecStats'timelineLabel</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-273"></a>                          <span class='hs-sel'>_NodeExecStats'scheduledMicros</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-274"></a>                          <span class='hs-sel'>_NodeExecStats'threadId</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-275"></a>                          <span class='hs-sel'>_NodeExecStats'referencedTensor</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>}</span>
+<a name="line-276"></a>
+<a name="line-277"></a><a name="instance%20Data.ProtoLens.Message%20NodeExecStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>NodeExecStats</span> <span class='hs-keyword'>where</span>
+<a name="line-278"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-279"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>nodeName__field_descriptor</span>
+<a name="line-280"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"node_name"</span>
+<a name="line-281"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-282"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-283"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>nodeName</span><span class='hs-layout'>)</span>
+<a name="line-284"></a>                <span class='hs-varid'>allStartMicros__field_descriptor</span>
+<a name="line-285"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"all_start_micros"</span>
+<a name="line-286"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-287"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-288"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>allStartMicros</span><span class='hs-layout'>)</span>
+<a name="line-289"></a>                <span class='hs-varid'>opStartRelMicros__field_descriptor</span>
+<a name="line-290"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"op_start_rel_micros"</span>
+<a name="line-291"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-292"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-293"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-294"></a>                         <span class='hs-varid'>opStartRelMicros</span><span class='hs-layout'>)</span>
+<a name="line-295"></a>                <span class='hs-varid'>opEndRelMicros__field_descriptor</span>
+<a name="line-296"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"op_end_rel_micros"</span>
+<a name="line-297"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-298"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-299"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>opEndRelMicros</span><span class='hs-layout'>)</span>
+<a name="line-300"></a>                <span class='hs-varid'>allEndRelMicros__field_descriptor</span>
+<a name="line-301"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"all_end_rel_micros"</span>
+<a name="line-302"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-303"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-304"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>allEndRelMicros</span><span class='hs-layout'>)</span>
+<a name="line-305"></a>                <span class='hs-varid'>memory__field_descriptor</span>
+<a name="line-306"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"memory"</span>
+<a name="line-307"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-308"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>AllocatorMemoryUsed</span><span class='hs-layout'>)</span>
+<a name="line-309"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>memory</span><span class='hs-layout'>)</span>
+<a name="line-310"></a>                <span class='hs-varid'>output__field_descriptor</span>
+<a name="line-311"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"output"</span>
+<a name="line-312"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-313"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>NodeOutput</span><span class='hs-layout'>)</span>
+<a name="line-314"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>output</span><span class='hs-layout'>)</span>
+<a name="line-315"></a>                <span class='hs-varid'>timelineLabel__field_descriptor</span>
+<a name="line-316"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"timeline_label"</span>
+<a name="line-317"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-318"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-319"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>timelineLabel</span><span class='hs-layout'>)</span>
+<a name="line-320"></a>                <span class='hs-varid'>scheduledMicros__field_descriptor</span>
+<a name="line-321"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"scheduled_micros"</span>
+<a name="line-322"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-323"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-324"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>scheduledMicros</span><span class='hs-layout'>)</span>
+<a name="line-325"></a>                <span class='hs-varid'>threadId__field_descriptor</span>
+<a name="line-326"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"thread_id"</span>
+<a name="line-327"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>UInt32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-328"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span><span class='hs-varop'>.</span><span class='hs-conid'>Word32</span><span class='hs-layout'>)</span>
+<a name="line-329"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>threadId</span><span class='hs-layout'>)</span>
+<a name="line-330"></a>                <span class='hs-varid'>referencedTensor__field_descriptor</span>
+<a name="line-331"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"referenced_tensor"</span>
+<a name="line-332"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-333"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-334"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span><span class='hs-layout'>)</span>
+<a name="line-335"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span>
+<a name="line-336"></a>                         <span class='hs-varid'>referencedTensor</span><span class='hs-layout'>)</span>
+<a name="line-337"></a>              <span class='hs-keyword'>in</span>
+<a name="line-338"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-339"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-340"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>nodeName__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-341"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>allStartMicros__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-342"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>opStartRelMicros__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-343"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>opEndRelMicros__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-344"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>5</span><span class='hs-layout'>,</span> <span class='hs-varid'>allEndRelMicros__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-345"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>6</span><span class='hs-layout'>,</span> <span class='hs-varid'>memory__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-346"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>7</span><span class='hs-layout'>,</span> <span class='hs-varid'>output__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-347"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>8</span><span class='hs-layout'>,</span> <span class='hs-varid'>timelineLabel__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-348"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>9</span><span class='hs-layout'>,</span> <span class='hs-varid'>scheduledMicros__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-349"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>10</span><span class='hs-layout'>,</span> <span class='hs-varid'>threadId__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-350"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>11</span><span class='hs-layout'>,</span> <span class='hs-varid'>referencedTensor__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-351"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-352"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"node_name"</span><span class='hs-layout'>,</span> <span class='hs-varid'>nodeName__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-353"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"all_start_micros"</span><span class='hs-layout'>,</span> <span class='hs-varid'>allStartMicros__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-354"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"op_start_rel_micros"</span><span class='hs-layout'>,</span> <span class='hs-varid'>opStartRelMicros__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-355"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"op_end_rel_micros"</span><span class='hs-layout'>,</span> <span class='hs-varid'>opEndRelMicros__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-356"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"all_end_rel_micros"</span><span class='hs-layout'>,</span> <span class='hs-varid'>allEndRelMicros__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-357"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"memory"</span><span class='hs-layout'>,</span> <span class='hs-varid'>memory__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-358"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"output"</span><span class='hs-layout'>,</span> <span class='hs-varid'>output__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-359"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"timeline_label"</span><span class='hs-layout'>,</span> <span class='hs-varid'>timelineLabel__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-360"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"scheduled_micros"</span><span class='hs-layout'>,</span> <span class='hs-varid'>scheduledMicros__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-361"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"thread_id"</span><span class='hs-layout'>,</span> <span class='hs-varid'>threadId__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-362"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"referenced_tensor"</span><span class='hs-layout'>,</span> <span class='hs-varid'>referencedTensor__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-363"></a>
+<a name="line-364"></a><a name="NodeOutput"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>NodeOutput</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NodeOutput</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeOutput'slot</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-365"></a>                             <span class='hs-sel'>_NodeOutput'tensorDescription</span> <span class='hs-keyglyph'>::</span>
+<a name="line-366"></a>                             <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-367"></a>                               <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorDescription</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorDescription</span><span class='hs-layout'>}</span>
+<a name="line-368"></a>                <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-369"></a>
+<a name="line-370"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"slot"</span> <span class='hs-conid'>NodeOutput</span> <span class='hs-keyglyph'>=</span>
+<a name="line-371"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-372"></a>
+<a name="line-373"></a><a name="instance%20Data.ProtoLens.HasField%20%22slot%22%20NodeOutput%20NodeOutput"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"slot"</span> <span class='hs-conid'>NodeOutput</span> <span class='hs-conid'>NodeOutput</span> <span class='hs-keyword'>where</span>
+<a name="line-374"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-375"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeOutput'slot</span>
+<a name="line-376"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeOutput'slot</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-377"></a>
+<a name="line-378"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"tensorDescription"</span> <span class='hs-conid'>NodeOutput</span> <span class='hs-keyglyph'>=</span>
+<a name="line-379"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorDescription</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorDescription</span>
+<a name="line-380"></a>
+<a name="line-381"></a><a name="instance%20Data.ProtoLens.HasField%20%22tensorDescription%22%20NodeOutput%20NodeOutput"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"tensorDescription"</span> <span class='hs-conid'>NodeOutput</span>
+<a name="line-382"></a>         <span class='hs-conid'>NodeOutput</span> <span class='hs-keyword'>where</span>
+<a name="line-383"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-384"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'tensorDescription</span>
+<a name="line-385"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-386"></a>
+<a name="line-387"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-388"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'tensorDescription"</span> <span class='hs-conid'>NodeOutput</span> <span class='hs-keyglyph'>=</span>
+<a name="line-389"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-390"></a>       <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorDescription</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorDescription</span>
+<a name="line-391"></a>
+<a name="line-392"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'tensorDescription%22%20NodeOutput%20NodeOutput"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'tensorDescription"</span>
+<a name="line-393"></a>         <span class='hs-conid'>NodeOutput</span> <span class='hs-conid'>NodeOutput</span> <span class='hs-keyword'>where</span>
+<a name="line-394"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-395"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_NodeOutput'tensorDescription</span>
+<a name="line-396"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeOutput'tensorDescription</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-397"></a>
+<a name="line-398"></a><a name="instance%20Data.Default.Class.Default%20NodeOutput"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>NodeOutput</span> <span class='hs-keyword'>where</span>
+<a name="line-399"></a>        <span class='hs-varid'>def</span>
+<a name="line-400"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>NodeOutput</span><span class='hs-layout'>{</span><span class='hs-sel'>_NodeOutput'slot</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-401"></a>                       <span class='hs-sel'>_NodeOutput'tensorDescription</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>}</span>
+<a name="line-402"></a>
+<a name="line-403"></a><a name="instance%20Data.ProtoLens.Message%20NodeOutput"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>NodeOutput</span> <span class='hs-keyword'>where</span>
+<a name="line-404"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-405"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>slot__field_descriptor</span>
+<a name="line-406"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"slot"</span>
+<a name="line-407"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-408"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-409"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>slot</span><span class='hs-layout'>)</span>
+<a name="line-410"></a>                <span class='hs-varid'>tensorDescription__field_descriptor</span>
+<a name="line-411"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"tensor_description"</span>
+<a name="line-412"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-413"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-414"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorDescription</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorDescription</span><span class='hs-layout'>)</span>
+<a name="line-415"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'tensorDescription</span><span class='hs-layout'>)</span>
+<a name="line-416"></a>              <span class='hs-keyword'>in</span>
+<a name="line-417"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-418"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-419"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>slot__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-420"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>tensorDescription__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-421"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-422"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"slot"</span><span class='hs-layout'>,</span> <span class='hs-varid'>slot__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-423"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"tensor_description"</span><span class='hs-layout'>,</span> <span class='hs-varid'>tensorDescription__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-424"></a>
+<a name="line-425"></a><a name="StepStats"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>StepStats</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>StepStats</span><span class='hs-layout'>{</span><span class='hs-sel'>_StepStats'devStats</span> <span class='hs-keyglyph'>::</span>
+<a name="line-426"></a>                           <span class='hs-keyglyph'>[</span><span class='hs-conid'>DeviceStepStats</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>}</span>
+<a name="line-427"></a>               <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-428"></a>
+<a name="line-429"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"devStats"</span> <span class='hs-conid'>StepStats</span> <span class='hs-keyglyph'>=</span>
+<a name="line-430"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>DeviceStepStats</span><span class='hs-keyglyph'>]</span>
+<a name="line-431"></a>
+<a name="line-432"></a><a name="instance%20Data.ProtoLens.HasField%20%22devStats%22%20StepStats%20StepStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"devStats"</span> <span class='hs-conid'>StepStats</span> <span class='hs-conid'>StepStats</span>
+<a name="line-433"></a>         <span class='hs-keyword'>where</span>
+<a name="line-434"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-435"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_StepStats'devStats</span>
+<a name="line-436"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_StepStats'devStats</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-437"></a>
+<a name="line-438"></a><a name="instance%20Data.Default.Class.Default%20StepStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>StepStats</span> <span class='hs-keyword'>where</span>
+<a name="line-439"></a>        <span class='hs-varid'>def</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>StepStats</span><span class='hs-layout'>{</span><span class='hs-sel'>_StepStats'devStats</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>}</span>
+<a name="line-440"></a>
+<a name="line-441"></a><a name="instance%20Data.ProtoLens.Message%20StepStats"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>StepStats</span> <span class='hs-keyword'>where</span>
+<a name="line-442"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-443"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>devStats__field_descriptor</span>
+<a name="line-444"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"dev_stats"</span>
+<a name="line-445"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-446"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>DeviceStepStats</span><span class='hs-layout'>)</span>
+<a name="line-447"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>devStats</span><span class='hs-layout'>)</span>
+<a name="line-448"></a>              <span class='hs-keyword'>in</span>
+<a name="line-449"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-450"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-451"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>devStats__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-452"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"dev_stats"</span><span class='hs-layout'>,</span> <span class='hs-varid'>devStats__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-453"></a>
+<a name="line-454"></a><a name="allEndRelMicros"></a><span class='hs-definition'>allEndRelMicros</span> <span class='hs-keyglyph'>::</span>
+<a name="line-455"></a>                <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-456"></a>                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allEndRelMicros"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-457"></a>                  <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-458"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allEndRelMicros"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-459"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allEndRelMicros"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-460"></a><span class='hs-definition'>allEndRelMicros</span>
+<a name="line-461"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-462"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-463"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"allEndRelMicros"</span><span class='hs-layout'>)</span>
+<a name="line-464"></a>
+<a name="line-465"></a><a name="allStartMicros"></a><span class='hs-definition'>allStartMicros</span> <span class='hs-keyglyph'>::</span>
+<a name="line-466"></a>               <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-467"></a>                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allStartMicros"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-468"></a>                 <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-469"></a>                   <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allStartMicros"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-470"></a>                   <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allStartMicros"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-471"></a><span class='hs-definition'>allStartMicros</span>
+<a name="line-472"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-473"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-474"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"allStartMicros"</span><span class='hs-layout'>)</span>
+<a name="line-475"></a>
+<a name="line-476"></a><a name="allocatorName"></a><span class='hs-definition'>allocatorName</span> <span class='hs-keyglyph'>::</span>
+<a name="line-477"></a>              <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-478"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allocatorName"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-479"></a>                <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-480"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocatorName"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-481"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocatorName"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-482"></a><span class='hs-definition'>allocatorName</span>
+<a name="line-483"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-484"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-485"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"allocatorName"</span><span class='hs-layout'>)</span>
+<a name="line-486"></a>
+<a name="line-487"></a><a name="devStats"></a><span class='hs-definition'>devStats</span> <span class='hs-keyglyph'>::</span>
+<a name="line-488"></a>         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"devStats"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-489"></a>           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"devStats"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-490"></a>             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"devStats"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-491"></a><span class='hs-definition'>devStats</span>
+<a name="line-492"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-493"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"devStats"</span><span class='hs-layout'>)</span>
+<a name="line-494"></a>
+<a name="line-495"></a><a name="device"></a><span class='hs-definition'>device</span> <span class='hs-keyglyph'>::</span>
+<a name="line-496"></a>       <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"device"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-497"></a>         <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"device"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-498"></a>           <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"device"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-499"></a><span class='hs-definition'>device</span>
+<a name="line-500"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-501"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"device"</span><span class='hs-layout'>)</span>
+<a name="line-502"></a>
+<a name="line-503"></a><a name="maybe'tensorDescription"></a><span class='hs-definition'>maybe'tensorDescription</span> <span class='hs-keyglyph'>::</span>
+<a name="line-504"></a>                        <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-505"></a>                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'tensorDescription"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-506"></a>                          <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-507"></a>                            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'tensorDescription"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-508"></a>                            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'tensorDescription"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-509"></a><span class='hs-definition'>maybe'tensorDescription</span>
+<a name="line-510"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-511"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-512"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'tensorDescription"</span><span class='hs-layout'>)</span>
+<a name="line-513"></a>
+<a name="line-514"></a><a name="memory"></a><span class='hs-definition'>memory</span> <span class='hs-keyglyph'>::</span>
+<a name="line-515"></a>       <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"memory"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-516"></a>         <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"memory"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-517"></a>           <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"memory"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-518"></a><span class='hs-definition'>memory</span>
+<a name="line-519"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-520"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"memory"</span><span class='hs-layout'>)</span>
+<a name="line-521"></a>
+<a name="line-522"></a><a name="nodeName"></a><span class='hs-definition'>nodeName</span> <span class='hs-keyglyph'>::</span>
+<a name="line-523"></a>         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"nodeName"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-524"></a>           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"nodeName"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-525"></a>             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"nodeName"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-526"></a><span class='hs-definition'>nodeName</span>
+<a name="line-527"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-528"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"nodeName"</span><span class='hs-layout'>)</span>
+<a name="line-529"></a>
+<a name="line-530"></a><a name="nodeStats"></a><span class='hs-definition'>nodeStats</span> <span class='hs-keyglyph'>::</span>
+<a name="line-531"></a>          <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"nodeStats"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-532"></a>            <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"nodeStats"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-533"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"nodeStats"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-534"></a><span class='hs-definition'>nodeStats</span>
+<a name="line-535"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-536"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"nodeStats"</span><span class='hs-layout'>)</span>
+<a name="line-537"></a>
+<a name="line-538"></a><a name="opEndRelMicros"></a><span class='hs-definition'>opEndRelMicros</span> <span class='hs-keyglyph'>::</span>
+<a name="line-539"></a>               <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-540"></a>                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"opEndRelMicros"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-541"></a>                 <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-542"></a>                   <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"opEndRelMicros"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-543"></a>                   <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"opEndRelMicros"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-544"></a><span class='hs-definition'>opEndRelMicros</span>
+<a name="line-545"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-546"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-547"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"opEndRelMicros"</span><span class='hs-layout'>)</span>
+<a name="line-548"></a>
+<a name="line-549"></a><a name="opStartRelMicros"></a><span class='hs-definition'>opStartRelMicros</span> <span class='hs-keyglyph'>::</span>
+<a name="line-550"></a>                 <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-551"></a>                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"opStartRelMicros"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-552"></a>                   <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-553"></a>                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"opStartRelMicros"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-554"></a>                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"opStartRelMicros"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-555"></a><span class='hs-definition'>opStartRelMicros</span>
+<a name="line-556"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-557"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-558"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"opStartRelMicros"</span><span class='hs-layout'>)</span>
+<a name="line-559"></a>
+<a name="line-560"></a><a name="output"></a><span class='hs-definition'>output</span> <span class='hs-keyglyph'>::</span>
+<a name="line-561"></a>       <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"output"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-562"></a>         <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"output"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-563"></a>           <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"output"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-564"></a><span class='hs-definition'>output</span>
+<a name="line-565"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-566"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"output"</span><span class='hs-layout'>)</span>
+<a name="line-567"></a>
+<a name="line-568"></a><a name="peakBytes"></a><span class='hs-definition'>peakBytes</span> <span class='hs-keyglyph'>::</span>
+<a name="line-569"></a>          <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"peakBytes"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-570"></a>            <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"peakBytes"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-571"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"peakBytes"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-572"></a><span class='hs-definition'>peakBytes</span>
+<a name="line-573"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-574"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"peakBytes"</span><span class='hs-layout'>)</span>
+<a name="line-575"></a>
+<a name="line-576"></a><a name="referencedTensor"></a><span class='hs-definition'>referencedTensor</span> <span class='hs-keyglyph'>::</span>
+<a name="line-577"></a>                 <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-578"></a>                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"referencedTensor"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-579"></a>                   <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-580"></a>                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"referencedTensor"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-581"></a>                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"referencedTensor"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-582"></a><span class='hs-definition'>referencedTensor</span>
+<a name="line-583"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-584"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-585"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"referencedTensor"</span><span class='hs-layout'>)</span>
+<a name="line-586"></a>
+<a name="line-587"></a><a name="scheduledMicros"></a><span class='hs-definition'>scheduledMicros</span> <span class='hs-keyglyph'>::</span>
+<a name="line-588"></a>                <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-589"></a>                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"scheduledMicros"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-590"></a>                  <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-591"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"scheduledMicros"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-592"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"scheduledMicros"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-593"></a><span class='hs-definition'>scheduledMicros</span>
+<a name="line-594"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-595"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-596"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"scheduledMicros"</span><span class='hs-layout'>)</span>
+<a name="line-597"></a>
+<a name="line-598"></a><a name="slot"></a><span class='hs-definition'>slot</span> <span class='hs-keyglyph'>::</span>
+<a name="line-599"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"slot"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-600"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"slot"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-601"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"slot"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-602"></a><span class='hs-definition'>slot</span>
+<a name="line-603"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-604"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"slot"</span><span class='hs-layout'>)</span>
+<a name="line-605"></a>
+<a name="line-606"></a><a name="tensorDescription"></a><span class='hs-definition'>tensorDescription</span> <span class='hs-keyglyph'>::</span>
+<a name="line-607"></a>                  <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-608"></a>                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"tensorDescription"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-609"></a>                    <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-610"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"tensorDescription"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-611"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"tensorDescription"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-612"></a><span class='hs-definition'>tensorDescription</span>
+<a name="line-613"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-614"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-615"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"tensorDescription"</span><span class='hs-layout'>)</span>
+<a name="line-616"></a>
+<a name="line-617"></a><a name="threadId"></a><span class='hs-definition'>threadId</span> <span class='hs-keyglyph'>::</span>
+<a name="line-618"></a>         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"threadId"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-619"></a>           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"threadId"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-620"></a>             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"threadId"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-621"></a><span class='hs-definition'>threadId</span>
+<a name="line-622"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-623"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"threadId"</span><span class='hs-layout'>)</span>
+<a name="line-624"></a>
+<a name="line-625"></a><a name="timelineLabel"></a><span class='hs-definition'>timelineLabel</span> <span class='hs-keyglyph'>::</span>
+<a name="line-626"></a>              <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-627"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"timelineLabel"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-628"></a>                <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-629"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"timelineLabel"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-630"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"timelineLabel"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-631"></a><span class='hs-definition'>timelineLabel</span>
+<a name="line-632"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-633"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-634"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"timelineLabel"</span><span class='hs-layout'>)</span>
+<a name="line-635"></a>
+<a name="line-636"></a><a name="totalBytes"></a><span class='hs-definition'>totalBytes</span> <span class='hs-keyglyph'>::</span>
+<a name="line-637"></a>           <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"totalBytes"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-638"></a>             <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"totalBytes"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-639"></a>               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"totalBytes"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-640"></a><span class='hs-definition'>totalBytes</span>
+<a name="line-641"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-642"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"totalBytes"</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Tensor.html b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Tensor.html
new file mode 100644
index 0000000..5d54c36
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Tensor.html
@@ -0,0 +1,448 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Proto/Tensorflow/Core/Framework/Tensor.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>{- This file was auto-generated from tensorflow/core/framework/tensor.proto by the proto-lens-protoc program. -}</span>
+<a name="line-2"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables, DataKinds, TypeFamilies,
+<a name="line-3"></a>  MultiParamTypeClasses, FlexibleContexts, FlexibleInstances,
+<a name="line-4"></a>  PatternSynonyms #-}</span>
+<a name="line-5"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-unused-imports#-}</span>
+<a name="line-6"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span> <span class='hs-keyword'>where</span>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Prelude</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-11"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-12"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-13"></a>       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-14"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-15"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-16"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-17"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-18"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-19"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-20"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-24"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-25"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>ResourceHandle</span>
+<a name="line-26"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span>
+<a name="line-27"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span>
+<a name="line-28"></a>
+<a name="line-29"></a><a name="TensorProto"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TensorProto</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorProto'dtype</span> <span class='hs-keyglyph'>::</span>
+<a name="line-30"></a>                               <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span><span class='hs-layout'>,</span>
+<a name="line-31"></a>                               <span class='hs-sel'>_TensorProto'tensorShape</span> <span class='hs-keyglyph'>::</span>
+<a name="line-32"></a>                               <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-33"></a>                                 <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span><span class='hs-layout'>,</span>
+<a name="line-34"></a>                               <span class='hs-sel'>_TensorProto'versionNumber</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-35"></a>                               <span class='hs-sel'>_TensorProto'tensorContent</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>,</span>
+<a name="line-36"></a>                               <span class='hs-sel'>_TensorProto'halfVal</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-37"></a>                               <span class='hs-sel'>_TensorProto'floatVal</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-38"></a>                               <span class='hs-sel'>_TensorProto'doubleVal</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Double</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-39"></a>                               <span class='hs-sel'>_TensorProto'intVal</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-40"></a>                               <span class='hs-sel'>_TensorProto'stringVal</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-41"></a>                               <span class='hs-sel'>_TensorProto'scomplexVal</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-42"></a>                               <span class='hs-sel'>_TensorProto'int64Val</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-43"></a>                               <span class='hs-sel'>_TensorProto'boolVal</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-44"></a>                               <span class='hs-sel'>_TensorProto'dcomplexVal</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Double</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-45"></a>                               <span class='hs-sel'>_TensorProto'resourceHandleVal</span> <span class='hs-keyglyph'>::</span>
+<a name="line-46"></a>                               <span class='hs-keyglyph'>[</span><span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>ResourceHandle</span><span class='hs-varop'>.</span><span class='hs-conid'>ResourceHandle</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>}</span>
+<a name="line-47"></a>                 <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-48"></a>
+<a name="line-49"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dtype"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-50"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span>
+<a name="line-51"></a>
+<a name="line-52"></a><a name="instance%20Data.ProtoLens.HasField%20%22dtype%22%20TensorProto%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"dtype"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-53"></a>         <span class='hs-keyword'>where</span>
+<a name="line-54"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-55"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorProto'dtype</span>
+<a name="line-56"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorProto'dtype</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-57"></a>
+<a name="line-58"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"tensorShape"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-59"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span>
+<a name="line-60"></a>
+<a name="line-61"></a><a name="instance%20Data.ProtoLens.HasField%20%22tensorShape%22%20TensorProto%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"tensorShape"</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-62"></a>         <span class='hs-conid'>TensorProto</span> <span class='hs-keyword'>where</span>
+<a name="line-63"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-64"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'tensorShape</span>
+<a name="line-65"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-66"></a>
+<a name="line-67"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'tensorShape"</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-68"></a>     <span class='hs-keyglyph'>=</span>
+<a name="line-69"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-70"></a>       <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span>
+<a name="line-71"></a>
+<a name="line-72"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'tensorShape%22%20TensorProto%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'tensorShape"</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-73"></a>         <span class='hs-conid'>TensorProto</span> <span class='hs-keyword'>where</span>
+<a name="line-74"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-75"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorProto'tensorShape</span>
+<a name="line-76"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorProto'tensorShape</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-77"></a>
+<a name="line-78"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"versionNumber"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-79"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-80"></a>
+<a name="line-81"></a><a name="instance%20Data.ProtoLens.HasField%20%22versionNumber%22%20TensorProto%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"versionNumber"</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-82"></a>         <span class='hs-conid'>TensorProto</span> <span class='hs-keyword'>where</span>
+<a name="line-83"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-84"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorProto'versionNumber</span>
+<a name="line-85"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorProto'versionNumber</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-86"></a>
+<a name="line-87"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"tensorContent"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-88"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-89"></a>
+<a name="line-90"></a><a name="instance%20Data.ProtoLens.HasField%20%22tensorContent%22%20TensorProto%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"tensorContent"</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-91"></a>         <span class='hs-conid'>TensorProto</span> <span class='hs-keyword'>where</span>
+<a name="line-92"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-93"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorProto'tensorContent</span>
+<a name="line-94"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorProto'tensorContent</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-95"></a>
+<a name="line-96"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"halfVal"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-97"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-keyglyph'>]</span>
+<a name="line-98"></a>
+<a name="line-99"></a><a name="instance%20Data.ProtoLens.HasField%20%22halfVal%22%20TensorProto%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"halfVal"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-100"></a>         <span class='hs-keyword'>where</span>
+<a name="line-101"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-102"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorProto'halfVal</span>
+<a name="line-103"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorProto'halfVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-104"></a>
+<a name="line-105"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"floatVal"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-106"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span>
+<a name="line-107"></a>
+<a name="line-108"></a><a name="instance%20Data.ProtoLens.HasField%20%22floatVal%22%20TensorProto%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"floatVal"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-109"></a>         <span class='hs-keyword'>where</span>
+<a name="line-110"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-111"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorProto'floatVal</span>
+<a name="line-112"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorProto'floatVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-113"></a>
+<a name="line-114"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"doubleVal"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-115"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Double</span><span class='hs-keyglyph'>]</span>
+<a name="line-116"></a>
+<a name="line-117"></a><a name="instance%20Data.ProtoLens.HasField%20%22doubleVal%22%20TensorProto%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"doubleVal"</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-118"></a>         <span class='hs-conid'>TensorProto</span> <span class='hs-keyword'>where</span>
+<a name="line-119"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-120"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorProto'doubleVal</span>
+<a name="line-121"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorProto'doubleVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-122"></a>
+<a name="line-123"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"intVal"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-124"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-keyglyph'>]</span>
+<a name="line-125"></a>
+<a name="line-126"></a><a name="instance%20Data.ProtoLens.HasField%20%22intVal%22%20TensorProto%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"intVal"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-127"></a>         <span class='hs-keyword'>where</span>
+<a name="line-128"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-129"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorProto'intVal</span>
+<a name="line-130"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorProto'intVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-131"></a>
+<a name="line-132"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"stringVal"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-133"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-keyglyph'>]</span>
+<a name="line-134"></a>
+<a name="line-135"></a><a name="instance%20Data.ProtoLens.HasField%20%22stringVal%22%20TensorProto%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"stringVal"</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-136"></a>         <span class='hs-conid'>TensorProto</span> <span class='hs-keyword'>where</span>
+<a name="line-137"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-138"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorProto'stringVal</span>
+<a name="line-139"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorProto'stringVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-140"></a>
+<a name="line-141"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"scomplexVal"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-142"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Float</span><span class='hs-keyglyph'>]</span>
+<a name="line-143"></a>
+<a name="line-144"></a><a name="instance%20Data.ProtoLens.HasField%20%22scomplexVal%22%20TensorProto%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"scomplexVal"</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-145"></a>         <span class='hs-conid'>TensorProto</span> <span class='hs-keyword'>where</span>
+<a name="line-146"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-147"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorProto'scomplexVal</span>
+<a name="line-148"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorProto'scomplexVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-149"></a>
+<a name="line-150"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"int64Val"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-151"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-keyglyph'>]</span>
+<a name="line-152"></a>
+<a name="line-153"></a><a name="instance%20Data.ProtoLens.HasField%20%22int64Val%22%20TensorProto%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"int64Val"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-154"></a>         <span class='hs-keyword'>where</span>
+<a name="line-155"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-156"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorProto'int64Val</span>
+<a name="line-157"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorProto'int64Val</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-158"></a>
+<a name="line-159"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"boolVal"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-160"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-keyglyph'>]</span>
+<a name="line-161"></a>
+<a name="line-162"></a><a name="instance%20Data.ProtoLens.HasField%20%22boolVal%22%20TensorProto%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"boolVal"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-163"></a>         <span class='hs-keyword'>where</span>
+<a name="line-164"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-165"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorProto'boolVal</span>
+<a name="line-166"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorProto'boolVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-167"></a>
+<a name="line-168"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dcomplexVal"</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-169"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Double</span><span class='hs-keyglyph'>]</span>
+<a name="line-170"></a>
+<a name="line-171"></a><a name="instance%20Data.ProtoLens.HasField%20%22dcomplexVal%22%20TensorProto%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"dcomplexVal"</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-172"></a>         <span class='hs-conid'>TensorProto</span> <span class='hs-keyword'>where</span>
+<a name="line-173"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-174"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorProto'dcomplexVal</span>
+<a name="line-175"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorProto'dcomplexVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-176"></a>
+<a name="line-177"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"resourceHandleVal"</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-178"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>ResourceHandle</span><span class='hs-varop'>.</span><span class='hs-conid'>ResourceHandle</span><span class='hs-keyglyph'>]</span>
+<a name="line-179"></a>
+<a name="line-180"></a><a name="instance%20Data.ProtoLens.HasField%20%22resourceHandleVal%22%20TensorProto%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"resourceHandleVal"</span> <span class='hs-conid'>TensorProto</span>
+<a name="line-181"></a>         <span class='hs-conid'>TensorProto</span> <span class='hs-keyword'>where</span>
+<a name="line-182"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-183"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorProto'resourceHandleVal</span>
+<a name="line-184"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorProto'resourceHandleVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-185"></a>
+<a name="line-186"></a><a name="instance%20Data.Default.Class.Default%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyword'>where</span>
+<a name="line-187"></a>        <span class='hs-varid'>def</span>
+<a name="line-188"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TensorProto</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorProto'dtype</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>,</span>
+<a name="line-189"></a>                        <span class='hs-sel'>_TensorProto'tensorShape</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-190"></a>                        <span class='hs-sel'>_TensorProto'versionNumber</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-191"></a>                        <span class='hs-sel'>_TensorProto'tensorContent</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-192"></a>                        <span class='hs-sel'>_TensorProto'halfVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span> <span class='hs-sel'>_TensorProto'floatVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-193"></a>                        <span class='hs-sel'>_TensorProto'doubleVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span> <span class='hs-sel'>_TensorProto'intVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-194"></a>                        <span class='hs-sel'>_TensorProto'stringVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span> <span class='hs-sel'>_TensorProto'scomplexVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-195"></a>                        <span class='hs-sel'>_TensorProto'int64Val</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span> <span class='hs-sel'>_TensorProto'boolVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-196"></a>                        <span class='hs-sel'>_TensorProto'dcomplexVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span> <span class='hs-sel'>_TensorProto'resourceHandleVal</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>}</span>
+<a name="line-197"></a>
+<a name="line-198"></a><a name="instance%20Data.ProtoLens.Message%20TensorProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>TensorProto</span> <span class='hs-keyword'>where</span>
+<a name="line-199"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-200"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>dtype__field_descriptor</span>
+<a name="line-201"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"dtype"</span>
+<a name="line-202"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>EnumField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-203"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-204"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span><span class='hs-layout'>)</span>
+<a name="line-205"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span>
+<a name="line-206"></a>                <span class='hs-varid'>tensorShape__field_descriptor</span>
+<a name="line-207"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"tensor_shape"</span>
+<a name="line-208"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-209"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-210"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span><span class='hs-layout'>)</span>
+<a name="line-211"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'tensorShape</span><span class='hs-layout'>)</span>
+<a name="line-212"></a>                <span class='hs-varid'>versionNumber__field_descriptor</span>
+<a name="line-213"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"version_number"</span>
+<a name="line-214"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-215"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-216"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>versionNumber</span><span class='hs-layout'>)</span>
+<a name="line-217"></a>                <span class='hs-varid'>tensorContent__field_descriptor</span>
+<a name="line-218"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"tensor_content"</span>
+<a name="line-219"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BytesField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-220"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>
+<a name="line-221"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>tensorContent</span><span class='hs-layout'>)</span>
+<a name="line-222"></a>                <span class='hs-varid'>halfVal__field_descriptor</span>
+<a name="line-223"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"half_val"</span>
+<a name="line-224"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-225"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-226"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Packed</span> <span class='hs-varid'>halfVal</span><span class='hs-layout'>)</span>
+<a name="line-227"></a>                <span class='hs-varid'>floatVal__field_descriptor</span>
+<a name="line-228"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"float_val"</span>
+<a name="line-229"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FloatField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-230"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Float</span><span class='hs-layout'>)</span>
+<a name="line-231"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Packed</span> <span class='hs-varid'>floatVal</span><span class='hs-layout'>)</span>
+<a name="line-232"></a>                <span class='hs-varid'>doubleVal__field_descriptor</span>
+<a name="line-233"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"double_val"</span>
+<a name="line-234"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>DoubleField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-235"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Double</span><span class='hs-layout'>)</span>
+<a name="line-236"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Packed</span> <span class='hs-varid'>doubleVal</span><span class='hs-layout'>)</span>
+<a name="line-237"></a>                <span class='hs-varid'>intVal__field_descriptor</span>
+<a name="line-238"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"int_val"</span>
+<a name="line-239"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-240"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-241"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Packed</span> <span class='hs-varid'>intVal</span><span class='hs-layout'>)</span>
+<a name="line-242"></a>                <span class='hs-varid'>stringVal__field_descriptor</span>
+<a name="line-243"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"string_val"</span>
+<a name="line-244"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BytesField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-245"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>
+<a name="line-246"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>stringVal</span><span class='hs-layout'>)</span>
+<a name="line-247"></a>                <span class='hs-varid'>scomplexVal__field_descriptor</span>
+<a name="line-248"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"scomplex_val"</span>
+<a name="line-249"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FloatField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-250"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Float</span><span class='hs-layout'>)</span>
+<a name="line-251"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Packed</span> <span class='hs-varid'>scomplexVal</span><span class='hs-layout'>)</span>
+<a name="line-252"></a>                <span class='hs-varid'>int64Val__field_descriptor</span>
+<a name="line-253"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"int64_val"</span>
+<a name="line-254"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-255"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-256"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Packed</span> <span class='hs-varid'>int64Val</span><span class='hs-layout'>)</span>
+<a name="line-257"></a>                <span class='hs-varid'>boolVal__field_descriptor</span>
+<a name="line-258"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"bool_val"</span>
+<a name="line-259"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-260"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-261"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Packed</span> <span class='hs-varid'>boolVal</span><span class='hs-layout'>)</span>
+<a name="line-262"></a>                <span class='hs-varid'>dcomplexVal__field_descriptor</span>
+<a name="line-263"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"dcomplex_val"</span>
+<a name="line-264"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>DoubleField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-265"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Double</span><span class='hs-layout'>)</span>
+<a name="line-266"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Packed</span> <span class='hs-varid'>dcomplexVal</span><span class='hs-layout'>)</span>
+<a name="line-267"></a>                <span class='hs-varid'>resourceHandleVal__field_descriptor</span>
+<a name="line-268"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"resource_handle_val"</span>
+<a name="line-269"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-270"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-271"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>ResourceHandle</span><span class='hs-varop'>.</span><span class='hs-conid'>ResourceHandle</span><span class='hs-layout'>)</span>
+<a name="line-272"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span>
+<a name="line-273"></a>                         <span class='hs-varid'>resourceHandleVal</span><span class='hs-layout'>)</span>
+<a name="line-274"></a>              <span class='hs-keyword'>in</span>
+<a name="line-275"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-276"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-277"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>dtype__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-278"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>tensorShape__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-279"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>versionNumber__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-280"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>tensorContent__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-281"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>13</span><span class='hs-layout'>,</span> <span class='hs-varid'>halfVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-282"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>5</span><span class='hs-layout'>,</span> <span class='hs-varid'>floatVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-283"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>6</span><span class='hs-layout'>,</span> <span class='hs-varid'>doubleVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-284"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>7</span><span class='hs-layout'>,</span> <span class='hs-varid'>intVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-285"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>8</span><span class='hs-layout'>,</span> <span class='hs-varid'>stringVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-286"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>9</span><span class='hs-layout'>,</span> <span class='hs-varid'>scomplexVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-287"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>10</span><span class='hs-layout'>,</span> <span class='hs-varid'>int64Val__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-288"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>11</span><span class='hs-layout'>,</span> <span class='hs-varid'>boolVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-289"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>12</span><span class='hs-layout'>,</span> <span class='hs-varid'>dcomplexVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-290"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>14</span><span class='hs-layout'>,</span> <span class='hs-varid'>resourceHandleVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-291"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-292"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"dtype"</span><span class='hs-layout'>,</span> <span class='hs-varid'>dtype__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-293"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"tensor_shape"</span><span class='hs-layout'>,</span> <span class='hs-varid'>tensorShape__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-294"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"version_number"</span><span class='hs-layout'>,</span> <span class='hs-varid'>versionNumber__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-295"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"tensor_content"</span><span class='hs-layout'>,</span> <span class='hs-varid'>tensorContent__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-296"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"half_val"</span><span class='hs-layout'>,</span> <span class='hs-varid'>halfVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-297"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"float_val"</span><span class='hs-layout'>,</span> <span class='hs-varid'>floatVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-298"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"double_val"</span><span class='hs-layout'>,</span> <span class='hs-varid'>doubleVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-299"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"int_val"</span><span class='hs-layout'>,</span> <span class='hs-varid'>intVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-300"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"string_val"</span><span class='hs-layout'>,</span> <span class='hs-varid'>stringVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-301"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"scomplex_val"</span><span class='hs-layout'>,</span> <span class='hs-varid'>scomplexVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-302"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"int64_val"</span><span class='hs-layout'>,</span> <span class='hs-varid'>int64Val__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-303"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"bool_val"</span><span class='hs-layout'>,</span> <span class='hs-varid'>boolVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-304"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"dcomplex_val"</span><span class='hs-layout'>,</span> <span class='hs-varid'>dcomplexVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-305"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"resource_handle_val"</span><span class='hs-layout'>,</span> <span class='hs-varid'>resourceHandleVal__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-306"></a>
+<a name="line-307"></a><a name="boolVal"></a><span class='hs-definition'>boolVal</span> <span class='hs-keyglyph'>::</span>
+<a name="line-308"></a>        <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"boolVal"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-309"></a>          <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"boolVal"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-310"></a>            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"boolVal"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-311"></a><span class='hs-definition'>boolVal</span>
+<a name="line-312"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-313"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"boolVal"</span><span class='hs-layout'>)</span>
+<a name="line-314"></a>
+<a name="line-315"></a><a name="dcomplexVal"></a><span class='hs-definition'>dcomplexVal</span> <span class='hs-keyglyph'>::</span>
+<a name="line-316"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"dcomplexVal"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-317"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dcomplexVal"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-318"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dcomplexVal"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-319"></a><span class='hs-definition'>dcomplexVal</span>
+<a name="line-320"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-321"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"dcomplexVal"</span><span class='hs-layout'>)</span>
+<a name="line-322"></a>
+<a name="line-323"></a><a name="doubleVal"></a><span class='hs-definition'>doubleVal</span> <span class='hs-keyglyph'>::</span>
+<a name="line-324"></a>          <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"doubleVal"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-325"></a>            <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"doubleVal"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-326"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"doubleVal"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-327"></a><span class='hs-definition'>doubleVal</span>
+<a name="line-328"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-329"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"doubleVal"</span><span class='hs-layout'>)</span>
+<a name="line-330"></a>
+<a name="line-331"></a><a name="dtype"></a><span class='hs-definition'>dtype</span> <span class='hs-keyglyph'>::</span>
+<a name="line-332"></a>      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"dtype"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-333"></a>        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dtype"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-334"></a>          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dtype"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-335"></a><span class='hs-definition'>dtype</span>
+<a name="line-336"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-337"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"dtype"</span><span class='hs-layout'>)</span>
+<a name="line-338"></a>
+<a name="line-339"></a><a name="floatVal"></a><span class='hs-definition'>floatVal</span> <span class='hs-keyglyph'>::</span>
+<a name="line-340"></a>         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"floatVal"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-341"></a>           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"floatVal"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-342"></a>             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"floatVal"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-343"></a><span class='hs-definition'>floatVal</span>
+<a name="line-344"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-345"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"floatVal"</span><span class='hs-layout'>)</span>
+<a name="line-346"></a>
+<a name="line-347"></a><a name="halfVal"></a><span class='hs-definition'>halfVal</span> <span class='hs-keyglyph'>::</span>
+<a name="line-348"></a>        <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"halfVal"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-349"></a>          <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"halfVal"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-350"></a>            <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"halfVal"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-351"></a><span class='hs-definition'>halfVal</span>
+<a name="line-352"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-353"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"halfVal"</span><span class='hs-layout'>)</span>
+<a name="line-354"></a>
+<a name="line-355"></a><a name="int64Val"></a><span class='hs-definition'>int64Val</span> <span class='hs-keyglyph'>::</span>
+<a name="line-356"></a>         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"int64Val"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-357"></a>           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"int64Val"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-358"></a>             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"int64Val"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-359"></a><span class='hs-definition'>int64Val</span>
+<a name="line-360"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-361"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"int64Val"</span><span class='hs-layout'>)</span>
+<a name="line-362"></a>
+<a name="line-363"></a><a name="intVal"></a><span class='hs-definition'>intVal</span> <span class='hs-keyglyph'>::</span>
+<a name="line-364"></a>       <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"intVal"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-365"></a>         <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"intVal"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-366"></a>           <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"intVal"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-367"></a><span class='hs-definition'>intVal</span>
+<a name="line-368"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-369"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"intVal"</span><span class='hs-layout'>)</span>
+<a name="line-370"></a>
+<a name="line-371"></a><a name="maybe'tensorShape"></a><span class='hs-definition'>maybe'tensorShape</span> <span class='hs-keyglyph'>::</span>
+<a name="line-372"></a>                  <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-373"></a>                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'tensorShape"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-374"></a>                    <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-375"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'tensorShape"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-376"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'tensorShape"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-377"></a><span class='hs-definition'>maybe'tensorShape</span>
+<a name="line-378"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-379"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-380"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'tensorShape"</span><span class='hs-layout'>)</span>
+<a name="line-381"></a>
+<a name="line-382"></a><a name="resourceHandleVal"></a><span class='hs-definition'>resourceHandleVal</span> <span class='hs-keyglyph'>::</span>
+<a name="line-383"></a>                  <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-384"></a>                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"resourceHandleVal"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-385"></a>                    <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-386"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"resourceHandleVal"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-387"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"resourceHandleVal"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-388"></a><span class='hs-definition'>resourceHandleVal</span>
+<a name="line-389"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-390"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-391"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"resourceHandleVal"</span><span class='hs-layout'>)</span>
+<a name="line-392"></a>
+<a name="line-393"></a><a name="scomplexVal"></a><span class='hs-definition'>scomplexVal</span> <span class='hs-keyglyph'>::</span>
+<a name="line-394"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"scomplexVal"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-395"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"scomplexVal"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-396"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"scomplexVal"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-397"></a><span class='hs-definition'>scomplexVal</span>
+<a name="line-398"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-399"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"scomplexVal"</span><span class='hs-layout'>)</span>
+<a name="line-400"></a>
+<a name="line-401"></a><a name="stringVal"></a><span class='hs-definition'>stringVal</span> <span class='hs-keyglyph'>::</span>
+<a name="line-402"></a>          <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"stringVal"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-403"></a>            <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"stringVal"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-404"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"stringVal"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-405"></a><span class='hs-definition'>stringVal</span>
+<a name="line-406"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-407"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"stringVal"</span><span class='hs-layout'>)</span>
+<a name="line-408"></a>
+<a name="line-409"></a><a name="tensorContent"></a><span class='hs-definition'>tensorContent</span> <span class='hs-keyglyph'>::</span>
+<a name="line-410"></a>              <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-411"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"tensorContent"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-412"></a>                <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-413"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"tensorContent"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-414"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"tensorContent"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-415"></a><span class='hs-definition'>tensorContent</span>
+<a name="line-416"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-417"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-418"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"tensorContent"</span><span class='hs-layout'>)</span>
+<a name="line-419"></a>
+<a name="line-420"></a><a name="tensorShape"></a><span class='hs-definition'>tensorShape</span> <span class='hs-keyglyph'>::</span>
+<a name="line-421"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"tensorShape"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-422"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"tensorShape"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-423"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"tensorShape"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-424"></a><span class='hs-definition'>tensorShape</span>
+<a name="line-425"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-426"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"tensorShape"</span><span class='hs-layout'>)</span>
+<a name="line-427"></a>
+<a name="line-428"></a><a name="versionNumber"></a><span class='hs-definition'>versionNumber</span> <span class='hs-keyglyph'>::</span>
+<a name="line-429"></a>              <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-430"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"versionNumber"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-431"></a>                <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-432"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"versionNumber"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-433"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"versionNumber"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-434"></a><span class='hs-definition'>versionNumber</span>
+<a name="line-435"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-436"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-437"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"versionNumber"</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-TensorDescription.html b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-TensorDescription.html
new file mode 100644
index 0000000..d9bcc9b
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-TensorDescription.html
@@ -0,0 +1,187 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Proto/Tensorflow/Core/Framework/TensorDescription.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>{- This file was auto-generated from tensorflow/core/framework/tensor_description.proto by the proto-lens-protoc program. -}</span>
+<a name="line-2"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables, DataKinds, TypeFamilies,
+<a name="line-3"></a>  MultiParamTypeClasses, FlexibleContexts, FlexibleInstances,
+<a name="line-4"></a>  PatternSynonyms #-}</span>
+<a name="line-5"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-unused-imports#-}</span>
+<a name="line-6"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Prelude</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-11"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-12"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-13"></a>       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-14"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-15"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-16"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-17"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-18"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-19"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-20"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-24"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-25"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-26"></a>       <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span>
+<a name="line-27"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span>
+<a name="line-28"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span>
+<a name="line-29"></a>
+<a name="line-30"></a><a name="TensorDescription"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>TensorDescription</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TensorDescription</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorDescription'dtype</span>
+<a name="line-31"></a>                                           <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span><span class='hs-layout'>,</span>
+<a name="line-32"></a>                                           <span class='hs-sel'>_TensorDescription'shape</span> <span class='hs-keyglyph'>::</span>
+<a name="line-33"></a>                                           <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-34"></a>                                             <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span><span class='hs-layout'>,</span>
+<a name="line-35"></a>                                           <span class='hs-sel'>_TensorDescription'allocationDescription</span> <span class='hs-keyglyph'>::</span>
+<a name="line-36"></a>                                           <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-37"></a>                                             <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span><span class='hs-layout'>}</span>
+<a name="line-38"></a>                       <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-39"></a>
+<a name="line-40"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dtype"</span> <span class='hs-conid'>TensorDescription</span> <span class='hs-keyglyph'>=</span>
+<a name="line-41"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span>
+<a name="line-42"></a>
+<a name="line-43"></a><a name="instance%20Data.ProtoLens.HasField%20%22dtype%22%20TensorDescription%20TensorDescription"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"dtype"</span> <span class='hs-conid'>TensorDescription</span>
+<a name="line-44"></a>         <span class='hs-conid'>TensorDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-45"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-46"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorDescription'dtype</span>
+<a name="line-47"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorDescription'dtype</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-48"></a>
+<a name="line-49"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"shape"</span> <span class='hs-conid'>TensorDescription</span> <span class='hs-keyglyph'>=</span>
+<a name="line-50"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span>
+<a name="line-51"></a>
+<a name="line-52"></a><a name="instance%20Data.ProtoLens.HasField%20%22shape%22%20TensorDescription%20TensorDescription"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"shape"</span> <span class='hs-conid'>TensorDescription</span>
+<a name="line-53"></a>         <span class='hs-conid'>TensorDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-54"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-55"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'shape</span>
+<a name="line-56"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-57"></a>
+<a name="line-58"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'shape"</span> <span class='hs-conid'>TensorDescription</span>
+<a name="line-59"></a>     <span class='hs-keyglyph'>=</span>
+<a name="line-60"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-61"></a>       <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span>
+<a name="line-62"></a>
+<a name="line-63"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'shape%22%20TensorDescription%20TensorDescription"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'shape"</span> <span class='hs-conid'>TensorDescription</span>
+<a name="line-64"></a>         <span class='hs-conid'>TensorDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-65"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-66"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorDescription'shape</span>
+<a name="line-67"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorDescription'shape</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-68"></a>
+<a name="line-69"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-70"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocationDescription"</span> <span class='hs-conid'>TensorDescription</span> <span class='hs-keyglyph'>=</span>
+<a name="line-71"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span>
+<a name="line-72"></a>
+<a name="line-73"></a><a name="instance%20Data.ProtoLens.HasField%20%22allocationDescription%22%20TensorDescription%20TensorDescription"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allocationDescription"</span>
+<a name="line-74"></a>         <span class='hs-conid'>TensorDescription</span> <span class='hs-conid'>TensorDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-75"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-76"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'allocationDescription</span>
+<a name="line-77"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-78"></a>
+<a name="line-79"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-80"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'allocationDescription"</span>
+<a name="line-81"></a>       <span class='hs-conid'>TensorDescription</span>
+<a name="line-82"></a>     <span class='hs-keyglyph'>=</span>
+<a name="line-83"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-84"></a>       <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span>
+<a name="line-85"></a>
+<a name="line-86"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'allocationDescription%22%20TensorDescription%20TensorDescription"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'allocationDescription"</span>
+<a name="line-87"></a>         <span class='hs-conid'>TensorDescription</span> <span class='hs-conid'>TensorDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-88"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-89"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span>
+<a name="line-90"></a>              <span class='hs-sel'>_TensorDescription'allocationDescription</span>
+<a name="line-91"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorDescription'allocationDescription</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-92"></a>
+<a name="line-93"></a><a name="instance%20Data.Default.Class.Default%20TensorDescription"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>TensorDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-94"></a>        <span class='hs-varid'>def</span>
+<a name="line-95"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TensorDescription</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorDescription'dtype</span> <span class='hs-keyglyph'>=</span>
+<a name="line-96"></a>                                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>,</span>
+<a name="line-97"></a>                              <span class='hs-sel'>_TensorDescription'shape</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-98"></a>                              <span class='hs-sel'>_TensorDescription'allocationDescription</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>}</span>
+<a name="line-99"></a>
+<a name="line-100"></a><a name="instance%20Data.ProtoLens.Message%20TensorDescription"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>TensorDescription</span> <span class='hs-keyword'>where</span>
+<a name="line-101"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-102"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>dtype__field_descriptor</span>
+<a name="line-103"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"dtype"</span>
+<a name="line-104"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>EnumField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-105"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-106"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span><span class='hs-varop'>.</span><span class='hs-conid'>DataType</span><span class='hs-layout'>)</span>
+<a name="line-107"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>dtype</span><span class='hs-layout'>)</span>
+<a name="line-108"></a>                <span class='hs-varid'>shape__field_descriptor</span>
+<a name="line-109"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"shape"</span>
+<a name="line-110"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-111"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-112"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShapeProto</span><span class='hs-layout'>)</span>
+<a name="line-113"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'shape</span><span class='hs-layout'>)</span>
+<a name="line-114"></a>                <span class='hs-varid'>allocationDescription__field_descriptor</span>
+<a name="line-115"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"allocation_description"</span>
+<a name="line-116"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-117"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-118"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span><span class='hs-varop'>.</span><span class='hs-conid'>AllocationDescription</span><span class='hs-layout'>)</span>
+<a name="line-119"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'allocationDescription</span><span class='hs-layout'>)</span>
+<a name="line-120"></a>              <span class='hs-keyword'>in</span>
+<a name="line-121"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-122"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-123"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>dtype__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-124"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>shape__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-125"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>allocationDescription__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-126"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-127"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"dtype"</span><span class='hs-layout'>,</span> <span class='hs-varid'>dtype__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-128"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"shape"</span><span class='hs-layout'>,</span> <span class='hs-varid'>shape__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-129"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"allocation_description"</span><span class='hs-layout'>,</span>
+<a name="line-130"></a>                     <span class='hs-varid'>allocationDescription__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-131"></a>
+<a name="line-132"></a><a name="allocationDescription"></a><span class='hs-definition'>allocationDescription</span> <span class='hs-keyglyph'>::</span>
+<a name="line-133"></a>                      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-134"></a>                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allocationDescription"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-135"></a>                        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-136"></a>                          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocationDescription"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-137"></a>                          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocationDescription"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-138"></a><span class='hs-definition'>allocationDescription</span>
+<a name="line-139"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-140"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-141"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"allocationDescription"</span><span class='hs-layout'>)</span>
+<a name="line-142"></a>
+<a name="line-143"></a><a name="dtype"></a><span class='hs-definition'>dtype</span> <span class='hs-keyglyph'>::</span>
+<a name="line-144"></a>      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"dtype"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-145"></a>        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dtype"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-146"></a>          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dtype"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-147"></a><span class='hs-definition'>dtype</span>
+<a name="line-148"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-149"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"dtype"</span><span class='hs-layout'>)</span>
+<a name="line-150"></a>
+<a name="line-151"></a><a name="maybe'allocationDescription"></a><span class='hs-definition'>maybe'allocationDescription</span> <span class='hs-keyglyph'>::</span>
+<a name="line-152"></a>                            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-153"></a>                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'allocationDescription"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-154"></a>                              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-155"></a>                                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'allocationDescription"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-156"></a>                                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'allocationDescription"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-157"></a><span class='hs-definition'>maybe'allocationDescription</span>
+<a name="line-158"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-159"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-160"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'allocationDescription"</span><span class='hs-layout'>)</span>
+<a name="line-161"></a>
+<a name="line-162"></a><a name="maybe'shape"></a><span class='hs-definition'>maybe'shape</span> <span class='hs-keyglyph'>::</span>
+<a name="line-163"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'shape"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-164"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'shape"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-165"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'shape"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-166"></a><span class='hs-definition'>maybe'shape</span>
+<a name="line-167"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-168"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'shape"</span><span class='hs-layout'>)</span>
+<a name="line-169"></a>
+<a name="line-170"></a><a name="shape"></a><span class='hs-definition'>shape</span> <span class='hs-keyglyph'>::</span>
+<a name="line-171"></a>      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"shape"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-172"></a>        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"shape"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-173"></a>          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"shape"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-174"></a><span class='hs-definition'>shape</span>
+<a name="line-175"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-176"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"shape"</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-TensorShape.html b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-TensorShape.html
new file mode 100644
index 0000000..805a823
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-TensorShape.html
@@ -0,0 +1,166 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Proto/Tensorflow/Core/Framework/TensorShape.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>{- This file was auto-generated from tensorflow/core/framework/tensor_shape.proto by the proto-lens-protoc program. -}</span>
+<a name="line-2"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables, DataKinds, TypeFamilies,
+<a name="line-3"></a>  MultiParamTypeClasses, FlexibleContexts, FlexibleInstances,
+<a name="line-4"></a>  PatternSynonyms #-}</span>
+<a name="line-5"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-unused-imports#-}</span>
+<a name="line-6"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>TensorShape</span> <span class='hs-keyword'>where</span>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Prelude</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-11"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-12"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-13"></a>       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-14"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-15"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-16"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-17"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-18"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-19"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-20"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-24"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-25"></a>
+<a name="line-26"></a><a name="TensorShapeProto"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>TensorShapeProto</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TensorShapeProto</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorShapeProto'dim</span> <span class='hs-keyglyph'>::</span>
+<a name="line-27"></a>                                         <span class='hs-keyglyph'>[</span><span class='hs-conid'>TensorShapeProto'Dim</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-28"></a>                                         <span class='hs-sel'>_TensorShapeProto'unknownRank</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>}</span>
+<a name="line-29"></a>                      <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-30"></a>
+<a name="line-31"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dim"</span> <span class='hs-conid'>TensorShapeProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-32"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>TensorShapeProto'Dim</span><span class='hs-keyglyph'>]</span>
+<a name="line-33"></a>
+<a name="line-34"></a><a name="instance%20Data.ProtoLens.HasField%20%22dim%22%20TensorShapeProto%20TensorShapeProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"dim"</span> <span class='hs-conid'>TensorShapeProto</span>
+<a name="line-35"></a>         <span class='hs-conid'>TensorShapeProto</span> <span class='hs-keyword'>where</span>
+<a name="line-36"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-37"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorShapeProto'dim</span>
+<a name="line-38"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorShapeProto'dim</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-39"></a>
+<a name="line-40"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"unknownRank"</span> <span class='hs-conid'>TensorShapeProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-41"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-42"></a>
+<a name="line-43"></a><a name="instance%20Data.ProtoLens.HasField%20%22unknownRank%22%20TensorShapeProto%20TensorShapeProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"unknownRank"</span> <span class='hs-conid'>TensorShapeProto</span>
+<a name="line-44"></a>         <span class='hs-conid'>TensorShapeProto</span> <span class='hs-keyword'>where</span>
+<a name="line-45"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-46"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorShapeProto'unknownRank</span>
+<a name="line-47"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorShapeProto'unknownRank</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-48"></a>
+<a name="line-49"></a><a name="instance%20Data.Default.Class.Default%20TensorShapeProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>TensorShapeProto</span> <span class='hs-keyword'>where</span>
+<a name="line-50"></a>        <span class='hs-varid'>def</span>
+<a name="line-51"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TensorShapeProto</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorShapeProto'dim</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-52"></a>                             <span class='hs-sel'>_TensorShapeProto'unknownRank</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-53"></a>
+<a name="line-54"></a><a name="instance%20Data.ProtoLens.Message%20TensorShapeProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>TensorShapeProto</span> <span class='hs-keyword'>where</span>
+<a name="line-55"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-56"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>dim__field_descriptor</span>
+<a name="line-57"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"dim"</span>
+<a name="line-58"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-59"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>TensorShapeProto'Dim</span><span class='hs-layout'>)</span>
+<a name="line-60"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>dim</span><span class='hs-layout'>)</span>
+<a name="line-61"></a>                <span class='hs-varid'>unknownRank__field_descriptor</span>
+<a name="line-62"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"unknown_rank"</span>
+<a name="line-63"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-64"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-65"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>unknownRank</span><span class='hs-layout'>)</span>
+<a name="line-66"></a>              <span class='hs-keyword'>in</span>
+<a name="line-67"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-68"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-69"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>dim__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-70"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>unknownRank__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-71"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-72"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"dim"</span><span class='hs-layout'>,</span> <span class='hs-varid'>dim__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-73"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"unknown_rank"</span><span class='hs-layout'>,</span> <span class='hs-varid'>unknownRank__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-74"></a>
+<a name="line-75"></a><a name="TensorShapeProto'Dim"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>TensorShapeProto'Dim</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TensorShapeProto'Dim</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorShapeProto'Dim'size</span>
+<a name="line-76"></a>                                                 <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-77"></a>                                                 <span class='hs-sel'>_TensorShapeProto'Dim'name</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>}</span>
+<a name="line-78"></a>                          <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-79"></a>
+<a name="line-80"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"size"</span> <span class='hs-conid'>TensorShapeProto'Dim</span> <span class='hs-keyglyph'>=</span>
+<a name="line-81"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-82"></a>
+<a name="line-83"></a><a name="instance%20Data.ProtoLens.HasField%20%22size%22%20TensorShapeProto'Dim%20TensorShapeProto'Dim"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"size"</span> <span class='hs-conid'>TensorShapeProto'Dim</span>
+<a name="line-84"></a>         <span class='hs-conid'>TensorShapeProto'Dim</span> <span class='hs-keyword'>where</span>
+<a name="line-85"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-86"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorShapeProto'Dim'size</span>
+<a name="line-87"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorShapeProto'Dim'size</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-88"></a>
+<a name="line-89"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-conid'>TensorShapeProto'Dim</span> <span class='hs-keyglyph'>=</span>
+<a name="line-90"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-91"></a>
+<a name="line-92"></a><a name="instance%20Data.ProtoLens.HasField%20%22name%22%20TensorShapeProto'Dim%20TensorShapeProto'Dim"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"name"</span> <span class='hs-conid'>TensorShapeProto'Dim</span>
+<a name="line-93"></a>         <span class='hs-conid'>TensorShapeProto'Dim</span> <span class='hs-keyword'>where</span>
+<a name="line-94"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-95"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_TensorShapeProto'Dim'name</span>
+<a name="line-96"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorShapeProto'Dim'name</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-97"></a>
+<a name="line-98"></a><a name="instance%20Data.Default.Class.Default%20TensorShapeProto'Dim"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>TensorShapeProto'Dim</span> <span class='hs-keyword'>where</span>
+<a name="line-99"></a>        <span class='hs-varid'>def</span>
+<a name="line-100"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>TensorShapeProto'Dim</span><span class='hs-layout'>{</span><span class='hs-sel'>_TensorShapeProto'Dim'size</span> <span class='hs-keyglyph'>=</span>
+<a name="line-101"></a>                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-102"></a>                                 <span class='hs-sel'>_TensorShapeProto'Dim'name</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-103"></a>
+<a name="line-104"></a><a name="instance%20Data.ProtoLens.Message%20TensorShapeProto'Dim"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>TensorShapeProto'Dim</span> <span class='hs-keyword'>where</span>
+<a name="line-105"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-106"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>size__field_descriptor</span>
+<a name="line-107"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"size"</span>
+<a name="line-108"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-109"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-110"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>size</span><span class='hs-layout'>)</span>
+<a name="line-111"></a>                <span class='hs-varid'>name__field_descriptor</span>
+<a name="line-112"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"name"</span>
+<a name="line-113"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-114"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-115"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>name</span><span class='hs-layout'>)</span>
+<a name="line-116"></a>              <span class='hs-keyword'>in</span>
+<a name="line-117"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-118"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-119"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>size__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-120"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>name__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-121"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-122"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"size"</span><span class='hs-layout'>,</span> <span class='hs-varid'>size__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-123"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"name"</span><span class='hs-layout'>,</span> <span class='hs-varid'>name__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-124"></a>
+<a name="line-125"></a><a name="dim"></a><span class='hs-definition'>dim</span> <span class='hs-keyglyph'>::</span>
+<a name="line-126"></a>    <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"dim"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-127"></a>      <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dim"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-128"></a>        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"dim"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-129"></a><span class='hs-definition'>dim</span>
+<a name="line-130"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-131"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"dim"</span><span class='hs-layout'>)</span>
+<a name="line-132"></a>
+<a name="line-133"></a><a name="name"></a><span class='hs-definition'>name</span> <span class='hs-keyglyph'>::</span>
+<a name="line-134"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-135"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-136"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"name"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-137"></a><span class='hs-definition'>name</span>
+<a name="line-138"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-139"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"name"</span><span class='hs-layout'>)</span>
+<a name="line-140"></a>
+<a name="line-141"></a><a name="size"></a><span class='hs-definition'>size</span> <span class='hs-keyglyph'>::</span>
+<a name="line-142"></a>     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"size"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-143"></a>       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"size"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-144"></a>         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"size"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-145"></a><span class='hs-definition'>size</span>
+<a name="line-146"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-147"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"size"</span><span class='hs-layout'>)</span>
+<a name="line-148"></a>
+<a name="line-149"></a><a name="unknownRank"></a><span class='hs-definition'>unknownRank</span> <span class='hs-keyglyph'>::</span>
+<a name="line-150"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"unknownRank"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-151"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"unknownRank"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-152"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"unknownRank"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-153"></a><span class='hs-definition'>unknownRank</span>
+<a name="line-154"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-155"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"unknownRank"</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Types.html b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Types.html
new file mode 100644
index 0000000..cb2dca2
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Types.html
@@ -0,0 +1,355 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Proto/Tensorflow/Core/Framework/Types.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>{- This file was auto-generated from tensorflow/core/framework/types.proto by the proto-lens-protoc program. -}</span>
+<a name="line-2"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables, DataKinds, TypeFamilies,
+<a name="line-3"></a>  MultiParamTypeClasses, FlexibleContexts, FlexibleInstances,
+<a name="line-4"></a>  PatternSynonyms #-}</span>
+<a name="line-5"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-unused-imports#-}</span>
+<a name="line-6"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span> <span class='hs-keyword'>where</span>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Prelude</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-11"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-12"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-13"></a>       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-14"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-15"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-16"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-17"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-18"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-19"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-20"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-24"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-25"></a>
+<a name="line-26"></a><a name="DataType"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>DataType</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INVALID</span>
+<a name="line-27"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_FLOAT</span>
+<a name="line-28"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_DOUBLE</span>
+<a name="line-29"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_INT32</span>
+<a name="line-30"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_UINT8</span>
+<a name="line-31"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_INT16</span>
+<a name="line-32"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_INT8</span>
+<a name="line-33"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_STRING</span>
+<a name="line-34"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_COMPLEX64</span>
+<a name="line-35"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_INT64</span>
+<a name="line-36"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_BOOL</span>
+<a name="line-37"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_QINT8</span>
+<a name="line-38"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_QUINT8</span>
+<a name="line-39"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_QINT32</span>
+<a name="line-40"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_BFLOAT16</span>
+<a name="line-41"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_QINT16</span>
+<a name="line-42"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_QUINT16</span>
+<a name="line-43"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_UINT16</span>
+<a name="line-44"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_COMPLEX128</span>
+<a name="line-45"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_HALF</span>
+<a name="line-46"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_RESOURCE</span>
+<a name="line-47"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_FLOAT_REF</span>
+<a name="line-48"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_DOUBLE_REF</span>
+<a name="line-49"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_INT32_REF</span>
+<a name="line-50"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_UINT8_REF</span>
+<a name="line-51"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_INT16_REF</span>
+<a name="line-52"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_INT8_REF</span>
+<a name="line-53"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_STRING_REF</span>
+<a name="line-54"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_COMPLEX64_REF</span>
+<a name="line-55"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_INT64_REF</span>
+<a name="line-56"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_BOOL_REF</span>
+<a name="line-57"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_QINT8_REF</span>
+<a name="line-58"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_QUINT8_REF</span>
+<a name="line-59"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_QINT32_REF</span>
+<a name="line-60"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_BFLOAT16_REF</span>
+<a name="line-61"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_QINT16_REF</span>
+<a name="line-62"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_QUINT16_REF</span>
+<a name="line-63"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_UINT16_REF</span>
+<a name="line-64"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_COMPLEX128_REF</span>
+<a name="line-65"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_HALF_REF</span>
+<a name="line-66"></a>              <span class='hs-keyglyph'>|</span> <span class='hs-conid'>DT_RESOURCE_REF</span>
+<a name="line-67"></a>              <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-68"></a>
+<a name="line-69"></a><a name="instance%20Data.Default.Class.Default%20DataType"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>DataType</span> <span class='hs-keyword'>where</span>
+<a name="line-70"></a>        <span class='hs-varid'>def</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INVALID</span>
+<a name="line-71"></a>
+<a name="line-72"></a><a name="instance%20Data.ProtoLens.FieldDefault%20DataType"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDefault</span> <span class='hs-conid'>DataType</span> <span class='hs-keyword'>where</span>
+<a name="line-73"></a>        <span class='hs-varid'>fieldDefault</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INVALID</span>
+<a name="line-74"></a>
+<a name="line-75"></a><a name="instance%20Data.ProtoLens.MessageEnum%20DataType"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageEnum</span> <span class='hs-conid'>DataType</span> <span class='hs-keyword'>where</span>
+<a name="line-76"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>0</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INVALID</span>
+<a name="line-77"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>1</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_FLOAT</span>
+<a name="line-78"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>2</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_DOUBLE</span>
+<a name="line-79"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>3</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INT32</span>
+<a name="line-80"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>4</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_UINT8</span>
+<a name="line-81"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>5</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INT16</span>
+<a name="line-82"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>6</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INT8</span>
+<a name="line-83"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>7</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_STRING</span>
+<a name="line-84"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_COMPLEX64</span>
+<a name="line-85"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>9</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INT64</span>
+<a name="line-86"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>10</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_BOOL</span>
+<a name="line-87"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>11</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QINT8</span>
+<a name="line-88"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>12</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QUINT8</span>
+<a name="line-89"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>13</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QINT32</span>
+<a name="line-90"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>14</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_BFLOAT16</span>
+<a name="line-91"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>15</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QINT16</span>
+<a name="line-92"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QUINT16</span>
+<a name="line-93"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>17</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_UINT16</span>
+<a name="line-94"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>18</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_COMPLEX128</span>
+<a name="line-95"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>19</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_HALF</span>
+<a name="line-96"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>20</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_RESOURCE</span>
+<a name="line-97"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>101</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_FLOAT_REF</span>
+<a name="line-98"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>102</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_DOUBLE_REF</span>
+<a name="line-99"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>103</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INT32_REF</span>
+<a name="line-100"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>104</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_UINT8_REF</span>
+<a name="line-101"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>105</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INT16_REF</span>
+<a name="line-102"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>106</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INT8_REF</span>
+<a name="line-103"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>107</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_STRING_REF</span>
+<a name="line-104"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>108</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_COMPLEX64_REF</span>
+<a name="line-105"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>109</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INT64_REF</span>
+<a name="line-106"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>110</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_BOOL_REF</span>
+<a name="line-107"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>111</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QINT8_REF</span>
+<a name="line-108"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>112</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QUINT8_REF</span>
+<a name="line-109"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>113</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QINT32_REF</span>
+<a name="line-110"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>114</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_BFLOAT16_REF</span>
+<a name="line-111"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>115</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QINT16_REF</span>
+<a name="line-112"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>116</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QUINT16_REF</span>
+<a name="line-113"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>117</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_UINT16_REF</span>
+<a name="line-114"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>118</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_COMPLEX128_REF</span>
+<a name="line-115"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>119</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_HALF_REF</span>
+<a name="line-116"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>120</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_RESOURCE_REF</span>
+<a name="line-117"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span>
+<a name="line-118"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_INVALID</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_INVALID"</span>
+<a name="line-119"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_FLOAT</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_FLOAT"</span>
+<a name="line-120"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_DOUBLE</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_DOUBLE"</span>
+<a name="line-121"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_INT32</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_INT32"</span>
+<a name="line-122"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_UINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_UINT8"</span>
+<a name="line-123"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_INT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_INT16"</span>
+<a name="line-124"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_INT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_INT8"</span>
+<a name="line-125"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_STRING</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_STRING"</span>
+<a name="line-126"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_COMPLEX64</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_COMPLEX64"</span>
+<a name="line-127"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_INT64</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_INT64"</span>
+<a name="line-128"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_BOOL</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_BOOL"</span>
+<a name="line-129"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_QINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_QINT8"</span>
+<a name="line-130"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_QUINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_QUINT8"</span>
+<a name="line-131"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_QINT32</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_QINT32"</span>
+<a name="line-132"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_BFLOAT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_BFLOAT16"</span>
+<a name="line-133"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_QINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_QINT16"</span>
+<a name="line-134"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_QUINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_QUINT16"</span>
+<a name="line-135"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_UINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_UINT16"</span>
+<a name="line-136"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_COMPLEX128</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_COMPLEX128"</span>
+<a name="line-137"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_HALF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_HALF"</span>
+<a name="line-138"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_RESOURCE</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_RESOURCE"</span>
+<a name="line-139"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_FLOAT_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_FLOAT_REF"</span>
+<a name="line-140"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_DOUBLE_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_DOUBLE_REF"</span>
+<a name="line-141"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_INT32_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_INT32_REF"</span>
+<a name="line-142"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_UINT8_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_UINT8_REF"</span>
+<a name="line-143"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_INT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_INT16_REF"</span>
+<a name="line-144"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_INT8_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_INT8_REF"</span>
+<a name="line-145"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_STRING_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_STRING_REF"</span>
+<a name="line-146"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_COMPLEX64_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_COMPLEX64_REF"</span>
+<a name="line-147"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_INT64_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_INT64_REF"</span>
+<a name="line-148"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_BOOL_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_BOOL_REF"</span>
+<a name="line-149"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_QINT8_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_QINT8_REF"</span>
+<a name="line-150"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_QUINT8_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_QUINT8_REF"</span>
+<a name="line-151"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_QINT32_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_QINT32_REF"</span>
+<a name="line-152"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_BFLOAT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_BFLOAT16_REF"</span>
+<a name="line-153"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_QINT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_QINT16_REF"</span>
+<a name="line-154"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_QUINT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_QUINT16_REF"</span>
+<a name="line-155"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_UINT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_UINT16_REF"</span>
+<a name="line-156"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_COMPLEX128_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_COMPLEX128_REF"</span>
+<a name="line-157"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_HALF_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_HALF_REF"</span>
+<a name="line-158"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>DT_RESOURCE_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"DT_RESOURCE_REF"</span>
+<a name="line-159"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_INVALID"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INVALID</span>
+<a name="line-160"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_FLOAT"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_FLOAT</span>
+<a name="line-161"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_DOUBLE"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_DOUBLE</span>
+<a name="line-162"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_INT32"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INT32</span>
+<a name="line-163"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_UINT8"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_UINT8</span>
+<a name="line-164"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_INT16"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INT16</span>
+<a name="line-165"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_INT8"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INT8</span>
+<a name="line-166"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_STRING"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_STRING</span>
+<a name="line-167"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_COMPLEX64"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_COMPLEX64</span>
+<a name="line-168"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_INT64"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INT64</span>
+<a name="line-169"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_BOOL"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_BOOL</span>
+<a name="line-170"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_QINT8"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QINT8</span>
+<a name="line-171"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_QUINT8"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QUINT8</span>
+<a name="line-172"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_QINT32"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QINT32</span>
+<a name="line-173"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_BFLOAT16"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_BFLOAT16</span>
+<a name="line-174"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_QINT16"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QINT16</span>
+<a name="line-175"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_QUINT16"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QUINT16</span>
+<a name="line-176"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_UINT16"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_UINT16</span>
+<a name="line-177"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_COMPLEX128"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_COMPLEX128</span>
+<a name="line-178"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_HALF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_HALF</span>
+<a name="line-179"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_RESOURCE"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_RESOURCE</span>
+<a name="line-180"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_FLOAT_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_FLOAT_REF</span>
+<a name="line-181"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_DOUBLE_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_DOUBLE_REF</span>
+<a name="line-182"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_INT32_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INT32_REF</span>
+<a name="line-183"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_UINT8_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_UINT8_REF</span>
+<a name="line-184"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_INT16_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INT16_REF</span>
+<a name="line-185"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_INT8_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INT8_REF</span>
+<a name="line-186"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_STRING_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_STRING_REF</span>
+<a name="line-187"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_COMPLEX64_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_COMPLEX64_REF</span>
+<a name="line-188"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_INT64_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_INT64_REF</span>
+<a name="line-189"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_BOOL_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_BOOL_REF</span>
+<a name="line-190"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_QINT8_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QINT8_REF</span>
+<a name="line-191"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_QUINT8_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QUINT8_REF</span>
+<a name="line-192"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_QINT32_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QINT32_REF</span>
+<a name="line-193"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_BFLOAT16_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_BFLOAT16_REF</span>
+<a name="line-194"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_QINT16_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QINT16_REF</span>
+<a name="line-195"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_QUINT16_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_QUINT16_REF</span>
+<a name="line-196"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_UINT16_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_UINT16_REF</span>
+<a name="line-197"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_COMPLEX128_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_COMPLEX128_REF</span>
+<a name="line-198"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_HALF_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_HALF_REF</span>
+<a name="line-199"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"DT_RESOURCE_REF"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>DT_RESOURCE_REF</span>
+<a name="line-200"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span>
+<a name="line-201"></a>
+<a name="line-202"></a><a name="instance%20Prelude.Enum%20DataType"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span> <span class='hs-conid'>DataType</span> <span class='hs-keyword'>where</span>
+<a name="line-203"></a>        <span class='hs-varid'>toEnum</span> <span class='hs-varid'>k__</span>
+<a name="line-204"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>maybe</span>
+<a name="line-205"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>error</span>
+<a name="line-206"></a>                 <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.++</span><span class='hs-layout'>)</span> <span class='hs-str'>"toEnum: unknown value for enum DataType: "</span>
+<a name="line-207"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>show</span> <span class='hs-varid'>k__</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-208"></a>              <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>id</span>
+<a name="line-209"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeToEnum</span> <span class='hs-varid'>k__</span><span class='hs-layout'>)</span>
+<a name="line-210"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_INVALID</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>0</span>
+<a name="line-211"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_FLOAT</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>1</span>
+<a name="line-212"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_DOUBLE</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>2</span>
+<a name="line-213"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_INT32</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>3</span>
+<a name="line-214"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_UINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>4</span>
+<a name="line-215"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_INT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>5</span>
+<a name="line-216"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_INT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>6</span>
+<a name="line-217"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_STRING</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>7</span>
+<a name="line-218"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_COMPLEX64</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>8</span>
+<a name="line-219"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_INT64</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>9</span>
+<a name="line-220"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_BOOL</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>10</span>
+<a name="line-221"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_QINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>11</span>
+<a name="line-222"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_QUINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>12</span>
+<a name="line-223"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_QINT32</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>13</span>
+<a name="line-224"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_BFLOAT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>14</span>
+<a name="line-225"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_QINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>15</span>
+<a name="line-226"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_QUINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>16</span>
+<a name="line-227"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_UINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>17</span>
+<a name="line-228"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_COMPLEX128</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>18</span>
+<a name="line-229"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_HALF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>19</span>
+<a name="line-230"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_RESOURCE</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>20</span>
+<a name="line-231"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_FLOAT_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>101</span>
+<a name="line-232"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_DOUBLE_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>102</span>
+<a name="line-233"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_INT32_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>103</span>
+<a name="line-234"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_UINT8_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>104</span>
+<a name="line-235"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_INT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>105</span>
+<a name="line-236"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_INT8_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>106</span>
+<a name="line-237"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_STRING_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>107</span>
+<a name="line-238"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_COMPLEX64_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>108</span>
+<a name="line-239"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_INT64_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>109</span>
+<a name="line-240"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_BOOL_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>110</span>
+<a name="line-241"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_QINT8_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>111</span>
+<a name="line-242"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_QUINT8_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>112</span>
+<a name="line-243"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_QINT32_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>113</span>
+<a name="line-244"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_BFLOAT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>114</span>
+<a name="line-245"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_QINT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>115</span>
+<a name="line-246"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_QUINT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>116</span>
+<a name="line-247"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_UINT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>117</span>
+<a name="line-248"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_COMPLEX128_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>118</span>
+<a name="line-249"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_HALF_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>119</span>
+<a name="line-250"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>DT_RESOURCE_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>120</span>
+<a name="line-251"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_RESOURCE_REF</span>
+<a name="line-252"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>error</span>
+<a name="line-253"></a>              <span class='hs-str'>"Ident \"DataType\".Ident \"succ\": bad argument Ident \"DT_RESOURCE_REF\". This value would be out of bounds."</span>
+<a name="line-254"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_INVALID</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_FLOAT</span>
+<a name="line-255"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_FLOAT</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_DOUBLE</span>
+<a name="line-256"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_DOUBLE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT32</span>
+<a name="line-257"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_INT32</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_UINT8</span>
+<a name="line-258"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_UINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT16</span>
+<a name="line-259"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_INT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT8</span>
+<a name="line-260"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_INT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_STRING</span>
+<a name="line-261"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_STRING</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_COMPLEX64</span>
+<a name="line-262"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_COMPLEX64</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT64</span>
+<a name="line-263"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_INT64</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_BOOL</span>
+<a name="line-264"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_BOOL</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QINT8</span>
+<a name="line-265"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_QINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QUINT8</span>
+<a name="line-266"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_QUINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QINT32</span>
+<a name="line-267"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_QINT32</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_BFLOAT16</span>
+<a name="line-268"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_BFLOAT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QINT16</span>
+<a name="line-269"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_QINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QUINT16</span>
+<a name="line-270"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_QUINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_UINT16</span>
+<a name="line-271"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_UINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_COMPLEX128</span>
+<a name="line-272"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_COMPLEX128</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_HALF</span>
+<a name="line-273"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_HALF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_RESOURCE</span>
+<a name="line-274"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_RESOURCE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_FLOAT_REF</span>
+<a name="line-275"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_FLOAT_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_DOUBLE_REF</span>
+<a name="line-276"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_DOUBLE_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT32_REF</span>
+<a name="line-277"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_INT32_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_UINT8_REF</span>
+<a name="line-278"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_UINT8_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT16_REF</span>
+<a name="line-279"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_INT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT8_REF</span>
+<a name="line-280"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_INT8_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_STRING_REF</span>
+<a name="line-281"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_STRING_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_COMPLEX64_REF</span>
+<a name="line-282"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_COMPLEX64_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT64_REF</span>
+<a name="line-283"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_INT64_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_BOOL_REF</span>
+<a name="line-284"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_BOOL_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QINT8_REF</span>
+<a name="line-285"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_QINT8_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QUINT8_REF</span>
+<a name="line-286"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_QUINT8_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QINT32_REF</span>
+<a name="line-287"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_QINT32_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_BFLOAT16_REF</span>
+<a name="line-288"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_BFLOAT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QINT16_REF</span>
+<a name="line-289"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_QINT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QUINT16_REF</span>
+<a name="line-290"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_QUINT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_UINT16_REF</span>
+<a name="line-291"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_UINT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_COMPLEX128_REF</span>
+<a name="line-292"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_COMPLEX128_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_HALF_REF</span>
+<a name="line-293"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>DT_HALF_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_RESOURCE_REF</span>
+<a name="line-294"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_INVALID</span>
+<a name="line-295"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>error</span>
+<a name="line-296"></a>              <span class='hs-str'>"Ident \"DataType\".Ident \"pred\": bad argument Ident \"DT_INVALID\". This value would be out of bounds."</span>
+<a name="line-297"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_FLOAT</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INVALID</span>
+<a name="line-298"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_DOUBLE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_FLOAT</span>
+<a name="line-299"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_INT32</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_DOUBLE</span>
+<a name="line-300"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_UINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT32</span>
+<a name="line-301"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_INT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_UINT8</span>
+<a name="line-302"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_INT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT16</span>
+<a name="line-303"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_STRING</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT8</span>
+<a name="line-304"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_COMPLEX64</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_STRING</span>
+<a name="line-305"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_INT64</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_COMPLEX64</span>
+<a name="line-306"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_BOOL</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT64</span>
+<a name="line-307"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_QINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_BOOL</span>
+<a name="line-308"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_QUINT8</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QINT8</span>
+<a name="line-309"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_QINT32</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QUINT8</span>
+<a name="line-310"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_BFLOAT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QINT32</span>
+<a name="line-311"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_QINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_BFLOAT16</span>
+<a name="line-312"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_QUINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QINT16</span>
+<a name="line-313"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_UINT16</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QUINT16</span>
+<a name="line-314"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_COMPLEX128</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_UINT16</span>
+<a name="line-315"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_HALF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_COMPLEX128</span>
+<a name="line-316"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_RESOURCE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_HALF</span>
+<a name="line-317"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_FLOAT_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_RESOURCE</span>
+<a name="line-318"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_DOUBLE_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_FLOAT_REF</span>
+<a name="line-319"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_INT32_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_DOUBLE_REF</span>
+<a name="line-320"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_UINT8_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT32_REF</span>
+<a name="line-321"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_INT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_UINT8_REF</span>
+<a name="line-322"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_INT8_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT16_REF</span>
+<a name="line-323"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_STRING_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT8_REF</span>
+<a name="line-324"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_COMPLEX64_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_STRING_REF</span>
+<a name="line-325"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_INT64_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_COMPLEX64_REF</span>
+<a name="line-326"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_BOOL_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INT64_REF</span>
+<a name="line-327"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_QINT8_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_BOOL_REF</span>
+<a name="line-328"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_QUINT8_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QINT8_REF</span>
+<a name="line-329"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_QINT32_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QUINT8_REF</span>
+<a name="line-330"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_BFLOAT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QINT32_REF</span>
+<a name="line-331"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_QINT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_BFLOAT16_REF</span>
+<a name="line-332"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_QUINT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QINT16_REF</span>
+<a name="line-333"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_UINT16_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_QUINT16_REF</span>
+<a name="line-334"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_COMPLEX128_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_UINT16_REF</span>
+<a name="line-335"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_HALF_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_COMPLEX128_REF</span>
+<a name="line-336"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>DT_RESOURCE_REF</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_HALF_REF</span>
+<a name="line-337"></a>        <span class='hs-varid'>enumFrom</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span><span class='hs-varop'>.</span><span class='hs-varid'>messageEnumFrom</span>
+<a name="line-338"></a>        <span class='hs-varid'>enumFromTo</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span><span class='hs-varop'>.</span><span class='hs-varid'>messageEnumFromTo</span>
+<a name="line-339"></a>        <span class='hs-varid'>enumFromThen</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span><span class='hs-varop'>.</span><span class='hs-varid'>messageEnumFromThen</span>
+<a name="line-340"></a>        <span class='hs-varid'>enumFromThenTo</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span><span class='hs-varop'>.</span><span class='hs-varid'>messageEnumFromThenTo</span>
+<a name="line-341"></a>
+<a name="line-342"></a><a name="instance%20Prelude.Bounded%20DataType"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bounded</span> <span class='hs-conid'>DataType</span> <span class='hs-keyword'>where</span>
+<a name="line-343"></a>        <span class='hs-varid'>minBound</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_INVALID</span>
+<a name="line-344"></a>        <span class='hs-varid'>maxBound</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DT_RESOURCE_REF</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Versions.html b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Versions.html
new file mode 100644
index 0000000..aa80fb3
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Framework-Versions.html
@@ -0,0 +1,128 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Proto/Tensorflow/Core/Framework/Versions.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>{- This file was auto-generated from tensorflow/core/framework/versions.proto by the proto-lens-protoc program. -}</span>
+<a name="line-2"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables, DataKinds, TypeFamilies,
+<a name="line-3"></a>  MultiParamTypeClasses, FlexibleContexts, FlexibleInstances,
+<a name="line-4"></a>  PatternSynonyms #-}</span>
+<a name="line-5"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-unused-imports#-}</span>
+<a name="line-6"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Versions</span> <span class='hs-keyword'>where</span>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Prelude</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-11"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-12"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-13"></a>       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-14"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-15"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-16"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-17"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-18"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-19"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-20"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-24"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-25"></a>
+<a name="line-26"></a><a name="VersionDef"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>VersionDef</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>VersionDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_VersionDef'producer</span> <span class='hs-keyglyph'>::</span>
+<a name="line-27"></a>                             <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-28"></a>                             <span class='hs-sel'>_VersionDef'minConsumer</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-29"></a>                             <span class='hs-sel'>_VersionDef'badConsumers</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>}</span>
+<a name="line-30"></a>                <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-31"></a>
+<a name="line-32"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"producer"</span> <span class='hs-conid'>VersionDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-33"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-34"></a>
+<a name="line-35"></a><a name="instance%20Data.ProtoLens.HasField%20%22producer%22%20VersionDef%20VersionDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"producer"</span> <span class='hs-conid'>VersionDef</span> <span class='hs-conid'>VersionDef</span>
+<a name="line-36"></a>         <span class='hs-keyword'>where</span>
+<a name="line-37"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-38"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_VersionDef'producer</span>
+<a name="line-39"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_VersionDef'producer</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-40"></a>
+<a name="line-41"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"minConsumer"</span> <span class='hs-conid'>VersionDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-42"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-43"></a>
+<a name="line-44"></a><a name="instance%20Data.ProtoLens.HasField%20%22minConsumer%22%20VersionDef%20VersionDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"minConsumer"</span> <span class='hs-conid'>VersionDef</span>
+<a name="line-45"></a>         <span class='hs-conid'>VersionDef</span> <span class='hs-keyword'>where</span>
+<a name="line-46"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-47"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_VersionDef'minConsumer</span>
+<a name="line-48"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_VersionDef'minConsumer</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-49"></a>
+<a name="line-50"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"badConsumers"</span> <span class='hs-conid'>VersionDef</span> <span class='hs-keyglyph'>=</span>
+<a name="line-51"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-keyglyph'>]</span>
+<a name="line-52"></a>
+<a name="line-53"></a><a name="instance%20Data.ProtoLens.HasField%20%22badConsumers%22%20VersionDef%20VersionDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"badConsumers"</span> <span class='hs-conid'>VersionDef</span>
+<a name="line-54"></a>         <span class='hs-conid'>VersionDef</span> <span class='hs-keyword'>where</span>
+<a name="line-55"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-56"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_VersionDef'badConsumers</span>
+<a name="line-57"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_VersionDef'badConsumers</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-58"></a>
+<a name="line-59"></a><a name="instance%20Data.Default.Class.Default%20VersionDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>VersionDef</span> <span class='hs-keyword'>where</span>
+<a name="line-60"></a>        <span class='hs-varid'>def</span>
+<a name="line-61"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>VersionDef</span><span class='hs-layout'>{</span><span class='hs-sel'>_VersionDef'producer</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-62"></a>                       <span class='hs-sel'>_VersionDef'minConsumer</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-63"></a>                       <span class='hs-sel'>_VersionDef'badConsumers</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>}</span>
+<a name="line-64"></a>
+<a name="line-65"></a><a name="instance%20Data.ProtoLens.Message%20VersionDef"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>VersionDef</span> <span class='hs-keyword'>where</span>
+<a name="line-66"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-67"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>producer__field_descriptor</span>
+<a name="line-68"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"producer"</span>
+<a name="line-69"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-70"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-71"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>producer</span><span class='hs-layout'>)</span>
+<a name="line-72"></a>                <span class='hs-varid'>minConsumer__field_descriptor</span>
+<a name="line-73"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"min_consumer"</span>
+<a name="line-74"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-75"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-76"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>minConsumer</span><span class='hs-layout'>)</span>
+<a name="line-77"></a>                <span class='hs-varid'>badConsumers__field_descriptor</span>
+<a name="line-78"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"bad_consumers"</span>
+<a name="line-79"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-80"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-81"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>badConsumers</span><span class='hs-layout'>)</span>
+<a name="line-82"></a>              <span class='hs-keyword'>in</span>
+<a name="line-83"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-84"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-85"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>producer__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-86"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>minConsumer__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-87"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>badConsumers__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-88"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-89"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"producer"</span><span class='hs-layout'>,</span> <span class='hs-varid'>producer__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-90"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"min_consumer"</span><span class='hs-layout'>,</span> <span class='hs-varid'>minConsumer__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-91"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"bad_consumers"</span><span class='hs-layout'>,</span> <span class='hs-varid'>badConsumers__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-92"></a>
+<a name="line-93"></a><a name="badConsumers"></a><span class='hs-definition'>badConsumers</span> <span class='hs-keyglyph'>::</span>
+<a name="line-94"></a>             <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-95"></a>               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"badConsumers"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-96"></a>               <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-97"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"badConsumers"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-98"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"badConsumers"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-99"></a><span class='hs-definition'>badConsumers</span>
+<a name="line-100"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-101"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"badConsumers"</span><span class='hs-layout'>)</span>
+<a name="line-102"></a>
+<a name="line-103"></a><a name="minConsumer"></a><span class='hs-definition'>minConsumer</span> <span class='hs-keyglyph'>::</span>
+<a name="line-104"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"minConsumer"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-105"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"minConsumer"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-106"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"minConsumer"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-107"></a><span class='hs-definition'>minConsumer</span>
+<a name="line-108"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-109"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"minConsumer"</span><span class='hs-layout'>)</span>
+<a name="line-110"></a>
+<a name="line-111"></a><a name="producer"></a><span class='hs-definition'>producer</span> <span class='hs-keyglyph'>::</span>
+<a name="line-112"></a>         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"producer"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-113"></a>           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"producer"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-114"></a>             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"producer"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-115"></a><span class='hs-definition'>producer</span>
+<a name="line-116"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-117"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"producer"</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Protobuf-Config.html b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Protobuf-Config.html
new file mode 100644
index 0000000..e91a36d
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/src/Proto-Tensorflow-Core-Protobuf-Config.html
@@ -0,0 +1,1671 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>.stack-work/dist/x86_64-linux-dkd1ce2ff9c9560b648268df668d177711/Cabal-1.22.5.0/build/autogen/Proto/Tensorflow/Core/Protobuf/Config.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>{- This file was auto-generated from tensorflow/core/protobuf/config.proto by the proto-lens-protoc program. -}</span>
+<a name="line-2"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables, DataKinds, TypeFamilies,
+<a name="line-3"></a>  MultiParamTypeClasses, FlexibleContexts, FlexibleInstances,
+<a name="line-4"></a>  PatternSynonyms #-}</span>
+<a name="line-5"></a><span class='hs-comment'>{-# OPTIONS_GHC -fno-warn-unused-imports#-}</span>
+<a name="line-6"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Protobuf</span><span class='hs-varop'>.</span><span class='hs-conid'>Config</span> <span class='hs-keyword'>where</span>
+<a name="line-7"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Prelude</span>
+<a name="line-8"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span>
+<a name="line-9"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Word</span>
+<a name="line-10"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-11"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span>
+<a name="line-12"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span>
+<a name="line-13"></a>       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-14"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span>
+<a name="line-15"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-16"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span>
+<a name="line-17"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-18"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span>
+<a name="line-19"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-20"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Reexport</span><span class='hs-varop'>.</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-24"></a>       <span class='hs-keyword'>as</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span>
+<a name="line-25"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>CostGraph</span>
+<a name="line-26"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Graph</span>
+<a name="line-27"></a><span class='hs-keyword'>import</span> <span class='hs-keyword'>qualified</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>StepStats</span>
+<a name="line-28"></a>
+<a name="line-29"></a><a name="ConfigProto"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>ConfigProto</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'deviceCount</span> <span class='hs-keyglyph'>::</span>
+<a name="line-30"></a>                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-31"></a>                               <span class='hs-sel'>_ConfigProto'intraOpParallelismThreads</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-32"></a>                               <span class='hs-sel'>_ConfigProto'interOpParallelismThreads</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-33"></a>                               <span class='hs-sel'>_ConfigProto'usePerSessionThreads</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-34"></a>                               <span class='hs-sel'>_ConfigProto'sessionInterOpThreadPool</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>ThreadPoolOptionProto</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-35"></a>                               <span class='hs-sel'>_ConfigProto'placementPeriod</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-36"></a>                               <span class='hs-sel'>_ConfigProto'deviceFilters</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-37"></a>                               <span class='hs-sel'>_ConfigProto'gpuOptions</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>GPUOptions</span><span class='hs-layout'>,</span>
+<a name="line-38"></a>                               <span class='hs-sel'>_ConfigProto'allowSoftPlacement</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-39"></a>                               <span class='hs-sel'>_ConfigProto'logDevicePlacement</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-40"></a>                               <span class='hs-sel'>_ConfigProto'graphOptions</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>GraphOptions</span><span class='hs-layout'>,</span>
+<a name="line-41"></a>                               <span class='hs-sel'>_ConfigProto'operationTimeoutInMs</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>}</span>
+<a name="line-42"></a>                 <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-43"></a>
+<a name="line-44"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"deviceCount"</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-45"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-46"></a>
+<a name="line-47"></a><a name="instance%20Data.ProtoLens.HasField%20%22deviceCount%22%20ConfigProto%20ConfigProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"deviceCount"</span> <span class='hs-conid'>ConfigProto</span>
+<a name="line-48"></a>         <span class='hs-conid'>ConfigProto</span> <span class='hs-keyword'>where</span>
+<a name="line-49"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-50"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ConfigProto'deviceCount</span>
+<a name="line-51"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'deviceCount</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-52"></a>
+<a name="line-53"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-54"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"intraOpParallelismThreads"</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-55"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-56"></a>
+<a name="line-57"></a><a name="instance%20Data.ProtoLens.HasField%20%22intraOpParallelismThreads%22%20ConfigProto%20ConfigProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"intraOpParallelismThreads"</span>
+<a name="line-58"></a>         <span class='hs-conid'>ConfigProto</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyword'>where</span>
+<a name="line-59"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-60"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span>
+<a name="line-61"></a>              <span class='hs-sel'>_ConfigProto'intraOpParallelismThreads</span>
+<a name="line-62"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'intraOpParallelismThreads</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-63"></a>
+<a name="line-64"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-65"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"interOpParallelismThreads"</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-66"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-67"></a>
+<a name="line-68"></a><a name="instance%20Data.ProtoLens.HasField%20%22interOpParallelismThreads%22%20ConfigProto%20ConfigProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"interOpParallelismThreads"</span>
+<a name="line-69"></a>         <span class='hs-conid'>ConfigProto</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyword'>where</span>
+<a name="line-70"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-71"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span>
+<a name="line-72"></a>              <span class='hs-sel'>_ConfigProto'interOpParallelismThreads</span>
+<a name="line-73"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'interOpParallelismThreads</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-74"></a>
+<a name="line-75"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-76"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"usePerSessionThreads"</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-77"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-78"></a>
+<a name="line-79"></a><a name="instance%20Data.ProtoLens.HasField%20%22usePerSessionThreads%22%20ConfigProto%20ConfigProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"usePerSessionThreads"</span> <span class='hs-conid'>ConfigProto</span>
+<a name="line-80"></a>         <span class='hs-conid'>ConfigProto</span> <span class='hs-keyword'>where</span>
+<a name="line-81"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-82"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ConfigProto'usePerSessionThreads</span>
+<a name="line-83"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'usePerSessionThreads</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-84"></a>
+<a name="line-85"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-86"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"sessionInterOpThreadPool"</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-87"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>ThreadPoolOptionProto</span><span class='hs-keyglyph'>]</span>
+<a name="line-88"></a>
+<a name="line-89"></a><a name="instance%20Data.ProtoLens.HasField%20%22sessionInterOpThreadPool%22%20ConfigProto%20ConfigProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"sessionInterOpThreadPool"</span>
+<a name="line-90"></a>         <span class='hs-conid'>ConfigProto</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyword'>where</span>
+<a name="line-91"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-92"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ConfigProto'sessionInterOpThreadPool</span>
+<a name="line-93"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'sessionInterOpThreadPool</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-94"></a>
+<a name="line-95"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"placementPeriod"</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-96"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-97"></a>
+<a name="line-98"></a><a name="instance%20Data.ProtoLens.HasField%20%22placementPeriod%22%20ConfigProto%20ConfigProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"placementPeriod"</span> <span class='hs-conid'>ConfigProto</span>
+<a name="line-99"></a>         <span class='hs-conid'>ConfigProto</span> <span class='hs-keyword'>where</span>
+<a name="line-100"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-101"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ConfigProto'placementPeriod</span>
+<a name="line-102"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'placementPeriod</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-103"></a>
+<a name="line-104"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"deviceFilters"</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-105"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-keyglyph'>]</span>
+<a name="line-106"></a>
+<a name="line-107"></a><a name="instance%20Data.ProtoLens.HasField%20%22deviceFilters%22%20ConfigProto%20ConfigProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"deviceFilters"</span> <span class='hs-conid'>ConfigProto</span>
+<a name="line-108"></a>         <span class='hs-conid'>ConfigProto</span> <span class='hs-keyword'>where</span>
+<a name="line-109"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-110"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ConfigProto'deviceFilters</span>
+<a name="line-111"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'deviceFilters</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-112"></a>
+<a name="line-113"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"gpuOptions"</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-114"></a>     <span class='hs-conid'>GPUOptions</span>
+<a name="line-115"></a>
+<a name="line-116"></a><a name="instance%20Data.ProtoLens.HasField%20%22gpuOptions%22%20ConfigProto%20ConfigProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"gpuOptions"</span> <span class='hs-conid'>ConfigProto</span>
+<a name="line-117"></a>         <span class='hs-conid'>ConfigProto</span> <span class='hs-keyword'>where</span>
+<a name="line-118"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-119"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'gpuOptions</span>
+<a name="line-120"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-121"></a>
+<a name="line-122"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'gpuOptions"</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-123"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>GPUOptions</span>
+<a name="line-124"></a>
+<a name="line-125"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'gpuOptions%22%20ConfigProto%20ConfigProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'gpuOptions"</span> <span class='hs-conid'>ConfigProto</span>
+<a name="line-126"></a>         <span class='hs-conid'>ConfigProto</span> <span class='hs-keyword'>where</span>
+<a name="line-127"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-128"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ConfigProto'gpuOptions</span>
+<a name="line-129"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'gpuOptions</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-130"></a>
+<a name="line-131"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allowSoftPlacement"</span> <span class='hs-conid'>ConfigProto</span>
+<a name="line-132"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-133"></a>
+<a name="line-134"></a><a name="instance%20Data.ProtoLens.HasField%20%22allowSoftPlacement%22%20ConfigProto%20ConfigProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allowSoftPlacement"</span> <span class='hs-conid'>ConfigProto</span>
+<a name="line-135"></a>         <span class='hs-conid'>ConfigProto</span> <span class='hs-keyword'>where</span>
+<a name="line-136"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-137"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ConfigProto'allowSoftPlacement</span>
+<a name="line-138"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'allowSoftPlacement</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-139"></a>
+<a name="line-140"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"logDevicePlacement"</span> <span class='hs-conid'>ConfigProto</span>
+<a name="line-141"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-142"></a>
+<a name="line-143"></a><a name="instance%20Data.ProtoLens.HasField%20%22logDevicePlacement%22%20ConfigProto%20ConfigProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"logDevicePlacement"</span> <span class='hs-conid'>ConfigProto</span>
+<a name="line-144"></a>         <span class='hs-conid'>ConfigProto</span> <span class='hs-keyword'>where</span>
+<a name="line-145"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-146"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ConfigProto'logDevicePlacement</span>
+<a name="line-147"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'logDevicePlacement</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-148"></a>
+<a name="line-149"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"graphOptions"</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-150"></a>     <span class='hs-conid'>GraphOptions</span>
+<a name="line-151"></a>
+<a name="line-152"></a><a name="instance%20Data.ProtoLens.HasField%20%22graphOptions%22%20ConfigProto%20ConfigProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"graphOptions"</span> <span class='hs-conid'>ConfigProto</span>
+<a name="line-153"></a>         <span class='hs-conid'>ConfigProto</span> <span class='hs-keyword'>where</span>
+<a name="line-154"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-155"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'graphOptions</span>
+<a name="line-156"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-157"></a>
+<a name="line-158"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'graphOptions"</span> <span class='hs-conid'>ConfigProto</span>
+<a name="line-159"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>GraphOptions</span>
+<a name="line-160"></a>
+<a name="line-161"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'graphOptions%22%20ConfigProto%20ConfigProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'graphOptions"</span> <span class='hs-conid'>ConfigProto</span>
+<a name="line-162"></a>         <span class='hs-conid'>ConfigProto</span> <span class='hs-keyword'>where</span>
+<a name="line-163"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-164"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ConfigProto'graphOptions</span>
+<a name="line-165"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'graphOptions</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-166"></a>
+<a name="line-167"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-168"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"operationTimeoutInMs"</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-169"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-170"></a>
+<a name="line-171"></a><a name="instance%20Data.ProtoLens.HasField%20%22operationTimeoutInMs%22%20ConfigProto%20ConfigProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"operationTimeoutInMs"</span> <span class='hs-conid'>ConfigProto</span>
+<a name="line-172"></a>         <span class='hs-conid'>ConfigProto</span> <span class='hs-keyword'>where</span>
+<a name="line-173"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-174"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ConfigProto'operationTimeoutInMs</span>
+<a name="line-175"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'operationTimeoutInMs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-176"></a>
+<a name="line-177"></a><a name="instance%20Data.Default.Class.Default%20ConfigProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyword'>where</span>
+<a name="line-178"></a>        <span class='hs-varid'>def</span>
+<a name="line-179"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>ConfigProto</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'deviceCount</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>empty</span><span class='hs-layout'>,</span>
+<a name="line-180"></a>                        <span class='hs-sel'>_ConfigProto'intraOpParallelismThreads</span> <span class='hs-keyglyph'>=</span>
+<a name="line-181"></a>                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-182"></a>                        <span class='hs-sel'>_ConfigProto'interOpParallelismThreads</span> <span class='hs-keyglyph'>=</span>
+<a name="line-183"></a>                          <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-184"></a>                        <span class='hs-sel'>_ConfigProto'usePerSessionThreads</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-185"></a>                        <span class='hs-sel'>_ConfigProto'sessionInterOpThreadPool</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-186"></a>                        <span class='hs-sel'>_ConfigProto'placementPeriod</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-187"></a>                        <span class='hs-sel'>_ConfigProto'deviceFilters</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-188"></a>                        <span class='hs-sel'>_ConfigProto'gpuOptions</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-189"></a>                        <span class='hs-sel'>_ConfigProto'allowSoftPlacement</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-190"></a>                        <span class='hs-sel'>_ConfigProto'logDevicePlacement</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-191"></a>                        <span class='hs-sel'>_ConfigProto'graphOptions</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-192"></a>                        <span class='hs-sel'>_ConfigProto'operationTimeoutInMs</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-193"></a>
+<a name="line-194"></a><a name="instance%20Data.ProtoLens.Message%20ConfigProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>ConfigProto</span> <span class='hs-keyword'>where</span>
+<a name="line-195"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-196"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>deviceCount__field_descriptor</span>
+<a name="line-197"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"device_count"</span>
+<a name="line-198"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-199"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>ConfigProto'DeviceCountEntry</span><span class='hs-layout'>)</span>
+<a name="line-200"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MapField</span> <span class='hs-varid'>key</span> <span class='hs-varid'>value</span> <span class='hs-varid'>deviceCount</span><span class='hs-layout'>)</span>
+<a name="line-201"></a>                <span class='hs-varid'>intraOpParallelismThreads__field_descriptor</span>
+<a name="line-202"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"intra_op_parallelism_threads"</span>
+<a name="line-203"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-204"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-205"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-206"></a>                         <span class='hs-varid'>intraOpParallelismThreads</span><span class='hs-layout'>)</span>
+<a name="line-207"></a>                <span class='hs-varid'>interOpParallelismThreads__field_descriptor</span>
+<a name="line-208"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"inter_op_parallelism_threads"</span>
+<a name="line-209"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-210"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-211"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-212"></a>                         <span class='hs-varid'>interOpParallelismThreads</span><span class='hs-layout'>)</span>
+<a name="line-213"></a>                <span class='hs-varid'>usePerSessionThreads__field_descriptor</span>
+<a name="line-214"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"use_per_session_threads"</span>
+<a name="line-215"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-216"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-217"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-218"></a>                         <span class='hs-varid'>usePerSessionThreads</span><span class='hs-layout'>)</span>
+<a name="line-219"></a>                <span class='hs-varid'>sessionInterOpThreadPool__field_descriptor</span>
+<a name="line-220"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"session_inter_op_thread_pool"</span>
+<a name="line-221"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-222"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>ThreadPoolOptionProto</span><span class='hs-layout'>)</span>
+<a name="line-223"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span>
+<a name="line-224"></a>                         <span class='hs-varid'>sessionInterOpThreadPool</span><span class='hs-layout'>)</span>
+<a name="line-225"></a>                <span class='hs-varid'>placementPeriod__field_descriptor</span>
+<a name="line-226"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"placement_period"</span>
+<a name="line-227"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-228"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-229"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>placementPeriod</span><span class='hs-layout'>)</span>
+<a name="line-230"></a>                <span class='hs-varid'>deviceFilters__field_descriptor</span>
+<a name="line-231"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"device_filters"</span>
+<a name="line-232"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-233"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-234"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span>
+<a name="line-235"></a>                         <span class='hs-varid'>deviceFilters</span><span class='hs-layout'>)</span>
+<a name="line-236"></a>                <span class='hs-varid'>gpuOptions__field_descriptor</span>
+<a name="line-237"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"gpu_options"</span>
+<a name="line-238"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-239"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>GPUOptions</span><span class='hs-layout'>)</span>
+<a name="line-240"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'gpuOptions</span><span class='hs-layout'>)</span>
+<a name="line-241"></a>                <span class='hs-varid'>allowSoftPlacement__field_descriptor</span>
+<a name="line-242"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"allow_soft_placement"</span>
+<a name="line-243"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-244"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-245"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-246"></a>                         <span class='hs-varid'>allowSoftPlacement</span><span class='hs-layout'>)</span>
+<a name="line-247"></a>                <span class='hs-varid'>logDevicePlacement__field_descriptor</span>
+<a name="line-248"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"log_device_placement"</span>
+<a name="line-249"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-250"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-251"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-252"></a>                         <span class='hs-varid'>logDevicePlacement</span><span class='hs-layout'>)</span>
+<a name="line-253"></a>                <span class='hs-varid'>graphOptions__field_descriptor</span>
+<a name="line-254"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"graph_options"</span>
+<a name="line-255"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-256"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>GraphOptions</span><span class='hs-layout'>)</span>
+<a name="line-257"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'graphOptions</span><span class='hs-layout'>)</span>
+<a name="line-258"></a>                <span class='hs-varid'>operationTimeoutInMs__field_descriptor</span>
+<a name="line-259"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"operation_timeout_in_ms"</span>
+<a name="line-260"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-261"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-262"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-263"></a>                         <span class='hs-varid'>operationTimeoutInMs</span><span class='hs-layout'>)</span>
+<a name="line-264"></a>              <span class='hs-keyword'>in</span>
+<a name="line-265"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-266"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-267"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>deviceCount__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-268"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span>
+<a name="line-269"></a>                     <span class='hs-varid'>intraOpParallelismThreads__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-270"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>5</span><span class='hs-layout'>,</span>
+<a name="line-271"></a>                     <span class='hs-varid'>interOpParallelismThreads__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-272"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>9</span><span class='hs-layout'>,</span> <span class='hs-varid'>usePerSessionThreads__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-273"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>12</span><span class='hs-layout'>,</span>
+<a name="line-274"></a>                     <span class='hs-varid'>sessionInterOpThreadPool__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-275"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>placementPeriod__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-276"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>deviceFilters__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-277"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>6</span><span class='hs-layout'>,</span> <span class='hs-varid'>gpuOptions__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-278"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>7</span><span class='hs-layout'>,</span> <span class='hs-varid'>allowSoftPlacement__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-279"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>8</span><span class='hs-layout'>,</span> <span class='hs-varid'>logDevicePlacement__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-280"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>10</span><span class='hs-layout'>,</span> <span class='hs-varid'>graphOptions__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-281"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>11</span><span class='hs-layout'>,</span> <span class='hs-varid'>operationTimeoutInMs__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-282"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-283"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"device_count"</span><span class='hs-layout'>,</span> <span class='hs-varid'>deviceCount__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-284"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"intra_op_parallelism_threads"</span><span class='hs-layout'>,</span>
+<a name="line-285"></a>                     <span class='hs-varid'>intraOpParallelismThreads__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-286"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"inter_op_parallelism_threads"</span><span class='hs-layout'>,</span>
+<a name="line-287"></a>                     <span class='hs-varid'>interOpParallelismThreads__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-288"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"use_per_session_threads"</span><span class='hs-layout'>,</span>
+<a name="line-289"></a>                     <span class='hs-varid'>usePerSessionThreads__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-290"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"session_inter_op_thread_pool"</span><span class='hs-layout'>,</span>
+<a name="line-291"></a>                     <span class='hs-varid'>sessionInterOpThreadPool__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-292"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"placement_period"</span><span class='hs-layout'>,</span> <span class='hs-varid'>placementPeriod__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-293"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"device_filters"</span><span class='hs-layout'>,</span> <span class='hs-varid'>deviceFilters__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-294"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"gpu_options"</span><span class='hs-layout'>,</span> <span class='hs-varid'>gpuOptions__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-295"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"allow_soft_placement"</span><span class='hs-layout'>,</span> <span class='hs-varid'>allowSoftPlacement__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-296"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"log_device_placement"</span><span class='hs-layout'>,</span> <span class='hs-varid'>logDevicePlacement__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-297"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"graph_options"</span><span class='hs-layout'>,</span> <span class='hs-varid'>graphOptions__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-298"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"operation_timeout_in_ms"</span><span class='hs-layout'>,</span>
+<a name="line-299"></a>                     <span class='hs-varid'>operationTimeoutInMs__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-300"></a>
+<a name="line-301"></a><a name="ConfigProto'DeviceCountEntry"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>ConfigProto'DeviceCountEntry</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>ConfigProto'DeviceCountEntry</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'DeviceCountEntry'key</span>
+<a name="line-302"></a>                                                                 <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-303"></a>                                                                 <span class='hs-sel'>_ConfigProto'DeviceCountEntry'value</span>
+<a name="line-304"></a>                                                                 <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>}</span>
+<a name="line-305"></a>                                  <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-306"></a>
+<a name="line-307"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-308"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"key"</span> <span class='hs-conid'>ConfigProto'DeviceCountEntry</span> <span class='hs-keyglyph'>=</span>
+<a name="line-309"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-310"></a>
+<a name="line-311"></a><a name="instance%20Data.ProtoLens.HasField%20%22key%22%20ConfigProto'DeviceCountEntry%20ConfigProto'DeviceCountEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"key"</span> <span class='hs-conid'>ConfigProto'DeviceCountEntry</span>
+<a name="line-312"></a>         <span class='hs-conid'>ConfigProto'DeviceCountEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-313"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-314"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ConfigProto'DeviceCountEntry'key</span>
+<a name="line-315"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'DeviceCountEntry'key</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-316"></a>
+<a name="line-317"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-318"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"value"</span> <span class='hs-conid'>ConfigProto'DeviceCountEntry</span> <span class='hs-keyglyph'>=</span>
+<a name="line-319"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-320"></a>
+<a name="line-321"></a><a name="instance%20Data.ProtoLens.HasField%20%22value%22%20ConfigProto'DeviceCountEntry%20ConfigProto'DeviceCountEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"value"</span>
+<a name="line-322"></a>         <span class='hs-conid'>ConfigProto'DeviceCountEntry</span> <span class='hs-conid'>ConfigProto'DeviceCountEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-323"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-324"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ConfigProto'DeviceCountEntry'value</span>
+<a name="line-325"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'DeviceCountEntry'value</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-326"></a>
+<a name="line-327"></a><a name="instance%20Data.Default.Class.Default%20ConfigProto'DeviceCountEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>ConfigProto'DeviceCountEntry</span>
+<a name="line-328"></a>         <span class='hs-keyword'>where</span>
+<a name="line-329"></a>        <span class='hs-varid'>def</span>
+<a name="line-330"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>ConfigProto'DeviceCountEntry</span><span class='hs-layout'>{</span><span class='hs-sel'>_ConfigProto'DeviceCountEntry'key</span> <span class='hs-keyglyph'>=</span>
+<a name="line-331"></a>                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-332"></a>                                         <span class='hs-sel'>_ConfigProto'DeviceCountEntry'value</span> <span class='hs-keyglyph'>=</span>
+<a name="line-333"></a>                                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-334"></a>
+<a name="line-335"></a><a name="instance%20Data.ProtoLens.Message%20ConfigProto'DeviceCountEntry"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>ConfigProto'DeviceCountEntry</span> <span class='hs-keyword'>where</span>
+<a name="line-336"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-337"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>key__field_descriptor</span>
+<a name="line-338"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"key"</span>
+<a name="line-339"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-340"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-341"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>key</span><span class='hs-layout'>)</span>
+<a name="line-342"></a>                <span class='hs-varid'>value__field_descriptor</span>
+<a name="line-343"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"value"</span>
+<a name="line-344"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-345"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-346"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>value</span><span class='hs-layout'>)</span>
+<a name="line-347"></a>              <span class='hs-keyword'>in</span>
+<a name="line-348"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-349"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-350"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>key__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-351"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>value__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-352"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-353"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"key"</span><span class='hs-layout'>,</span> <span class='hs-varid'>key__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-354"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"value"</span><span class='hs-layout'>,</span> <span class='hs-varid'>value__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-355"></a>
+<a name="line-356"></a><a name="DebugTensorWatch"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>DebugTensorWatch</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DebugTensorWatch</span><span class='hs-layout'>{</span><span class='hs-sel'>_DebugTensorWatch'nodeName</span>
+<a name="line-357"></a>                                         <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-358"></a>                                         <span class='hs-sel'>_DebugTensorWatch'outputSlot</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-359"></a>                                         <span class='hs-sel'>_DebugTensorWatch'debugOps</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-360"></a>                                         <span class='hs-sel'>_DebugTensorWatch'debugUrls</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>}</span>
+<a name="line-361"></a>                      <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-362"></a>
+<a name="line-363"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"nodeName"</span> <span class='hs-conid'>DebugTensorWatch</span> <span class='hs-keyglyph'>=</span>
+<a name="line-364"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-365"></a>
+<a name="line-366"></a><a name="instance%20Data.ProtoLens.HasField%20%22nodeName%22%20DebugTensorWatch%20DebugTensorWatch"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"nodeName"</span> <span class='hs-conid'>DebugTensorWatch</span>
+<a name="line-367"></a>         <span class='hs-conid'>DebugTensorWatch</span> <span class='hs-keyword'>where</span>
+<a name="line-368"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-369"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_DebugTensorWatch'nodeName</span>
+<a name="line-370"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_DebugTensorWatch'nodeName</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-371"></a>
+<a name="line-372"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"outputSlot"</span> <span class='hs-conid'>DebugTensorWatch</span> <span class='hs-keyglyph'>=</span>
+<a name="line-373"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-374"></a>
+<a name="line-375"></a><a name="instance%20Data.ProtoLens.HasField%20%22outputSlot%22%20DebugTensorWatch%20DebugTensorWatch"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"outputSlot"</span> <span class='hs-conid'>DebugTensorWatch</span>
+<a name="line-376"></a>         <span class='hs-conid'>DebugTensorWatch</span> <span class='hs-keyword'>where</span>
+<a name="line-377"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-378"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_DebugTensorWatch'outputSlot</span>
+<a name="line-379"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_DebugTensorWatch'outputSlot</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-380"></a>
+<a name="line-381"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"debugOps"</span> <span class='hs-conid'>DebugTensorWatch</span> <span class='hs-keyglyph'>=</span>
+<a name="line-382"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-keyglyph'>]</span>
+<a name="line-383"></a>
+<a name="line-384"></a><a name="instance%20Data.ProtoLens.HasField%20%22debugOps%22%20DebugTensorWatch%20DebugTensorWatch"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"debugOps"</span> <span class='hs-conid'>DebugTensorWatch</span>
+<a name="line-385"></a>         <span class='hs-conid'>DebugTensorWatch</span> <span class='hs-keyword'>where</span>
+<a name="line-386"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-387"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_DebugTensorWatch'debugOps</span>
+<a name="line-388"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_DebugTensorWatch'debugOps</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-389"></a>
+<a name="line-390"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"debugUrls"</span> <span class='hs-conid'>DebugTensorWatch</span> <span class='hs-keyglyph'>=</span>
+<a name="line-391"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-keyglyph'>]</span>
+<a name="line-392"></a>
+<a name="line-393"></a><a name="instance%20Data.ProtoLens.HasField%20%22debugUrls%22%20DebugTensorWatch%20DebugTensorWatch"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"debugUrls"</span> <span class='hs-conid'>DebugTensorWatch</span>
+<a name="line-394"></a>         <span class='hs-conid'>DebugTensorWatch</span> <span class='hs-keyword'>where</span>
+<a name="line-395"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-396"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_DebugTensorWatch'debugUrls</span>
+<a name="line-397"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_DebugTensorWatch'debugUrls</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-398"></a>
+<a name="line-399"></a><a name="instance%20Data.Default.Class.Default%20DebugTensorWatch"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>DebugTensorWatch</span> <span class='hs-keyword'>where</span>
+<a name="line-400"></a>        <span class='hs-varid'>def</span>
+<a name="line-401"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>DebugTensorWatch</span><span class='hs-layout'>{</span><span class='hs-sel'>_DebugTensorWatch'nodeName</span> <span class='hs-keyglyph'>=</span>
+<a name="line-402"></a>                               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-403"></a>                             <span class='hs-sel'>_DebugTensorWatch'outputSlot</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-404"></a>                             <span class='hs-sel'>_DebugTensorWatch'debugOps</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span> <span class='hs-sel'>_DebugTensorWatch'debugUrls</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>}</span>
+<a name="line-405"></a>
+<a name="line-406"></a><a name="instance%20Data.ProtoLens.Message%20DebugTensorWatch"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>DebugTensorWatch</span> <span class='hs-keyword'>where</span>
+<a name="line-407"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-408"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>nodeName__field_descriptor</span>
+<a name="line-409"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"node_name"</span>
+<a name="line-410"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-411"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-412"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>nodeName</span><span class='hs-layout'>)</span>
+<a name="line-413"></a>                <span class='hs-varid'>outputSlot__field_descriptor</span>
+<a name="line-414"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"output_slot"</span>
+<a name="line-415"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-416"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-417"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>outputSlot</span><span class='hs-layout'>)</span>
+<a name="line-418"></a>                <span class='hs-varid'>debugOps__field_descriptor</span>
+<a name="line-419"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"debug_ops"</span>
+<a name="line-420"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-421"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-422"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>debugOps</span><span class='hs-layout'>)</span>
+<a name="line-423"></a>                <span class='hs-varid'>debugUrls__field_descriptor</span>
+<a name="line-424"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"debug_urls"</span>
+<a name="line-425"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-426"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-427"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span> <span class='hs-varid'>debugUrls</span><span class='hs-layout'>)</span>
+<a name="line-428"></a>              <span class='hs-keyword'>in</span>
+<a name="line-429"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-430"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-431"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>nodeName__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-432"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>outputSlot__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-433"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>debugOps__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-434"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>debugUrls__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-435"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-436"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"node_name"</span><span class='hs-layout'>,</span> <span class='hs-varid'>nodeName__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-437"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"output_slot"</span><span class='hs-layout'>,</span> <span class='hs-varid'>outputSlot__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-438"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"debug_ops"</span><span class='hs-layout'>,</span> <span class='hs-varid'>debugOps__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-439"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"debug_urls"</span><span class='hs-layout'>,</span> <span class='hs-varid'>debugUrls__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-440"></a>
+<a name="line-441"></a><a name="GPUOptions"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>GPUOptions</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>GPUOptions</span><span class='hs-layout'>{</span><span class='hs-sel'>_GPUOptions'perProcessGpuMemoryFraction</span>
+<a name="line-442"></a>                             <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Double</span><span class='hs-layout'>,</span>
+<a name="line-443"></a>                             <span class='hs-sel'>_GPUOptions'allocatorType</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>,</span>
+<a name="line-444"></a>                             <span class='hs-sel'>_GPUOptions'deferredDeletionBytes</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-445"></a>                             <span class='hs-sel'>_GPUOptions'allowGrowth</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-446"></a>                             <span class='hs-sel'>_GPUOptions'visibleDeviceList</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>}</span>
+<a name="line-447"></a>                <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-448"></a>
+<a name="line-449"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-450"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"perProcessGpuMemoryFraction"</span> <span class='hs-conid'>GPUOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-451"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Double</span>
+<a name="line-452"></a>
+<a name="line-453"></a><a name="instance%20Data.ProtoLens.HasField%20%22perProcessGpuMemoryFraction%22%20GPUOptions%20GPUOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"perProcessGpuMemoryFraction"</span>
+<a name="line-454"></a>         <span class='hs-conid'>GPUOptions</span> <span class='hs-conid'>GPUOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-455"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-456"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span>
+<a name="line-457"></a>              <span class='hs-sel'>_GPUOptions'perProcessGpuMemoryFraction</span>
+<a name="line-458"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GPUOptions'perProcessGpuMemoryFraction</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-459"></a>
+<a name="line-460"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocatorType"</span> <span class='hs-conid'>GPUOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-461"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-462"></a>
+<a name="line-463"></a><a name="instance%20Data.ProtoLens.HasField%20%22allocatorType%22%20GPUOptions%20GPUOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allocatorType"</span> <span class='hs-conid'>GPUOptions</span>
+<a name="line-464"></a>         <span class='hs-conid'>GPUOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-465"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-466"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GPUOptions'allocatorType</span>
+<a name="line-467"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GPUOptions'allocatorType</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-468"></a>
+<a name="line-469"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-470"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"deferredDeletionBytes"</span> <span class='hs-conid'>GPUOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-471"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-472"></a>
+<a name="line-473"></a><a name="instance%20Data.ProtoLens.HasField%20%22deferredDeletionBytes%22%20GPUOptions%20GPUOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"deferredDeletionBytes"</span> <span class='hs-conid'>GPUOptions</span>
+<a name="line-474"></a>         <span class='hs-conid'>GPUOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-475"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-476"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GPUOptions'deferredDeletionBytes</span>
+<a name="line-477"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GPUOptions'deferredDeletionBytes</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-478"></a>
+<a name="line-479"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allowGrowth"</span> <span class='hs-conid'>GPUOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-480"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-481"></a>
+<a name="line-482"></a><a name="instance%20Data.ProtoLens.HasField%20%22allowGrowth%22%20GPUOptions%20GPUOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allowGrowth"</span> <span class='hs-conid'>GPUOptions</span>
+<a name="line-483"></a>         <span class='hs-conid'>GPUOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-484"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-485"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GPUOptions'allowGrowth</span>
+<a name="line-486"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GPUOptions'allowGrowth</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-487"></a>
+<a name="line-488"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"visibleDeviceList"</span> <span class='hs-conid'>GPUOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-489"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span>
+<a name="line-490"></a>
+<a name="line-491"></a><a name="instance%20Data.ProtoLens.HasField%20%22visibleDeviceList%22%20GPUOptions%20GPUOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"visibleDeviceList"</span> <span class='hs-conid'>GPUOptions</span>
+<a name="line-492"></a>         <span class='hs-conid'>GPUOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-493"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-494"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GPUOptions'visibleDeviceList</span>
+<a name="line-495"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GPUOptions'visibleDeviceList</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-496"></a>
+<a name="line-497"></a><a name="instance%20Data.Default.Class.Default%20GPUOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>GPUOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-498"></a>        <span class='hs-varid'>def</span>
+<a name="line-499"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>GPUOptions</span><span class='hs-layout'>{</span><span class='hs-sel'>_GPUOptions'perProcessGpuMemoryFraction</span> <span class='hs-keyglyph'>=</span>
+<a name="line-500"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-501"></a>                       <span class='hs-sel'>_GPUOptions'allocatorType</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-502"></a>                       <span class='hs-sel'>_GPUOptions'deferredDeletionBytes</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-503"></a>                       <span class='hs-sel'>_GPUOptions'allowGrowth</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-504"></a>                       <span class='hs-sel'>_GPUOptions'visibleDeviceList</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-505"></a>
+<a name="line-506"></a><a name="instance%20Data.ProtoLens.Message%20GPUOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>GPUOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-507"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-508"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>perProcessGpuMemoryFraction__field_descriptor</span>
+<a name="line-509"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"per_process_gpu_memory_fraction"</span>
+<a name="line-510"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>DoubleField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-511"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Double</span><span class='hs-layout'>)</span>
+<a name="line-512"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-513"></a>                         <span class='hs-varid'>perProcessGpuMemoryFraction</span><span class='hs-layout'>)</span>
+<a name="line-514"></a>                <span class='hs-varid'>allocatorType__field_descriptor</span>
+<a name="line-515"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"allocator_type"</span>
+<a name="line-516"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-517"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-518"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>allocatorType</span><span class='hs-layout'>)</span>
+<a name="line-519"></a>                <span class='hs-varid'>deferredDeletionBytes__field_descriptor</span>
+<a name="line-520"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"deferred_deletion_bytes"</span>
+<a name="line-521"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-522"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-523"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-524"></a>                         <span class='hs-varid'>deferredDeletionBytes</span><span class='hs-layout'>)</span>
+<a name="line-525"></a>                <span class='hs-varid'>allowGrowth__field_descriptor</span>
+<a name="line-526"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"allow_growth"</span>
+<a name="line-527"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-528"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-529"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>allowGrowth</span><span class='hs-layout'>)</span>
+<a name="line-530"></a>                <span class='hs-varid'>visibleDeviceList__field_descriptor</span>
+<a name="line-531"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"visible_device_list"</span>
+<a name="line-532"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>StringField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-533"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-varop'>.</span><span class='hs-conid'>Text</span><span class='hs-layout'>)</span>
+<a name="line-534"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-535"></a>                         <span class='hs-varid'>visibleDeviceList</span><span class='hs-layout'>)</span>
+<a name="line-536"></a>              <span class='hs-keyword'>in</span>
+<a name="line-537"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-538"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-539"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span>
+<a name="line-540"></a>                     <span class='hs-varid'>perProcessGpuMemoryFraction__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-541"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>allocatorType__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-542"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>deferredDeletionBytes__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-543"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>allowGrowth__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-544"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>5</span><span class='hs-layout'>,</span> <span class='hs-varid'>visibleDeviceList__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-545"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-546"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"per_process_gpu_memory_fraction"</span><span class='hs-layout'>,</span>
+<a name="line-547"></a>                     <span class='hs-varid'>perProcessGpuMemoryFraction__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-548"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"allocator_type"</span><span class='hs-layout'>,</span> <span class='hs-varid'>allocatorType__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-549"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"deferred_deletion_bytes"</span><span class='hs-layout'>,</span>
+<a name="line-550"></a>                     <span class='hs-varid'>deferredDeletionBytes__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-551"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"allow_growth"</span><span class='hs-layout'>,</span> <span class='hs-varid'>allowGrowth__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-552"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"visible_device_list"</span><span class='hs-layout'>,</span> <span class='hs-varid'>visibleDeviceList__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-553"></a>
+<a name="line-554"></a><a name="GraphOptions"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>GraphOptions</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>GraphOptions</span><span class='hs-layout'>{</span><span class='hs-sel'>_GraphOptions'enableRecvScheduling</span>
+<a name="line-555"></a>                                 <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-556"></a>                                 <span class='hs-sel'>_GraphOptions'optimizerOptions</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>OptimizerOptions</span><span class='hs-layout'>,</span>
+<a name="line-557"></a>                                 <span class='hs-sel'>_GraphOptions'buildCostModel</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-558"></a>                                 <span class='hs-sel'>_GraphOptions'buildCostModelAfter</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-559"></a>                                 <span class='hs-sel'>_GraphOptions'inferShapes</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-560"></a>                                 <span class='hs-sel'>_GraphOptions'placePrunedGraph</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-561"></a>                                 <span class='hs-sel'>_GraphOptions'enableBfloat16Sendrecv</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-562"></a>                                 <span class='hs-sel'>_GraphOptions'timelineStep</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>}</span>
+<a name="line-563"></a>                  <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-564"></a>
+<a name="line-565"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-566"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"enableRecvScheduling"</span> <span class='hs-conid'>GraphOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-567"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-568"></a>
+<a name="line-569"></a><a name="instance%20Data.ProtoLens.HasField%20%22enableRecvScheduling%22%20GraphOptions%20GraphOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"enableRecvScheduling"</span>
+<a name="line-570"></a>         <span class='hs-conid'>GraphOptions</span> <span class='hs-conid'>GraphOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-571"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-572"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GraphOptions'enableRecvScheduling</span>
+<a name="line-573"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GraphOptions'enableRecvScheduling</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-574"></a>
+<a name="line-575"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"optimizerOptions"</span> <span class='hs-conid'>GraphOptions</span>
+<a name="line-576"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OptimizerOptions</span>
+<a name="line-577"></a>
+<a name="line-578"></a><a name="instance%20Data.ProtoLens.HasField%20%22optimizerOptions%22%20GraphOptions%20GraphOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"optimizerOptions"</span> <span class='hs-conid'>GraphOptions</span>
+<a name="line-579"></a>         <span class='hs-conid'>GraphOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-580"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-581"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'optimizerOptions</span>
+<a name="line-582"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-583"></a>
+<a name="line-584"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-585"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'optimizerOptions"</span> <span class='hs-conid'>GraphOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-586"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>OptimizerOptions</span>
+<a name="line-587"></a>
+<a name="line-588"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'optimizerOptions%22%20GraphOptions%20GraphOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'optimizerOptions"</span>
+<a name="line-589"></a>         <span class='hs-conid'>GraphOptions</span> <span class='hs-conid'>GraphOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-590"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-591"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GraphOptions'optimizerOptions</span>
+<a name="line-592"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GraphOptions'optimizerOptions</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-593"></a>
+<a name="line-594"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"buildCostModel"</span> <span class='hs-conid'>GraphOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-595"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-596"></a>
+<a name="line-597"></a><a name="instance%20Data.ProtoLens.HasField%20%22buildCostModel%22%20GraphOptions%20GraphOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"buildCostModel"</span> <span class='hs-conid'>GraphOptions</span>
+<a name="line-598"></a>         <span class='hs-conid'>GraphOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-599"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-600"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GraphOptions'buildCostModel</span>
+<a name="line-601"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GraphOptions'buildCostModel</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-602"></a>
+<a name="line-603"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-604"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"buildCostModelAfter"</span> <span class='hs-conid'>GraphOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-605"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-606"></a>
+<a name="line-607"></a><a name="instance%20Data.ProtoLens.HasField%20%22buildCostModelAfter%22%20GraphOptions%20GraphOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"buildCostModelAfter"</span> <span class='hs-conid'>GraphOptions</span>
+<a name="line-608"></a>         <span class='hs-conid'>GraphOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-609"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-610"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GraphOptions'buildCostModelAfter</span>
+<a name="line-611"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GraphOptions'buildCostModelAfter</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-612"></a>
+<a name="line-613"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"inferShapes"</span> <span class='hs-conid'>GraphOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-614"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-615"></a>
+<a name="line-616"></a><a name="instance%20Data.ProtoLens.HasField%20%22inferShapes%22%20GraphOptions%20GraphOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"inferShapes"</span> <span class='hs-conid'>GraphOptions</span>
+<a name="line-617"></a>         <span class='hs-conid'>GraphOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-618"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-619"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GraphOptions'inferShapes</span>
+<a name="line-620"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GraphOptions'inferShapes</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-621"></a>
+<a name="line-622"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"placePrunedGraph"</span> <span class='hs-conid'>GraphOptions</span>
+<a name="line-623"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-624"></a>
+<a name="line-625"></a><a name="instance%20Data.ProtoLens.HasField%20%22placePrunedGraph%22%20GraphOptions%20GraphOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"placePrunedGraph"</span> <span class='hs-conid'>GraphOptions</span>
+<a name="line-626"></a>         <span class='hs-conid'>GraphOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-627"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-628"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GraphOptions'placePrunedGraph</span>
+<a name="line-629"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GraphOptions'placePrunedGraph</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-630"></a>
+<a name="line-631"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-632"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"enableBfloat16Sendrecv"</span> <span class='hs-conid'>GraphOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-633"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-634"></a>
+<a name="line-635"></a><a name="instance%20Data.ProtoLens.HasField%20%22enableBfloat16Sendrecv%22%20GraphOptions%20GraphOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"enableBfloat16Sendrecv"</span>
+<a name="line-636"></a>         <span class='hs-conid'>GraphOptions</span> <span class='hs-conid'>GraphOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-637"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-638"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GraphOptions'enableBfloat16Sendrecv</span>
+<a name="line-639"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GraphOptions'enableBfloat16Sendrecv</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-640"></a>
+<a name="line-641"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"timelineStep"</span> <span class='hs-conid'>GraphOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-642"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-643"></a>
+<a name="line-644"></a><a name="instance%20Data.ProtoLens.HasField%20%22timelineStep%22%20GraphOptions%20GraphOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"timelineStep"</span> <span class='hs-conid'>GraphOptions</span>
+<a name="line-645"></a>         <span class='hs-conid'>GraphOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-646"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-647"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_GraphOptions'timelineStep</span>
+<a name="line-648"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_GraphOptions'timelineStep</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-649"></a>
+<a name="line-650"></a><a name="instance%20Data.Default.Class.Default%20GraphOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>GraphOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-651"></a>        <span class='hs-varid'>def</span>
+<a name="line-652"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>GraphOptions</span><span class='hs-layout'>{</span><span class='hs-sel'>_GraphOptions'enableRecvScheduling</span> <span class='hs-keyglyph'>=</span>
+<a name="line-653"></a>                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-654"></a>                         <span class='hs-sel'>_GraphOptions'optimizerOptions</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-655"></a>                         <span class='hs-sel'>_GraphOptions'buildCostModel</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-656"></a>                         <span class='hs-sel'>_GraphOptions'buildCostModelAfter</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-657"></a>                         <span class='hs-sel'>_GraphOptions'inferShapes</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-658"></a>                         <span class='hs-sel'>_GraphOptions'placePrunedGraph</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-659"></a>                         <span class='hs-sel'>_GraphOptions'enableBfloat16Sendrecv</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-660"></a>                         <span class='hs-sel'>_GraphOptions'timelineStep</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-661"></a>
+<a name="line-662"></a><a name="instance%20Data.ProtoLens.Message%20GraphOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>GraphOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-663"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-664"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>enableRecvScheduling__field_descriptor</span>
+<a name="line-665"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"enable_recv_scheduling"</span>
+<a name="line-666"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-667"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-668"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-669"></a>                         <span class='hs-varid'>enableRecvScheduling</span><span class='hs-layout'>)</span>
+<a name="line-670"></a>                <span class='hs-varid'>optimizerOptions__field_descriptor</span>
+<a name="line-671"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"optimizer_options"</span>
+<a name="line-672"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-673"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>OptimizerOptions</span><span class='hs-layout'>)</span>
+<a name="line-674"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'optimizerOptions</span><span class='hs-layout'>)</span>
+<a name="line-675"></a>                <span class='hs-varid'>buildCostModel__field_descriptor</span>
+<a name="line-676"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"build_cost_model"</span>
+<a name="line-677"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-678"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-679"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>buildCostModel</span><span class='hs-layout'>)</span>
+<a name="line-680"></a>                <span class='hs-varid'>buildCostModelAfter__field_descriptor</span>
+<a name="line-681"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"build_cost_model_after"</span>
+<a name="line-682"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-683"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-684"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-685"></a>                         <span class='hs-varid'>buildCostModelAfter</span><span class='hs-layout'>)</span>
+<a name="line-686"></a>                <span class='hs-varid'>inferShapes__field_descriptor</span>
+<a name="line-687"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"infer_shapes"</span>
+<a name="line-688"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-689"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-690"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>inferShapes</span><span class='hs-layout'>)</span>
+<a name="line-691"></a>                <span class='hs-varid'>placePrunedGraph__field_descriptor</span>
+<a name="line-692"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"place_pruned_graph"</span>
+<a name="line-693"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-694"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-695"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-696"></a>                         <span class='hs-varid'>placePrunedGraph</span><span class='hs-layout'>)</span>
+<a name="line-697"></a>                <span class='hs-varid'>enableBfloat16Sendrecv__field_descriptor</span>
+<a name="line-698"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"enable_bfloat16_sendrecv"</span>
+<a name="line-699"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-700"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-701"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-702"></a>                         <span class='hs-varid'>enableBfloat16Sendrecv</span><span class='hs-layout'>)</span>
+<a name="line-703"></a>                <span class='hs-varid'>timelineStep__field_descriptor</span>
+<a name="line-704"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"timeline_step"</span>
+<a name="line-705"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-706"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-707"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>timelineStep</span><span class='hs-layout'>)</span>
+<a name="line-708"></a>              <span class='hs-keyword'>in</span>
+<a name="line-709"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-710"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-711"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>enableRecvScheduling__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-712"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>optimizerOptions__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-713"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>buildCostModel__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-714"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>9</span><span class='hs-layout'>,</span> <span class='hs-varid'>buildCostModelAfter__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-715"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>5</span><span class='hs-layout'>,</span> <span class='hs-varid'>inferShapes__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-716"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>6</span><span class='hs-layout'>,</span> <span class='hs-varid'>placePrunedGraph__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-717"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>7</span><span class='hs-layout'>,</span> <span class='hs-varid'>enableBfloat16Sendrecv__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-718"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>8</span><span class='hs-layout'>,</span> <span class='hs-varid'>timelineStep__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-719"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-720"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"enable_recv_scheduling"</span><span class='hs-layout'>,</span>
+<a name="line-721"></a>                     <span class='hs-varid'>enableRecvScheduling__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-722"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"optimizer_options"</span><span class='hs-layout'>,</span> <span class='hs-varid'>optimizerOptions__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-723"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"build_cost_model"</span><span class='hs-layout'>,</span> <span class='hs-varid'>buildCostModel__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-724"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"build_cost_model_after"</span><span class='hs-layout'>,</span> <span class='hs-varid'>buildCostModelAfter__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-725"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"infer_shapes"</span><span class='hs-layout'>,</span> <span class='hs-varid'>inferShapes__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-726"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"place_pruned_graph"</span><span class='hs-layout'>,</span> <span class='hs-varid'>placePrunedGraph__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-727"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"enable_bfloat16_sendrecv"</span><span class='hs-layout'>,</span>
+<a name="line-728"></a>                     <span class='hs-varid'>enableBfloat16Sendrecv__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-729"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"timeline_step"</span><span class='hs-layout'>,</span> <span class='hs-varid'>timelineStep__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-730"></a>
+<a name="line-731"></a><a name="OptimizerOptions"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>OptimizerOptions</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OptimizerOptions</span><span class='hs-layout'>{</span><span class='hs-sel'>_OptimizerOptions'doCommonSubexpressionElimination</span>
+<a name="line-732"></a>                                         <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-733"></a>                                         <span class='hs-sel'>_OptimizerOptions'doConstantFolding</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-734"></a>                                         <span class='hs-sel'>_OptimizerOptions'doFunctionInlining</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>,</span>
+<a name="line-735"></a>                                         <span class='hs-sel'>_OptimizerOptions'optLevel</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>OptimizerOptions'Level</span><span class='hs-layout'>}</span>
+<a name="line-736"></a>                      <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-737"></a>
+<a name="line-738"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-739"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"doCommonSubexpressionElimination"</span>
+<a name="line-740"></a>       <span class='hs-conid'>OptimizerOptions</span>
+<a name="line-741"></a>     <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-742"></a>
+<a name="line-743"></a><a name="instance%20Data.ProtoLens.HasField%20%22doCommonSubexpressionElimination%22%20OptimizerOptions%20OptimizerOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"doCommonSubexpressionElimination"</span>
+<a name="line-744"></a>         <span class='hs-conid'>OptimizerOptions</span> <span class='hs-conid'>OptimizerOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-745"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-746"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span>
+<a name="line-747"></a>              <span class='hs-sel'>_OptimizerOptions'doCommonSubexpressionElimination</span>
+<a name="line-748"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span>
+<a name="line-749"></a>                 <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OptimizerOptions'doCommonSubexpressionElimination</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-750"></a>
+<a name="line-751"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-752"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"doConstantFolding"</span> <span class='hs-conid'>OptimizerOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-753"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-754"></a>
+<a name="line-755"></a><a name="instance%20Data.ProtoLens.HasField%20%22doConstantFolding%22%20OptimizerOptions%20OptimizerOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"doConstantFolding"</span>
+<a name="line-756"></a>         <span class='hs-conid'>OptimizerOptions</span> <span class='hs-conid'>OptimizerOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-757"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-758"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OptimizerOptions'doConstantFolding</span>
+<a name="line-759"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OptimizerOptions'doConstantFolding</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-760"></a>
+<a name="line-761"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-762"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"doFunctionInlining"</span> <span class='hs-conid'>OptimizerOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-763"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-764"></a>
+<a name="line-765"></a><a name="instance%20Data.ProtoLens.HasField%20%22doFunctionInlining%22%20OptimizerOptions%20OptimizerOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"doFunctionInlining"</span>
+<a name="line-766"></a>         <span class='hs-conid'>OptimizerOptions</span> <span class='hs-conid'>OptimizerOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-767"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-768"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OptimizerOptions'doFunctionInlining</span>
+<a name="line-769"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OptimizerOptions'doFunctionInlining</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-770"></a>
+<a name="line-771"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"optLevel"</span> <span class='hs-conid'>OptimizerOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-772"></a>     <span class='hs-conid'>OptimizerOptions'Level</span>
+<a name="line-773"></a>
+<a name="line-774"></a><a name="instance%20Data.ProtoLens.HasField%20%22optLevel%22%20OptimizerOptions%20OptimizerOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"optLevel"</span> <span class='hs-conid'>OptimizerOptions</span>
+<a name="line-775"></a>         <span class='hs-conid'>OptimizerOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-776"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-777"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_OptimizerOptions'optLevel</span>
+<a name="line-778"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_OptimizerOptions'optLevel</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-779"></a>
+<a name="line-780"></a><a name="instance%20Data.Default.Class.Default%20OptimizerOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>OptimizerOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-781"></a>        <span class='hs-varid'>def</span>
+<a name="line-782"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OptimizerOptions</span><span class='hs-layout'>{</span><span class='hs-sel'>_OptimizerOptions'doCommonSubexpressionElimination</span>
+<a name="line-783"></a>                               <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-784"></a>                             <span class='hs-sel'>_OptimizerOptions'doConstantFolding</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-785"></a>                             <span class='hs-sel'>_OptimizerOptions'doFunctionInlining</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-786"></a>                             <span class='hs-sel'>_OptimizerOptions'optLevel</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>}</span>
+<a name="line-787"></a>
+<a name="line-788"></a><a name="instance%20Data.ProtoLens.Message%20OptimizerOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>OptimizerOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-789"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-790"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>doCommonSubexpressionElimination__field_descriptor</span>
+<a name="line-791"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span>
+<a name="line-792"></a>                      <span class='hs-str'>"do_common_subexpression_elimination"</span>
+<a name="line-793"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-794"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-795"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-796"></a>                         <span class='hs-varid'>doCommonSubexpressionElimination</span><span class='hs-layout'>)</span>
+<a name="line-797"></a>                <span class='hs-varid'>doConstantFolding__field_descriptor</span>
+<a name="line-798"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"do_constant_folding"</span>
+<a name="line-799"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-800"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-801"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-802"></a>                         <span class='hs-varid'>doConstantFolding</span><span class='hs-layout'>)</span>
+<a name="line-803"></a>                <span class='hs-varid'>doFunctionInlining__field_descriptor</span>
+<a name="line-804"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"do_function_inlining"</span>
+<a name="line-805"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-806"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-807"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-808"></a>                         <span class='hs-varid'>doFunctionInlining</span><span class='hs-layout'>)</span>
+<a name="line-809"></a>                <span class='hs-varid'>optLevel__field_descriptor</span>
+<a name="line-810"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"opt_level"</span>
+<a name="line-811"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>EnumField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-812"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>OptimizerOptions'Level</span><span class='hs-layout'>)</span>
+<a name="line-813"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>optLevel</span><span class='hs-layout'>)</span>
+<a name="line-814"></a>              <span class='hs-keyword'>in</span>
+<a name="line-815"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-816"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-817"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span>
+<a name="line-818"></a>                     <span class='hs-varid'>doCommonSubexpressionElimination__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-819"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>doConstantFolding__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-820"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>doFunctionInlining__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-821"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>optLevel__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-822"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-823"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"do_common_subexpression_elimination"</span><span class='hs-layout'>,</span>
+<a name="line-824"></a>                     <span class='hs-varid'>doCommonSubexpressionElimination__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-825"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"do_constant_folding"</span><span class='hs-layout'>,</span> <span class='hs-varid'>doConstantFolding__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-826"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"do_function_inlining"</span><span class='hs-layout'>,</span> <span class='hs-varid'>doFunctionInlining__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-827"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"opt_level"</span><span class='hs-layout'>,</span> <span class='hs-varid'>optLevel__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-828"></a>
+<a name="line-829"></a><a name="OptimizerOptions'Level"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>OptimizerOptions'Level</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OptimizerOptions'L0</span>
+<a name="line-830"></a>                            <span class='hs-keyglyph'>|</span> <span class='hs-conid'>OptimizerOptions'L1</span>
+<a name="line-831"></a>                            <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-832"></a>
+<a name="line-833"></a><a name="instance%20Data.Default.Class.Default%20OptimizerOptions'Level"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>OptimizerOptions'Level</span> <span class='hs-keyword'>where</span>
+<a name="line-834"></a>        <span class='hs-varid'>def</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OptimizerOptions'L0</span>
+<a name="line-835"></a>
+<a name="line-836"></a><a name="instance%20Data.ProtoLens.FieldDefault%20OptimizerOptions'Level"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDefault</span> <span class='hs-conid'>OptimizerOptions'Level</span> <span class='hs-keyword'>where</span>
+<a name="line-837"></a>        <span class='hs-varid'>fieldDefault</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OptimizerOptions'L0</span>
+<a name="line-838"></a>
+<a name="line-839"></a><a name="instance%20Data.ProtoLens.MessageEnum%20OptimizerOptions'Level"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageEnum</span> <span class='hs-conid'>OptimizerOptions'Level</span> <span class='hs-keyword'>where</span>
+<a name="line-840"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-layout'>(</span><span class='hs-comment'>-</span><span class='hs-num'>1</span><span class='hs-layout'>)</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>OptimizerOptions'L0</span>
+<a name="line-841"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>0</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>OptimizerOptions'L1</span>
+<a name="line-842"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span>
+<a name="line-843"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>OptimizerOptions'L0</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"L0"</span>
+<a name="line-844"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>OptimizerOptions'L1</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"L1"</span>
+<a name="line-845"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"L0"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>OptimizerOptions'L0</span>
+<a name="line-846"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"L1"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>OptimizerOptions'L1</span>
+<a name="line-847"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span>
+<a name="line-848"></a>
+<a name="line-849"></a><a name="instance%20Prelude.Enum%20OptimizerOptions'Level"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span> <span class='hs-conid'>OptimizerOptions'Level</span> <span class='hs-keyword'>where</span>
+<a name="line-850"></a>        <span class='hs-varid'>toEnum</span> <span class='hs-varid'>k__</span>
+<a name="line-851"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>maybe</span>
+<a name="line-852"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>error</span>
+<a name="line-853"></a>                 <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.++</span><span class='hs-layout'>)</span> <span class='hs-str'>"toEnum: unknown value for enum Level: "</span>
+<a name="line-854"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>show</span> <span class='hs-varid'>k__</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-855"></a>              <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>id</span>
+<a name="line-856"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeToEnum</span> <span class='hs-varid'>k__</span><span class='hs-layout'>)</span>
+<a name="line-857"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>OptimizerOptions'L0</span> <span class='hs-keyglyph'>=</span> <span class='hs-comment'>-</span><span class='hs-num'>1</span>
+<a name="line-858"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>OptimizerOptions'L1</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>0</span>
+<a name="line-859"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>OptimizerOptions'L1</span>
+<a name="line-860"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>error</span>
+<a name="line-861"></a>              <span class='hs-str'>"Ident \"OptimizerOptions'Level\".Ident \"succ\": bad argument Ident \"OptimizerOptions'L1\". This value would be out of bounds."</span>
+<a name="line-862"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>OptimizerOptions'L0</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OptimizerOptions'L1</span>
+<a name="line-863"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>OptimizerOptions'L0</span>
+<a name="line-864"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>error</span>
+<a name="line-865"></a>              <span class='hs-str'>"Ident \"OptimizerOptions'Level\".Ident \"pred\": bad argument Ident \"OptimizerOptions'L0\". This value would be out of bounds."</span>
+<a name="line-866"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>OptimizerOptions'L1</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OptimizerOptions'L0</span>
+<a name="line-867"></a>        <span class='hs-varid'>enumFrom</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span><span class='hs-varop'>.</span><span class='hs-varid'>messageEnumFrom</span>
+<a name="line-868"></a>        <span class='hs-varid'>enumFromTo</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span><span class='hs-varop'>.</span><span class='hs-varid'>messageEnumFromTo</span>
+<a name="line-869"></a>        <span class='hs-varid'>enumFromThen</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span><span class='hs-varop'>.</span><span class='hs-varid'>messageEnumFromThen</span>
+<a name="line-870"></a>        <span class='hs-varid'>enumFromThenTo</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span><span class='hs-varop'>.</span><span class='hs-varid'>messageEnumFromThenTo</span>
+<a name="line-871"></a>
+<a name="line-872"></a><a name="instance%20Prelude.Bounded%20OptimizerOptions'Level"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bounded</span> <span class='hs-conid'>OptimizerOptions'Level</span> <span class='hs-keyword'>where</span>
+<a name="line-873"></a>        <span class='hs-varid'>minBound</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OptimizerOptions'L0</span>
+<a name="line-874"></a>        <span class='hs-varid'>maxBound</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>OptimizerOptions'L1</span>
+<a name="line-875"></a>
+<a name="line-876"></a><a name="RunMetadata"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>RunMetadata</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>RunMetadata</span><span class='hs-layout'>{</span><span class='hs-sel'>_RunMetadata'stepStats</span> <span class='hs-keyglyph'>::</span>
+<a name="line-877"></a>                               <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>StepStats</span><span class='hs-varop'>.</span><span class='hs-conid'>StepStats</span><span class='hs-layout'>,</span>
+<a name="line-878"></a>                               <span class='hs-sel'>_RunMetadata'costGraph</span> <span class='hs-keyglyph'>::</span>
+<a name="line-879"></a>                               <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-880"></a>                                 <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>CostGraph</span><span class='hs-varop'>.</span><span class='hs-conid'>CostGraphDef</span><span class='hs-layout'>,</span>
+<a name="line-881"></a>                               <span class='hs-sel'>_RunMetadata'partitionGraphs</span> <span class='hs-keyglyph'>::</span>
+<a name="line-882"></a>                               <span class='hs-keyglyph'>[</span><span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Graph</span><span class='hs-varop'>.</span><span class='hs-conid'>GraphDef</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>}</span>
+<a name="line-883"></a>                 <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-884"></a>
+<a name="line-885"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"stepStats"</span> <span class='hs-conid'>RunMetadata</span> <span class='hs-keyglyph'>=</span>
+<a name="line-886"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>StepStats</span><span class='hs-varop'>.</span><span class='hs-conid'>StepStats</span>
+<a name="line-887"></a>
+<a name="line-888"></a><a name="instance%20Data.ProtoLens.HasField%20%22stepStats%22%20RunMetadata%20RunMetadata"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"stepStats"</span> <span class='hs-conid'>RunMetadata</span>
+<a name="line-889"></a>         <span class='hs-conid'>RunMetadata</span> <span class='hs-keyword'>where</span>
+<a name="line-890"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-891"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'stepStats</span>
+<a name="line-892"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-893"></a>
+<a name="line-894"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'stepStats"</span> <span class='hs-conid'>RunMetadata</span> <span class='hs-keyglyph'>=</span>
+<a name="line-895"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span> <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>StepStats</span><span class='hs-varop'>.</span><span class='hs-conid'>StepStats</span>
+<a name="line-896"></a>
+<a name="line-897"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'stepStats%22%20RunMetadata%20RunMetadata"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'stepStats"</span> <span class='hs-conid'>RunMetadata</span>
+<a name="line-898"></a>         <span class='hs-conid'>RunMetadata</span> <span class='hs-keyword'>where</span>
+<a name="line-899"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-900"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_RunMetadata'stepStats</span>
+<a name="line-901"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_RunMetadata'stepStats</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-902"></a>
+<a name="line-903"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"costGraph"</span> <span class='hs-conid'>RunMetadata</span> <span class='hs-keyglyph'>=</span>
+<a name="line-904"></a>     <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>CostGraph</span><span class='hs-varop'>.</span><span class='hs-conid'>CostGraphDef</span>
+<a name="line-905"></a>
+<a name="line-906"></a><a name="instance%20Data.ProtoLens.HasField%20%22costGraph%22%20RunMetadata%20RunMetadata"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"costGraph"</span> <span class='hs-conid'>RunMetadata</span>
+<a name="line-907"></a>         <span class='hs-conid'>RunMetadata</span> <span class='hs-keyword'>where</span>
+<a name="line-908"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-909"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-keyglyph'>..</span><span class='hs-layout'>)</span> <span class='hs-varid'>maybe'costGraph</span>
+<a name="line-910"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeLens</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>)</span>
+<a name="line-911"></a>
+<a name="line-912"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'costGraph"</span> <span class='hs-conid'>RunMetadata</span> <span class='hs-keyglyph'>=</span>
+<a name="line-913"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Maybe</span>
+<a name="line-914"></a>       <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>CostGraph</span><span class='hs-varop'>.</span><span class='hs-conid'>CostGraphDef</span>
+<a name="line-915"></a>
+<a name="line-916"></a><a name="instance%20Data.ProtoLens.HasField%20%22maybe'costGraph%22%20RunMetadata%20RunMetadata"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'costGraph"</span> <span class='hs-conid'>RunMetadata</span>
+<a name="line-917"></a>         <span class='hs-conid'>RunMetadata</span> <span class='hs-keyword'>where</span>
+<a name="line-918"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-919"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_RunMetadata'costGraph</span>
+<a name="line-920"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_RunMetadata'costGraph</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-921"></a>
+<a name="line-922"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"partitionGraphs"</span> <span class='hs-conid'>RunMetadata</span> <span class='hs-keyglyph'>=</span>
+<a name="line-923"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Graph</span><span class='hs-varop'>.</span><span class='hs-conid'>GraphDef</span><span class='hs-keyglyph'>]</span>
+<a name="line-924"></a>
+<a name="line-925"></a><a name="instance%20Data.ProtoLens.HasField%20%22partitionGraphs%22%20RunMetadata%20RunMetadata"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"partitionGraphs"</span> <span class='hs-conid'>RunMetadata</span>
+<a name="line-926"></a>         <span class='hs-conid'>RunMetadata</span> <span class='hs-keyword'>where</span>
+<a name="line-927"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-928"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_RunMetadata'partitionGraphs</span>
+<a name="line-929"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_RunMetadata'partitionGraphs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-930"></a>
+<a name="line-931"></a><a name="instance%20Data.Default.Class.Default%20RunMetadata"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>RunMetadata</span> <span class='hs-keyword'>where</span>
+<a name="line-932"></a>        <span class='hs-varid'>def</span>
+<a name="line-933"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>RunMetadata</span><span class='hs-layout'>{</span><span class='hs-sel'>_RunMetadata'stepStats</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-934"></a>                        <span class='hs-sel'>_RunMetadata'costGraph</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span><span class='hs-layout'>,</span>
+<a name="line-935"></a>                        <span class='hs-sel'>_RunMetadata'partitionGraphs</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>}</span>
+<a name="line-936"></a>
+<a name="line-937"></a><a name="instance%20Data.ProtoLens.Message%20RunMetadata"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>RunMetadata</span> <span class='hs-keyword'>where</span>
+<a name="line-938"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-939"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>stepStats__field_descriptor</span>
+<a name="line-940"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"step_stats"</span>
+<a name="line-941"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-942"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-943"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>StepStats</span><span class='hs-varop'>.</span><span class='hs-conid'>StepStats</span><span class='hs-layout'>)</span>
+<a name="line-944"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'stepStats</span><span class='hs-layout'>)</span>
+<a name="line-945"></a>                <span class='hs-varid'>costGraph__field_descriptor</span>
+<a name="line-946"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"cost_graph"</span>
+<a name="line-947"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-948"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-949"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>CostGraph</span><span class='hs-varop'>.</span><span class='hs-conid'>CostGraphDef</span><span class='hs-layout'>)</span>
+<a name="line-950"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>OptionalField</span> <span class='hs-varid'>maybe'costGraph</span><span class='hs-layout'>)</span>
+<a name="line-951"></a>                <span class='hs-varid'>partitionGraphs__field_descriptor</span>
+<a name="line-952"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"partition_graphs"</span>
+<a name="line-953"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-954"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span>
+<a name="line-955"></a>                           <span class='hs-conid'>Proto</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensorflow</span><span class='hs-varop'>.</span><span class='hs-conid'>Core</span><span class='hs-varop'>.</span><span class='hs-conid'>Framework</span><span class='hs-varop'>.</span><span class='hs-conid'>Graph</span><span class='hs-varop'>.</span><span class='hs-conid'>GraphDef</span><span class='hs-layout'>)</span>
+<a name="line-956"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span>
+<a name="line-957"></a>                         <span class='hs-varid'>partitionGraphs</span><span class='hs-layout'>)</span>
+<a name="line-958"></a>              <span class='hs-keyword'>in</span>
+<a name="line-959"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-960"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-961"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>stepStats__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-962"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>costGraph__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-963"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>partitionGraphs__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-964"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-965"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"step_stats"</span><span class='hs-layout'>,</span> <span class='hs-varid'>stepStats__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-966"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"cost_graph"</span><span class='hs-layout'>,</span> <span class='hs-varid'>costGraph__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-967"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"partition_graphs"</span><span class='hs-layout'>,</span> <span class='hs-varid'>partitionGraphs__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-968"></a>
+<a name="line-969"></a><a name="RunOptions"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>RunOptions</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>RunOptions</span><span class='hs-layout'>{</span><span class='hs-sel'>_RunOptions'traceLevel</span> <span class='hs-keyglyph'>::</span>
+<a name="line-970"></a>                             <span class='hs-conid'>RunOptions'TraceLevel</span><span class='hs-layout'>,</span>
+<a name="line-971"></a>                             <span class='hs-sel'>_RunOptions'timeoutInMs</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>,</span>
+<a name="line-972"></a>                             <span class='hs-sel'>_RunOptions'interOpThreadPool</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>,</span>
+<a name="line-973"></a>                             <span class='hs-sel'>_RunOptions'debugTensorWatchOpts</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyglyph'>[</span><span class='hs-conid'>DebugTensorWatch</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>,</span>
+<a name="line-974"></a>                             <span class='hs-sel'>_RunOptions'outputPartitionGraphs</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>}</span>
+<a name="line-975"></a>                <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-976"></a>
+<a name="line-977"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"traceLevel"</span> <span class='hs-conid'>RunOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-978"></a>     <span class='hs-conid'>RunOptions'TraceLevel</span>
+<a name="line-979"></a>
+<a name="line-980"></a><a name="instance%20Data.ProtoLens.HasField%20%22traceLevel%22%20RunOptions%20RunOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"traceLevel"</span> <span class='hs-conid'>RunOptions</span> <span class='hs-conid'>RunOptions</span>
+<a name="line-981"></a>         <span class='hs-keyword'>where</span>
+<a name="line-982"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-983"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_RunOptions'traceLevel</span>
+<a name="line-984"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_RunOptions'traceLevel</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-985"></a>
+<a name="line-986"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"timeoutInMs"</span> <span class='hs-conid'>RunOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-987"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span>
+<a name="line-988"></a>
+<a name="line-989"></a><a name="instance%20Data.ProtoLens.HasField%20%22timeoutInMs%22%20RunOptions%20RunOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"timeoutInMs"</span> <span class='hs-conid'>RunOptions</span>
+<a name="line-990"></a>         <span class='hs-conid'>RunOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-991"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-992"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_RunOptions'timeoutInMs</span>
+<a name="line-993"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_RunOptions'timeoutInMs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-994"></a>
+<a name="line-995"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"interOpThreadPool"</span> <span class='hs-conid'>RunOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-996"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-997"></a>
+<a name="line-998"></a><a name="instance%20Data.ProtoLens.HasField%20%22interOpThreadPool%22%20RunOptions%20RunOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"interOpThreadPool"</span> <span class='hs-conid'>RunOptions</span>
+<a name="line-999"></a>         <span class='hs-conid'>RunOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-1000"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-1001"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_RunOptions'interOpThreadPool</span>
+<a name="line-1002"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_RunOptions'interOpThreadPool</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-1003"></a>
+<a name="line-1004"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-1005"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"debugTensorWatchOpts"</span> <span class='hs-conid'>RunOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1006"></a>     <span class='hs-keyglyph'>[</span><span class='hs-conid'>DebugTensorWatch</span><span class='hs-keyglyph'>]</span>
+<a name="line-1007"></a>
+<a name="line-1008"></a><a name="instance%20Data.ProtoLens.HasField%20%22debugTensorWatchOpts%22%20RunOptions%20RunOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"debugTensorWatchOpts"</span> <span class='hs-conid'>RunOptions</span>
+<a name="line-1009"></a>         <span class='hs-conid'>RunOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-1010"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-1011"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_RunOptions'debugTensorWatchOpts</span>
+<a name="line-1012"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_RunOptions'debugTensorWatchOpts</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-1013"></a>
+<a name="line-1014"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-1015"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"outputPartitionGraphs"</span> <span class='hs-conid'>RunOptions</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1016"></a>     <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span>
+<a name="line-1017"></a>
+<a name="line-1018"></a><a name="instance%20Data.ProtoLens.HasField%20%22outputPartitionGraphs%22%20RunOptions%20RunOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"outputPartitionGraphs"</span> <span class='hs-conid'>RunOptions</span>
+<a name="line-1019"></a>         <span class='hs-conid'>RunOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-1020"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-1021"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_RunOptions'outputPartitionGraphs</span>
+<a name="line-1022"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_RunOptions'outputPartitionGraphs</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-1023"></a>
+<a name="line-1024"></a><a name="instance%20Data.Default.Class.Default%20RunOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>RunOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-1025"></a>        <span class='hs-varid'>def</span>
+<a name="line-1026"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>RunOptions</span><span class='hs-layout'>{</span><span class='hs-sel'>_RunOptions'traceLevel</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-varid'>def</span><span class='hs-layout'>,</span>
+<a name="line-1027"></a>                       <span class='hs-sel'>_RunOptions'timeoutInMs</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-1028"></a>                       <span class='hs-sel'>_RunOptions'interOpThreadPool</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>,</span>
+<a name="line-1029"></a>                       <span class='hs-sel'>_RunOptions'debugTensorWatchOpts</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>[]</span><span class='hs-layout'>,</span>
+<a name="line-1030"></a>                       <span class='hs-sel'>_RunOptions'outputPartitionGraphs</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-1031"></a>
+<a name="line-1032"></a><a name="instance%20Data.ProtoLens.Message%20RunOptions"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>RunOptions</span> <span class='hs-keyword'>where</span>
+<a name="line-1033"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-1034"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>traceLevel__field_descriptor</span>
+<a name="line-1035"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"trace_level"</span>
+<a name="line-1036"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>EnumField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1037"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>RunOptions'TraceLevel</span><span class='hs-layout'>)</span>
+<a name="line-1038"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>traceLevel</span><span class='hs-layout'>)</span>
+<a name="line-1039"></a>                <span class='hs-varid'>timeoutInMs__field_descriptor</span>
+<a name="line-1040"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"timeout_in_ms"</span>
+<a name="line-1041"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1042"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-1043"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>timeoutInMs</span><span class='hs-layout'>)</span>
+<a name="line-1044"></a>                <span class='hs-varid'>interOpThreadPool__field_descriptor</span>
+<a name="line-1045"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"inter_op_thread_pool"</span>
+<a name="line-1046"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1047"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-1048"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-1049"></a>                         <span class='hs-varid'>interOpThreadPool</span><span class='hs-layout'>)</span>
+<a name="line-1050"></a>                <span class='hs-varid'>debugTensorWatchOpts__field_descriptor</span>
+<a name="line-1051"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"debug_tensor_watch_opts"</span>
+<a name="line-1052"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1053"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>DebugTensorWatch</span><span class='hs-layout'>)</span>
+<a name="line-1054"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>RepeatedField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Unpacked</span>
+<a name="line-1055"></a>                         <span class='hs-varid'>debugTensorWatchOpts</span><span class='hs-layout'>)</span>
+<a name="line-1056"></a>                <span class='hs-varid'>outputPartitionGraphs__field_descriptor</span>
+<a name="line-1057"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"output_partition_graphs"</span>
+<a name="line-1058"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>BoolField</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1059"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bool</span><span class='hs-layout'>)</span>
+<a name="line-1060"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span>
+<a name="line-1061"></a>                         <span class='hs-varid'>outputPartitionGraphs</span><span class='hs-layout'>)</span>
+<a name="line-1062"></a>              <span class='hs-keyword'>in</span>
+<a name="line-1063"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-1064"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-1065"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>traceLevel__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1066"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>2</span><span class='hs-layout'>,</span> <span class='hs-varid'>timeoutInMs__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1067"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>3</span><span class='hs-layout'>,</span> <span class='hs-varid'>interOpThreadPool__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1068"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>4</span><span class='hs-layout'>,</span> <span class='hs-varid'>debugTensorWatchOpts__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1069"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>5</span><span class='hs-layout'>,</span> <span class='hs-varid'>outputPartitionGraphs__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-1070"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-1071"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"trace_level"</span><span class='hs-layout'>,</span> <span class='hs-varid'>traceLevel__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1072"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"timeout_in_ms"</span><span class='hs-layout'>,</span> <span class='hs-varid'>timeoutInMs__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1073"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"inter_op_thread_pool"</span><span class='hs-layout'>,</span> <span class='hs-varid'>interOpThreadPool__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1074"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"debug_tensor_watch_opts"</span><span class='hs-layout'>,</span>
+<a name="line-1075"></a>                     <span class='hs-varid'>debugTensorWatchOpts__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span>
+<a name="line-1076"></a>                    <span class='hs-layout'>(</span><span class='hs-str'>"output_partition_graphs"</span><span class='hs-layout'>,</span>
+<a name="line-1077"></a>                     <span class='hs-varid'>outputPartitionGraphs__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-1078"></a>
+<a name="line-1079"></a><a name="RunOptions'TraceLevel"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>RunOptions'TraceLevel</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>RunOptions'NO_TRACE</span>
+<a name="line-1080"></a>                           <span class='hs-keyglyph'>|</span> <span class='hs-conid'>RunOptions'SOFTWARE_TRACE</span>
+<a name="line-1081"></a>                           <span class='hs-keyglyph'>|</span> <span class='hs-conid'>RunOptions'HARDWARE_TRACE</span>
+<a name="line-1082"></a>                           <span class='hs-keyglyph'>|</span> <span class='hs-conid'>RunOptions'FULL_TRACE</span>
+<a name="line-1083"></a>                           <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-1084"></a>
+<a name="line-1085"></a><a name="instance%20Data.Default.Class.Default%20RunOptions'TraceLevel"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>RunOptions'TraceLevel</span> <span class='hs-keyword'>where</span>
+<a name="line-1086"></a>        <span class='hs-varid'>def</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>RunOptions'NO_TRACE</span>
+<a name="line-1087"></a>
+<a name="line-1088"></a><a name="instance%20Data.ProtoLens.FieldDefault%20RunOptions'TraceLevel"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDefault</span> <span class='hs-conid'>RunOptions'TraceLevel</span> <span class='hs-keyword'>where</span>
+<a name="line-1089"></a>        <span class='hs-varid'>fieldDefault</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>RunOptions'NO_TRACE</span>
+<a name="line-1090"></a>
+<a name="line-1091"></a><a name="instance%20Data.ProtoLens.MessageEnum%20RunOptions'TraceLevel"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageEnum</span> <span class='hs-conid'>RunOptions'TraceLevel</span> <span class='hs-keyword'>where</span>
+<a name="line-1092"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>0</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>RunOptions'NO_TRACE</span>
+<a name="line-1093"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>1</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>RunOptions'SOFTWARE_TRACE</span>
+<a name="line-1094"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>2</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>RunOptions'HARDWARE_TRACE</span>
+<a name="line-1095"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-num'>3</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>RunOptions'FULL_TRACE</span>
+<a name="line-1096"></a>        <span class='hs-varid'>maybeToEnum</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span>
+<a name="line-1097"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>RunOptions'NO_TRACE</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"NO_TRACE"</span>
+<a name="line-1098"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>RunOptions'SOFTWARE_TRACE</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"SOFTWARE_TRACE"</span>
+<a name="line-1099"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>RunOptions'HARDWARE_TRACE</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"HARDWARE_TRACE"</span>
+<a name="line-1100"></a>        <span class='hs-varid'>showEnum</span> <span class='hs-conid'>RunOptions'FULL_TRACE</span> <span class='hs-keyglyph'>=</span> <span class='hs-str'>"FULL_TRACE"</span>
+<a name="line-1101"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"NO_TRACE"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>RunOptions'NO_TRACE</span>
+<a name="line-1102"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"SOFTWARE_TRACE"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>RunOptions'SOFTWARE_TRACE</span>
+<a name="line-1103"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"HARDWARE_TRACE"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>RunOptions'HARDWARE_TRACE</span>
+<a name="line-1104"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-str'>"FULL_TRACE"</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Just</span> <span class='hs-conid'>RunOptions'FULL_TRACE</span>
+<a name="line-1105"></a>        <span class='hs-varid'>readEnum</span> <span class='hs-keyword'>_</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Nothing</span>
+<a name="line-1106"></a>
+<a name="line-1107"></a><a name="instance%20Prelude.Enum%20RunOptions'TraceLevel"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span> <span class='hs-conid'>RunOptions'TraceLevel</span> <span class='hs-keyword'>where</span>
+<a name="line-1108"></a>        <span class='hs-varid'>toEnum</span> <span class='hs-varid'>k__</span>
+<a name="line-1109"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>maybe</span>
+<a name="line-1110"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>error</span>
+<a name="line-1111"></a>                 <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.++</span><span class='hs-layout'>)</span> <span class='hs-str'>"toEnum: unknown value for enum TraceLevel: "</span>
+<a name="line-1112"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>show</span> <span class='hs-varid'>k__</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-1113"></a>              <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>id</span>
+<a name="line-1114"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>maybeToEnum</span> <span class='hs-varid'>k__</span><span class='hs-layout'>)</span>
+<a name="line-1115"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>RunOptions'NO_TRACE</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>0</span>
+<a name="line-1116"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>RunOptions'SOFTWARE_TRACE</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>1</span>
+<a name="line-1117"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>RunOptions'HARDWARE_TRACE</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>2</span>
+<a name="line-1118"></a>        <span class='hs-varid'>fromEnum</span> <span class='hs-conid'>RunOptions'FULL_TRACE</span> <span class='hs-keyglyph'>=</span> <span class='hs-num'>3</span>
+<a name="line-1119"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>RunOptions'FULL_TRACE</span>
+<a name="line-1120"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>error</span>
+<a name="line-1121"></a>              <span class='hs-str'>"Ident \"RunOptions'TraceLevel\".Ident \"succ\": bad argument Ident \"RunOptions'FULL_TRACE\". This value would be out of bounds."</span>
+<a name="line-1122"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>RunOptions'NO_TRACE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>RunOptions'SOFTWARE_TRACE</span>
+<a name="line-1123"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>RunOptions'SOFTWARE_TRACE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>RunOptions'HARDWARE_TRACE</span>
+<a name="line-1124"></a>        <span class='hs-varid'>succ</span> <span class='hs-conid'>RunOptions'HARDWARE_TRACE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>RunOptions'FULL_TRACE</span>
+<a name="line-1125"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>RunOptions'NO_TRACE</span>
+<a name="line-1126"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-varid'>error</span>
+<a name="line-1127"></a>              <span class='hs-str'>"Ident \"RunOptions'TraceLevel\".Ident \"pred\": bad argument Ident \"RunOptions'NO_TRACE\". This value would be out of bounds."</span>
+<a name="line-1128"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>RunOptions'SOFTWARE_TRACE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>RunOptions'NO_TRACE</span>
+<a name="line-1129"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>RunOptions'HARDWARE_TRACE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>RunOptions'SOFTWARE_TRACE</span>
+<a name="line-1130"></a>        <span class='hs-varid'>pred</span> <span class='hs-conid'>RunOptions'FULL_TRACE</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>RunOptions'HARDWARE_TRACE</span>
+<a name="line-1131"></a>        <span class='hs-varid'>enumFrom</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span><span class='hs-varop'>.</span><span class='hs-varid'>messageEnumFrom</span>
+<a name="line-1132"></a>        <span class='hs-varid'>enumFromTo</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span><span class='hs-varop'>.</span><span class='hs-varid'>messageEnumFromTo</span>
+<a name="line-1133"></a>        <span class='hs-varid'>enumFromThen</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span><span class='hs-varop'>.</span><span class='hs-varid'>messageEnumFromThen</span>
+<a name="line-1134"></a>        <span class='hs-varid'>enumFromThenTo</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span><span class='hs-varop'>.</span><span class='hs-conid'>Enum</span><span class='hs-varop'>.</span><span class='hs-varid'>messageEnumFromThenTo</span>
+<a name="line-1135"></a>
+<a name="line-1136"></a><a name="instance%20Prelude.Bounded%20RunOptions'TraceLevel"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Bounded</span> <span class='hs-conid'>RunOptions'TraceLevel</span> <span class='hs-keyword'>where</span>
+<a name="line-1137"></a>        <span class='hs-varid'>minBound</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>RunOptions'NO_TRACE</span>
+<a name="line-1138"></a>        <span class='hs-varid'>maxBound</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>RunOptions'FULL_TRACE</span>
+<a name="line-1139"></a>
+<a name="line-1140"></a><a name="ThreadPoolOptionProto"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>ThreadPoolOptionProto</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>ThreadPoolOptionProto</span><span class='hs-layout'>{</span><span class='hs-sel'>_ThreadPoolOptionProto'numThreads</span>
+<a name="line-1141"></a>                                                   <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>}</span>
+<a name="line-1142"></a>                           <span class='hs-keyword'>deriving</span> <span class='hs-layout'>(</span><span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Show</span><span class='hs-layout'>,</span> <span class='hs-conid'>Prelude</span><span class='hs-varop'>.</span><span class='hs-conid'>Eq</span><span class='hs-layout'>)</span>
+<a name="line-1143"></a>
+<a name="line-1144"></a><span class='hs-keyword'>type</span> <span class='hs-keyword'>instance</span>
+<a name="line-1145"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"numThreads"</span> <span class='hs-conid'>ThreadPoolOptionProto</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1146"></a>     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span>
+<a name="line-1147"></a>
+<a name="line-1148"></a><a name="instance%20Data.ProtoLens.HasField%20%22numThreads%22%20ThreadPoolOptionProto%20ThreadPoolOptionProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"numThreads"</span> <span class='hs-conid'>ThreadPoolOptionProto</span>
+<a name="line-1149"></a>         <span class='hs-conid'>ThreadPoolOptionProto</span> <span class='hs-keyword'>where</span>
+<a name="line-1150"></a>        <span class='hs-varid'>field</span> <span class='hs-keyword'>_</span>
+<a name="line-1151"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Unchecked</span><span class='hs-varop'>.</span><span class='hs-varid'>lens</span> <span class='hs-sel'>_ThreadPoolOptionProto'numThreads</span>
+<a name="line-1152"></a>              <span class='hs-layout'>(</span><span class='hs-keyglyph'>\</span> <span class='hs-varid'>x__</span> <span class='hs-varid'>y__</span> <span class='hs-keyglyph'>-&gt;</span> <span class='hs-varid'>x__</span><span class='hs-layout'>{</span><span class='hs-sel'>_ThreadPoolOptionProto'numThreads</span> <span class='hs-keyglyph'>=</span> <span class='hs-varid'>y__</span><span class='hs-layout'>}</span><span class='hs-layout'>)</span>
+<a name="line-1153"></a>
+<a name="line-1154"></a><a name="instance%20Data.Default.Class.Default%20ThreadPoolOptionProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span><span class='hs-varop'>.</span><span class='hs-conid'>Class</span><span class='hs-varop'>.</span><span class='hs-conid'>Default</span> <span class='hs-conid'>ThreadPoolOptionProto</span> <span class='hs-keyword'>where</span>
+<a name="line-1155"></a>        <span class='hs-varid'>def</span>
+<a name="line-1156"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-conid'>ThreadPoolOptionProto</span><span class='hs-layout'>{</span><span class='hs-sel'>_ThreadPoolOptionProto'numThreads</span> <span class='hs-keyglyph'>=</span>
+<a name="line-1157"></a>                                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>fieldDefault</span><span class='hs-layout'>}</span>
+<a name="line-1158"></a>
+<a name="line-1159"></a><a name="instance%20Data.ProtoLens.Message%20ThreadPoolOptionProto"></a><span class='hs-keyword'>instance</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Message</span> <span class='hs-conid'>ThreadPoolOptionProto</span> <span class='hs-keyword'>where</span>
+<a name="line-1160"></a>        <span class='hs-varid'>descriptor</span>
+<a name="line-1161"></a>          <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>let</span> <span class='hs-varid'>numThreads__field_descriptor</span>
+<a name="line-1162"></a>                  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldDescriptor</span> <span class='hs-str'>"num_threads"</span>
+<a name="line-1163"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32Field</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1164"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>FieldTypeDescriptor</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span><span class='hs-varop'>.</span><span class='hs-conid'>Int32</span><span class='hs-layout'>)</span>
+<a name="line-1165"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>PlainField</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Optional</span> <span class='hs-varid'>numThreads</span><span class='hs-layout'>)</span>
+<a name="line-1166"></a>              <span class='hs-keyword'>in</span>
+<a name="line-1167"></a>              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>MessageDescriptor</span>
+<a name="line-1168"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span>
+<a name="line-1169"></a>                   <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Tag</span> <span class='hs-num'>1</span><span class='hs-layout'>,</span> <span class='hs-varid'>numThreads__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-1170"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Map</span><span class='hs-varop'>.</span><span class='hs-varid'>fromList</span> <span class='hs-keyglyph'>[</span><span class='hs-layout'>(</span><span class='hs-str'>"num_threads"</span><span class='hs-layout'>,</span> <span class='hs-varid'>numThreads__field_descriptor</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-1171"></a>
+<a name="line-1172"></a><a name="allocatorType"></a><span class='hs-definition'>allocatorType</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1173"></a>              <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1174"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allocatorType"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1175"></a>                <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1176"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocatorType"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1177"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allocatorType"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1178"></a><span class='hs-definition'>allocatorType</span>
+<a name="line-1179"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1180"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1181"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"allocatorType"</span><span class='hs-layout'>)</span>
+<a name="line-1182"></a>
+<a name="line-1183"></a><a name="allowGrowth"></a><span class='hs-definition'>allowGrowth</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1184"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allowGrowth"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1185"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allowGrowth"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1186"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allowGrowth"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1187"></a><span class='hs-definition'>allowGrowth</span>
+<a name="line-1188"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1189"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"allowGrowth"</span><span class='hs-layout'>)</span>
+<a name="line-1190"></a>
+<a name="line-1191"></a><a name="allowSoftPlacement"></a><span class='hs-definition'>allowSoftPlacement</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1192"></a>                   <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1193"></a>                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"allowSoftPlacement"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1194"></a>                     <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1195"></a>                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allowSoftPlacement"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1196"></a>                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"allowSoftPlacement"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1197"></a><span class='hs-definition'>allowSoftPlacement</span>
+<a name="line-1198"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1199"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1200"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"allowSoftPlacement"</span><span class='hs-layout'>)</span>
+<a name="line-1201"></a>
+<a name="line-1202"></a><a name="buildCostModel"></a><span class='hs-definition'>buildCostModel</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1203"></a>               <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1204"></a>                 <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"buildCostModel"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1205"></a>                 <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1206"></a>                   <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"buildCostModel"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1207"></a>                   <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"buildCostModel"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1208"></a><span class='hs-definition'>buildCostModel</span>
+<a name="line-1209"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1210"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1211"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"buildCostModel"</span><span class='hs-layout'>)</span>
+<a name="line-1212"></a>
+<a name="line-1213"></a><a name="buildCostModelAfter"></a><span class='hs-definition'>buildCostModelAfter</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1214"></a>                    <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1215"></a>                      <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"buildCostModelAfter"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1216"></a>                      <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1217"></a>                        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"buildCostModelAfter"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1218"></a>                        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"buildCostModelAfter"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1219"></a><span class='hs-definition'>buildCostModelAfter</span>
+<a name="line-1220"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1221"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1222"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"buildCostModelAfter"</span><span class='hs-layout'>)</span>
+<a name="line-1223"></a>
+<a name="line-1224"></a><a name="costGraph"></a><span class='hs-definition'>costGraph</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1225"></a>          <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"costGraph"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1226"></a>            <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"costGraph"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1227"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"costGraph"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1228"></a><span class='hs-definition'>costGraph</span>
+<a name="line-1229"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1230"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"costGraph"</span><span class='hs-layout'>)</span>
+<a name="line-1231"></a>
+<a name="line-1232"></a><a name="debugOps"></a><span class='hs-definition'>debugOps</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1233"></a>         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"debugOps"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1234"></a>           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"debugOps"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1235"></a>             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"debugOps"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1236"></a><span class='hs-definition'>debugOps</span>
+<a name="line-1237"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1238"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"debugOps"</span><span class='hs-layout'>)</span>
+<a name="line-1239"></a>
+<a name="line-1240"></a><a name="debugTensorWatchOpts"></a><span class='hs-definition'>debugTensorWatchOpts</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1241"></a>                     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1242"></a>                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"debugTensorWatchOpts"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1243"></a>                       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1244"></a>                         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"debugTensorWatchOpts"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1245"></a>                         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"debugTensorWatchOpts"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1246"></a><span class='hs-definition'>debugTensorWatchOpts</span>
+<a name="line-1247"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1248"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1249"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"debugTensorWatchOpts"</span><span class='hs-layout'>)</span>
+<a name="line-1250"></a>
+<a name="line-1251"></a><a name="debugUrls"></a><span class='hs-definition'>debugUrls</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1252"></a>          <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"debugUrls"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1253"></a>            <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"debugUrls"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1254"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"debugUrls"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1255"></a><span class='hs-definition'>debugUrls</span>
+<a name="line-1256"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1257"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"debugUrls"</span><span class='hs-layout'>)</span>
+<a name="line-1258"></a>
+<a name="line-1259"></a><a name="deferredDeletionBytes"></a><span class='hs-definition'>deferredDeletionBytes</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1260"></a>                      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1261"></a>                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"deferredDeletionBytes"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1262"></a>                        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1263"></a>                          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"deferredDeletionBytes"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1264"></a>                          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"deferredDeletionBytes"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1265"></a><span class='hs-definition'>deferredDeletionBytes</span>
+<a name="line-1266"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1267"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1268"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"deferredDeletionBytes"</span><span class='hs-layout'>)</span>
+<a name="line-1269"></a>
+<a name="line-1270"></a><a name="deviceCount"></a><span class='hs-definition'>deviceCount</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1271"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"deviceCount"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1272"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"deviceCount"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1273"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"deviceCount"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1274"></a><span class='hs-definition'>deviceCount</span>
+<a name="line-1275"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1276"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"deviceCount"</span><span class='hs-layout'>)</span>
+<a name="line-1277"></a>
+<a name="line-1278"></a><a name="deviceFilters"></a><span class='hs-definition'>deviceFilters</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1279"></a>              <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1280"></a>                <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"deviceFilters"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1281"></a>                <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1282"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"deviceFilters"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1283"></a>                  <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"deviceFilters"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1284"></a><span class='hs-definition'>deviceFilters</span>
+<a name="line-1285"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1286"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1287"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"deviceFilters"</span><span class='hs-layout'>)</span>
+<a name="line-1288"></a>
+<a name="line-1289"></a><a name="doCommonSubexpressionElimination"></a><span class='hs-definition'>doCommonSubexpressionElimination</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1290"></a>                                 <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1291"></a>                                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"doCommonSubexpressionElimination"</span> <span class='hs-varid'>msg</span>
+<a name="line-1292"></a>                                     <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1293"></a>                                   <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1294"></a>                                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"doCommonSubexpressionElimination"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1295"></a>                                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"doCommonSubexpressionElimination"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1296"></a><span class='hs-definition'>doCommonSubexpressionElimination</span>
+<a name="line-1297"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1298"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1299"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"doCommonSubexpressionElimination"</span><span class='hs-layout'>)</span>
+<a name="line-1300"></a>
+<a name="line-1301"></a><a name="doConstantFolding"></a><span class='hs-definition'>doConstantFolding</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1302"></a>                  <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1303"></a>                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"doConstantFolding"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1304"></a>                    <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1305"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"doConstantFolding"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1306"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"doConstantFolding"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1307"></a><span class='hs-definition'>doConstantFolding</span>
+<a name="line-1308"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1309"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1310"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"doConstantFolding"</span><span class='hs-layout'>)</span>
+<a name="line-1311"></a>
+<a name="line-1312"></a><a name="doFunctionInlining"></a><span class='hs-definition'>doFunctionInlining</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1313"></a>                   <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1314"></a>                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"doFunctionInlining"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1315"></a>                     <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1316"></a>                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"doFunctionInlining"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1317"></a>                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"doFunctionInlining"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1318"></a><span class='hs-definition'>doFunctionInlining</span>
+<a name="line-1319"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1320"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1321"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"doFunctionInlining"</span><span class='hs-layout'>)</span>
+<a name="line-1322"></a>
+<a name="line-1323"></a><a name="enableBfloat16Sendrecv"></a><span class='hs-definition'>enableBfloat16Sendrecv</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1324"></a>                       <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1325"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"enableBfloat16Sendrecv"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1326"></a>                         <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1327"></a>                           <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"enableBfloat16Sendrecv"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1328"></a>                           <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"enableBfloat16Sendrecv"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1329"></a><span class='hs-definition'>enableBfloat16Sendrecv</span>
+<a name="line-1330"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1331"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1332"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"enableBfloat16Sendrecv"</span><span class='hs-layout'>)</span>
+<a name="line-1333"></a>
+<a name="line-1334"></a><a name="enableRecvScheduling"></a><span class='hs-definition'>enableRecvScheduling</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1335"></a>                     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1336"></a>                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"enableRecvScheduling"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1337"></a>                       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1338"></a>                         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"enableRecvScheduling"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1339"></a>                         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"enableRecvScheduling"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1340"></a><span class='hs-definition'>enableRecvScheduling</span>
+<a name="line-1341"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1342"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1343"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"enableRecvScheduling"</span><span class='hs-layout'>)</span>
+<a name="line-1344"></a>
+<a name="line-1345"></a><a name="gpuOptions"></a><span class='hs-definition'>gpuOptions</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1346"></a>           <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"gpuOptions"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1347"></a>             <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"gpuOptions"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1348"></a>               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"gpuOptions"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1349"></a><span class='hs-definition'>gpuOptions</span>
+<a name="line-1350"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1351"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"gpuOptions"</span><span class='hs-layout'>)</span>
+<a name="line-1352"></a>
+<a name="line-1353"></a><a name="graphOptions"></a><span class='hs-definition'>graphOptions</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1354"></a>             <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1355"></a>               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"graphOptions"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1356"></a>               <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1357"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"graphOptions"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1358"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"graphOptions"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1359"></a><span class='hs-definition'>graphOptions</span>
+<a name="line-1360"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1361"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"graphOptions"</span><span class='hs-layout'>)</span>
+<a name="line-1362"></a>
+<a name="line-1363"></a><a name="inferShapes"></a><span class='hs-definition'>inferShapes</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1364"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"inferShapes"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1365"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"inferShapes"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1366"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"inferShapes"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1367"></a><span class='hs-definition'>inferShapes</span>
+<a name="line-1368"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1369"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"inferShapes"</span><span class='hs-layout'>)</span>
+<a name="line-1370"></a>
+<a name="line-1371"></a><a name="interOpParallelismThreads"></a><span class='hs-definition'>interOpParallelismThreads</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1372"></a>                          <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1373"></a>                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"interOpParallelismThreads"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1374"></a>                            <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1375"></a>                              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"interOpParallelismThreads"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1376"></a>                              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"interOpParallelismThreads"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1377"></a><span class='hs-definition'>interOpParallelismThreads</span>
+<a name="line-1378"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1379"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1380"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"interOpParallelismThreads"</span><span class='hs-layout'>)</span>
+<a name="line-1381"></a>
+<a name="line-1382"></a><a name="interOpThreadPool"></a><span class='hs-definition'>interOpThreadPool</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1383"></a>                  <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1384"></a>                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"interOpThreadPool"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1385"></a>                    <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1386"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"interOpThreadPool"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1387"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"interOpThreadPool"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1388"></a><span class='hs-definition'>interOpThreadPool</span>
+<a name="line-1389"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1390"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1391"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"interOpThreadPool"</span><span class='hs-layout'>)</span>
+<a name="line-1392"></a>
+<a name="line-1393"></a><a name="intraOpParallelismThreads"></a><span class='hs-definition'>intraOpParallelismThreads</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1394"></a>                          <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1395"></a>                            <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"intraOpParallelismThreads"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1396"></a>                            <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1397"></a>                              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"intraOpParallelismThreads"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1398"></a>                              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"intraOpParallelismThreads"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1399"></a><span class='hs-definition'>intraOpParallelismThreads</span>
+<a name="line-1400"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1401"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1402"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"intraOpParallelismThreads"</span><span class='hs-layout'>)</span>
+<a name="line-1403"></a>
+<a name="line-1404"></a><a name="key"></a><span class='hs-definition'>key</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1405"></a>    <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"key"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1406"></a>      <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"key"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1407"></a>        <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"key"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1408"></a><span class='hs-definition'>key</span>
+<a name="line-1409"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1410"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"key"</span><span class='hs-layout'>)</span>
+<a name="line-1411"></a>
+<a name="line-1412"></a><a name="logDevicePlacement"></a><span class='hs-definition'>logDevicePlacement</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1413"></a>                   <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1414"></a>                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"logDevicePlacement"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1415"></a>                     <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1416"></a>                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"logDevicePlacement"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1417"></a>                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"logDevicePlacement"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1418"></a><span class='hs-definition'>logDevicePlacement</span>
+<a name="line-1419"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1420"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1421"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"logDevicePlacement"</span><span class='hs-layout'>)</span>
+<a name="line-1422"></a>
+<a name="line-1423"></a><a name="maybe'costGraph"></a><span class='hs-definition'>maybe'costGraph</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1424"></a>                <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1425"></a>                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'costGraph"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1426"></a>                  <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1427"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'costGraph"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1428"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'costGraph"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1429"></a><span class='hs-definition'>maybe'costGraph</span>
+<a name="line-1430"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1431"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1432"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'costGraph"</span><span class='hs-layout'>)</span>
+<a name="line-1433"></a>
+<a name="line-1434"></a><a name="maybe'gpuOptions"></a><span class='hs-definition'>maybe'gpuOptions</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1435"></a>                 <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1436"></a>                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'gpuOptions"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1437"></a>                   <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1438"></a>                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'gpuOptions"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1439"></a>                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'gpuOptions"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1440"></a><span class='hs-definition'>maybe'gpuOptions</span>
+<a name="line-1441"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1442"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1443"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'gpuOptions"</span><span class='hs-layout'>)</span>
+<a name="line-1444"></a>
+<a name="line-1445"></a><a name="maybe'graphOptions"></a><span class='hs-definition'>maybe'graphOptions</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1446"></a>                   <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1447"></a>                     <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'graphOptions"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1448"></a>                     <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1449"></a>                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'graphOptions"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1450"></a>                       <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'graphOptions"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1451"></a><span class='hs-definition'>maybe'graphOptions</span>
+<a name="line-1452"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1453"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1454"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'graphOptions"</span><span class='hs-layout'>)</span>
+<a name="line-1455"></a>
+<a name="line-1456"></a><a name="maybe'optimizerOptions"></a><span class='hs-definition'>maybe'optimizerOptions</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1457"></a>                       <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1458"></a>                         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'optimizerOptions"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1459"></a>                         <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1460"></a>                           <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'optimizerOptions"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1461"></a>                           <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'optimizerOptions"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1462"></a><span class='hs-definition'>maybe'optimizerOptions</span>
+<a name="line-1463"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1464"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1465"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'optimizerOptions"</span><span class='hs-layout'>)</span>
+<a name="line-1466"></a>
+<a name="line-1467"></a><a name="maybe'stepStats"></a><span class='hs-definition'>maybe'stepStats</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1468"></a>                <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1469"></a>                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"maybe'stepStats"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1470"></a>                  <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1471"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'stepStats"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1472"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"maybe'stepStats"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1473"></a><span class='hs-definition'>maybe'stepStats</span>
+<a name="line-1474"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1475"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1476"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"maybe'stepStats"</span><span class='hs-layout'>)</span>
+<a name="line-1477"></a>
+<a name="line-1478"></a><a name="nodeName"></a><span class='hs-definition'>nodeName</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1479"></a>         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"nodeName"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1480"></a>           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"nodeName"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1481"></a>             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"nodeName"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1482"></a><span class='hs-definition'>nodeName</span>
+<a name="line-1483"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1484"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"nodeName"</span><span class='hs-layout'>)</span>
+<a name="line-1485"></a>
+<a name="line-1486"></a><a name="numThreads"></a><span class='hs-definition'>numThreads</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1487"></a>           <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"numThreads"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1488"></a>             <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"numThreads"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1489"></a>               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"numThreads"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1490"></a><span class='hs-definition'>numThreads</span>
+<a name="line-1491"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1492"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"numThreads"</span><span class='hs-layout'>)</span>
+<a name="line-1493"></a>
+<a name="line-1494"></a><a name="operationTimeoutInMs"></a><span class='hs-definition'>operationTimeoutInMs</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1495"></a>                     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1496"></a>                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"operationTimeoutInMs"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1497"></a>                       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1498"></a>                         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"operationTimeoutInMs"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1499"></a>                         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"operationTimeoutInMs"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1500"></a><span class='hs-definition'>operationTimeoutInMs</span>
+<a name="line-1501"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1502"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1503"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"operationTimeoutInMs"</span><span class='hs-layout'>)</span>
+<a name="line-1504"></a>
+<a name="line-1505"></a><a name="optLevel"></a><span class='hs-definition'>optLevel</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1506"></a>         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"optLevel"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1507"></a>           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"optLevel"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1508"></a>             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"optLevel"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1509"></a><span class='hs-definition'>optLevel</span>
+<a name="line-1510"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1511"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"optLevel"</span><span class='hs-layout'>)</span>
+<a name="line-1512"></a>
+<a name="line-1513"></a><a name="optimizerOptions"></a><span class='hs-definition'>optimizerOptions</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1514"></a>                 <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1515"></a>                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"optimizerOptions"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1516"></a>                   <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1517"></a>                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"optimizerOptions"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1518"></a>                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"optimizerOptions"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1519"></a><span class='hs-definition'>optimizerOptions</span>
+<a name="line-1520"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1521"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1522"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"optimizerOptions"</span><span class='hs-layout'>)</span>
+<a name="line-1523"></a>
+<a name="line-1524"></a><a name="outputPartitionGraphs"></a><span class='hs-definition'>outputPartitionGraphs</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1525"></a>                      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1526"></a>                        <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"outputPartitionGraphs"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1527"></a>                        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1528"></a>                          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"outputPartitionGraphs"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1529"></a>                          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"outputPartitionGraphs"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1530"></a><span class='hs-definition'>outputPartitionGraphs</span>
+<a name="line-1531"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1532"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1533"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"outputPartitionGraphs"</span><span class='hs-layout'>)</span>
+<a name="line-1534"></a>
+<a name="line-1535"></a><a name="outputSlot"></a><span class='hs-definition'>outputSlot</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1536"></a>           <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"outputSlot"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1537"></a>             <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"outputSlot"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1538"></a>               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"outputSlot"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1539"></a><span class='hs-definition'>outputSlot</span>
+<a name="line-1540"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1541"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"outputSlot"</span><span class='hs-layout'>)</span>
+<a name="line-1542"></a>
+<a name="line-1543"></a><a name="partitionGraphs"></a><span class='hs-definition'>partitionGraphs</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1544"></a>                <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1545"></a>                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"partitionGraphs"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1546"></a>                  <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1547"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"partitionGraphs"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1548"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"partitionGraphs"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1549"></a><span class='hs-definition'>partitionGraphs</span>
+<a name="line-1550"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1551"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1552"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"partitionGraphs"</span><span class='hs-layout'>)</span>
+<a name="line-1553"></a>
+<a name="line-1554"></a><a name="perProcessGpuMemoryFraction"></a><span class='hs-definition'>perProcessGpuMemoryFraction</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1555"></a>                            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1556"></a>                              <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"perProcessGpuMemoryFraction"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1557"></a>                              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1558"></a>                                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"perProcessGpuMemoryFraction"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1559"></a>                                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"perProcessGpuMemoryFraction"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1560"></a><span class='hs-definition'>perProcessGpuMemoryFraction</span>
+<a name="line-1561"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1562"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1563"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"perProcessGpuMemoryFraction"</span><span class='hs-layout'>)</span>
+<a name="line-1564"></a>
+<a name="line-1565"></a><a name="placePrunedGraph"></a><span class='hs-definition'>placePrunedGraph</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1566"></a>                 <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1567"></a>                   <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"placePrunedGraph"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1568"></a>                   <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1569"></a>                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"placePrunedGraph"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1570"></a>                     <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"placePrunedGraph"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1571"></a><span class='hs-definition'>placePrunedGraph</span>
+<a name="line-1572"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1573"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1574"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"placePrunedGraph"</span><span class='hs-layout'>)</span>
+<a name="line-1575"></a>
+<a name="line-1576"></a><a name="placementPeriod"></a><span class='hs-definition'>placementPeriod</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1577"></a>                <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1578"></a>                  <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"placementPeriod"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1579"></a>                  <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1580"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"placementPeriod"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1581"></a>                    <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"placementPeriod"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1582"></a><span class='hs-definition'>placementPeriod</span>
+<a name="line-1583"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1584"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1585"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"placementPeriod"</span><span class='hs-layout'>)</span>
+<a name="line-1586"></a>
+<a name="line-1587"></a><a name="sessionInterOpThreadPool"></a><span class='hs-definition'>sessionInterOpThreadPool</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1588"></a>                         <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1589"></a>                           <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"sessionInterOpThreadPool"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1590"></a>                           <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1591"></a>                             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"sessionInterOpThreadPool"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1592"></a>                             <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"sessionInterOpThreadPool"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1593"></a><span class='hs-definition'>sessionInterOpThreadPool</span>
+<a name="line-1594"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1595"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1596"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"sessionInterOpThreadPool"</span><span class='hs-layout'>)</span>
+<a name="line-1597"></a>
+<a name="line-1598"></a><a name="stepStats"></a><span class='hs-definition'>stepStats</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1599"></a>          <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"stepStats"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1600"></a>            <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"stepStats"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1601"></a>              <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"stepStats"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1602"></a><span class='hs-definition'>stepStats</span>
+<a name="line-1603"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1604"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"stepStats"</span><span class='hs-layout'>)</span>
+<a name="line-1605"></a>
+<a name="line-1606"></a><a name="timelineStep"></a><span class='hs-definition'>timelineStep</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1607"></a>             <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1608"></a>               <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"timelineStep"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1609"></a>               <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1610"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"timelineStep"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1611"></a>                 <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"timelineStep"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1612"></a><span class='hs-definition'>timelineStep</span>
+<a name="line-1613"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1614"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"timelineStep"</span><span class='hs-layout'>)</span>
+<a name="line-1615"></a>
+<a name="line-1616"></a><a name="timeoutInMs"></a><span class='hs-definition'>timeoutInMs</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1617"></a>            <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"timeoutInMs"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1618"></a>              <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"timeoutInMs"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1619"></a>                <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"timeoutInMs"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1620"></a><span class='hs-definition'>timeoutInMs</span>
+<a name="line-1621"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1622"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"timeoutInMs"</span><span class='hs-layout'>)</span>
+<a name="line-1623"></a>
+<a name="line-1624"></a><a name="traceLevel"></a><span class='hs-definition'>traceLevel</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1625"></a>           <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"traceLevel"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1626"></a>             <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"traceLevel"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1627"></a>               <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"traceLevel"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1628"></a><span class='hs-definition'>traceLevel</span>
+<a name="line-1629"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1630"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"traceLevel"</span><span class='hs-layout'>)</span>
+<a name="line-1631"></a>
+<a name="line-1632"></a><a name="usePerSessionThreads"></a><span class='hs-definition'>usePerSessionThreads</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1633"></a>                     <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1634"></a>                       <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"usePerSessionThreads"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1635"></a>                       <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1636"></a>                         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"usePerSessionThreads"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1637"></a>                         <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"usePerSessionThreads"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1638"></a><span class='hs-definition'>usePerSessionThreads</span>
+<a name="line-1639"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1640"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1641"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"usePerSessionThreads"</span><span class='hs-layout'>)</span>
+<a name="line-1642"></a>
+<a name="line-1643"></a><a name="value"></a><span class='hs-definition'>value</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1644"></a>      <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"value"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1645"></a>        <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"value"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1646"></a>          <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"value"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1647"></a><span class='hs-definition'>value</span>
+<a name="line-1648"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1649"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"value"</span><span class='hs-layout'>)</span>
+<a name="line-1650"></a>
+<a name="line-1651"></a><a name="visibleDeviceList"></a><span class='hs-definition'>visibleDeviceList</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1652"></a>                  <span class='hs-keyword'>forall</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-varop'>.</span>
+<a name="line-1653"></a>                    <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>HasField</span> <span class='hs-str'>"visibleDeviceList"</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span> <span class='hs-keyglyph'>=&gt;</span>
+<a name="line-1654"></a>                    <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span><span class='hs-varop'>.</span><span class='hs-conid'>Lens</span> <span class='hs-varid'>msg</span> <span class='hs-varid'>msg'</span>
+<a name="line-1655"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"visibleDeviceList"</span> <span class='hs-varid'>msg</span><span class='hs-layout'>)</span>
+<a name="line-1656"></a>                      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>Field</span> <span class='hs-str'>"visibleDeviceList"</span> <span class='hs-varid'>msg'</span><span class='hs-layout'>)</span>
+<a name="line-1657"></a><span class='hs-definition'>visibleDeviceList</span>
+<a name="line-1658"></a>  <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-varid'>field</span>
+<a name="line-1659"></a>      <span class='hs-layout'>(</span><span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-keyglyph'>::</span>
+<a name="line-1660"></a>         <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ProtoLens</span><span class='hs-varop'>.</span><span class='hs-conid'>ProxySym</span> <span class='hs-str'>"visibleDeviceList"</span><span class='hs-layout'>)</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/src/hscolour.css b/docs/haddock/tensorflow-proto-0.1.0.0/src/hscolour.css
new file mode 100644
index 0000000..c15919e
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/src/hscolour.css
@@ -0,0 +1,5 @@
+.hs-keyglyph, .hs-layout {color: red;}
+.hs-keyword {color: blue;}
+.hs-comment, .hs-comment a {color: green;}
+.hs-str, .hs-chr {color: teal;}
+.hs-keyword, .hs-conid, .hs-varid, .hs-conop, .hs-varop, .hs-num, .hs-cpp, .hs-sel, .hs-definition {}
diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/synopsis.png b/docs/haddock/tensorflow-proto-0.1.0.0/synopsis.png
new file mode 100644
index 0000000000000000000000000000000000000000..85fb86ec84907bcc86531dc82871948ff4d471fa
GIT binary patch
literal 11327
zcmV-FEWp!=P)<h;3K|Lk000e1NJLTq004jh003wR1^@s6IVt!g000+IX+uL$Nkc;*
zP;zf(X>4Tx0C)k_S!GyNTeqHT_l8Y(cXyX`gGi?cY`Qxn1VID|MJXwjPC)?)F$h6K
zMMOd+6hs7sqbPzXbr*U(-*=zy-hcPcUC*=TdiNM(jyd-lv&OpsU|J&v2m2!^0SE{T
z54F(O;<U4O;KWV=44?vZKnREv-NJ%3&CHI1|Ml9M1kj;!OzhCt|7!L>E2!K(!rTCW
z%wV;vdzf1QjBf#e&~gh74F>?Z4a=WLg$KhJ^$5nap>PLbJadS>e&h8+?D`9%QNL`g
zEVKbYGXj7k5Q(8)0Fd#*a?VIMFW3*64geVHKzE-&0BG!BtmfuTbO(T`0Jaeg2nagF
z{V*1E{Wm{e|AvV~*MEExiC+KU-~R=!2{)|c6Bg`GjQ;iG|FQ`1kAUCTuZtQk34#8{
z4r4<P-^=bq9b@>(3g7#|{=Z@d+d#}7f!3C=>=26vx*jwA8>@MS>RG@Tt_zt3hie^T
z_?<zX!Qp0Bf6L=zytR+Pm;lRjJ;HSV_$l=z8Xd-r9G16)Mp#(Gm>0%9VUd=)Fos7I
z^ghPh%Jy%YZ|)vCf6EaFPai$Q-!=$ppK!y&wrJs)bNdAuANB!m3n34Tfj{s75g-&U
z1A!Pg3bcXF-=!Gv1VmU93G2duANT;{0JugFTqg*|oPXPC|A$2HS3NJd-hcPV3EW`Y
zh=1Dr-5Mv{<{zIvz#Yba<V3iQL)~HT!km4Q{?+Cmi$j0EdL!8&xgxpyr%xz;lr%~X
zrG-*OsiBkrCyE}$48%~fC`FVeN*$#F>y&^Vcn^E_`qRfl{{bzYkp)4~$~NAx_VB;E
z{?P)PU)DbV{Qi#~0H0@T9czDj06@6MNq8OrpdAz(9qQxd9nPr<&s+~tPQySqaZyfb
zNh!%g_5YjeaLxMN*$s<dYY{|0@-La=c>v_p;d%b#U$Wpz0Geb0U>E+EOsEQ;I<Re%
zqFY30<Zpkd!v`9W019}7G68nL0|epsEDhv=3eW($zzCQED_{?t;PK-PkHJt71>!&=
zNC6q(BFF<p;2Nj^wV)BSfDUjM^n!=r33v`Bz%-Zxi(nP3gDtQFK@b)qgJ>Wohy&t-
zL?CHM5mJM6p`(xmWDmJOUQi$u0mVUQpbRJ*DuT+OI;a`C4fR4p&?xj8nuk`Puh35f
z5<x)FBG?f82nmEDLK9((utJ<bcp*X%F^FWu1w;X&9C01dj_5@^K};a#5UYqyBtViP
z>5*JWF{C0=8)=GkKzbrWk@3iMWInPS*@Wyu4kE{pbI3L14-^JPgW^Pq!Q<2bWsPz}
zg`nb5nW!REEvg;Wj~YYGqt;RTXfiY_S_G|(HbmQ@z0gtU6m&ki8r_B-Ku@3-(OVb{
zh8`n;QNS2r>@mKWSWG773g!l;2Q!LUz-(f%SSG9pRuyZCC1S&|DcC~nb!<2G1$Gg;
zjU&Zz;G}VSI0sxHE(w>9tH<5Py}&KucJP#VKD;vC6z`6Y#%JLx@m=^4{33pbgo;Ff
zM3uyf#Fr$Iq=2M}WPoIbWP_BHl$%tE)ST3Z^fYM!=}po{r1PXd2-E~&f;PdC5J9*=
zs3G(aUK2LR$jJD~G{_vt!pSa>)sa0QdqcKOPD3tEZb<G<o<Lqq-cJ64{1XM9f{#Lz
z!ige=BA=p_VuWIi5>LrbsZB|wjHfK7yiNI%a+8XNN{Y&qDu61Js-9|yYMB~K%}=dM
z?M|IcT|xbTdVvN>!$YG@<3@9arjllWW|0;{D?n>V>r0zK+erJ2cAbuzPL|Gw?j&6?
z-95TFdL%tRy&=6neHMKS{UrTQ1~vvw1`mcbh9-s=4Br`97&RC@7}FVVFitT3Wa4Df
zW%6UX#MHqw%Zy?cW;SPzV!p~ez`Vvn%c8>K#*)s`!ZO8*U=?PyV2x$1V13HE$;Qs6
z&lb#9$o7D3jh&udgWZ=sm;FBb3I`2`8ix-@E=M=VM@~9UO-_H#0?vNUbuLye1Fi_J
zGOlM_JKO@?*4#<lH@IhcNO%-^Jb7|?`gqoPIe1NY<9Hi*r}*%EihSODm-z<yzVi$4
z+w!OKcknL@FbWt7#0WG9%m|VRstbk)mJ5yxVT2Tg{DiIuJr_m@%L#i67YmPwAVlOv
zd_=B@jEbT~l|%zY%SB&`5yUjbBE=fS=Edp7kBTRWcZh$I;FWNYxFj(su`ekn=`UF+
zIW0vkWh9j()g|>+T3Fgmx>$N#hD=6JCPAiC=8LR|tcUDX*;jHjawc-Aa(!}p@(S{y
z@=fw93cLy~3MC3J6=@aC6f+ecDWR3LloFKgD*aHFR}NQhQU0tVrsAhkud;kZ;E2bO
z$|DP^+^R&?GSxXXPBj;`QnfjCE_I@Mx%xW|9u0SmYK<jLAx&S+>zbdmB(*}d+O)oF
zD{G(9?$JT&=<B5GJkzDtwbm`teWS;%=cU)6x2`XzAFJPEfHp8TxMc9skj>D|u+DJZ
zNWtioQNJ<4*wVPj_}x+AqoGH;Ob{kUCOIZE$M}u~9_ug#riP|Drn6=OW+7&G%rWL>
z=Ede8ETk;rECwxUES)XuEw`++tg@`8tp%+ktov*zY#eRsY`)v-*k;?#*-6-)vU_6B
zZ0}>=>40^xaj16KJg$2@@A#sloMVdP<OzWjCr=DHF+2G?-E*dJCOWsepj~WT>Ron;
zro?jMrmLZAiR-$Xw%cX5Rd)^dT=x|ZRgY|sB~Mk)Y|mvcRj(Yc6>oL#eD5_MZJ#2a
zFTMu8*L=VGnflfE9r)Y&-w413xCGn|qz?28><?lOIvF$?EE1d&JQt!Ik{7ZbY8+Y}
z28Epn>kOxb4~I`91S8Hy%txw47DsMJ*+jLTq&gXR@@ceibXxRMj9yGtEGpJ5wl9t=
zE-`NYl;)|jcqraAzAu3%Avt03wEpSZM3O|m#Ni~#r0k?`XKc@OC9@@;PF^^xf3_io
zJS8;cWvWW*wR5O*KIfjL$)pvg?Wen^KhBWM$j{i#bjy5vUg~_o`GX6d7oKIwXI;IB
zxfpnH@{;<c+H9)qnCyidlbrTk&fN6eZ+XsnL;14#*DjM>j<`HmaI~Pakhkz+;ck(4
z(L}LU@r@GJlC+ZVSKP0>xT<lr=^ER$jB7tjeM(=I>6f*a^OxsWU@9UjK2+LN4pu2v
z)m1ZBXH@Ui1lG*eTGaN}Db&@~v({%dAQ~bXR<1i<f8MCm*x4l3RCa^@M#fEWGvemQ
zW|!v47UPz_R>jt)TYR@l+GyI++oAU8_Vo_$j=4_z&e7XOxBI$Oy4voD->JFFb+`B)
z-My^)B=?i=A9TlbZ}tTDto3^JF7!F~O+T=EFy3$8|7^f`;L$_9hYtod2fH7sKDs-k
zJaqf9;^U4d@=w~I$~|oxmK$z+CjYE`L}8@!xzh8l(IcbxU#P$69n%?mIBq!pWa8Mw
z=%n@JtCx;1=U%zLT7K>S`pZ=0)Xwzj8T3s0Eahze8`d}FZ-w68n3JEoH?K4Q^qu9q
z=>@li)%RiVcNddCkbTHs;#jI%mR`QQqPOz=<MEG+tHG-~Ybk$G{Z;Zw@YAi&8lRu8
z+pRBt3H@^LHFJY$qvo6Jx4uoY%{N;CTRY#=x0$wUe<=PK{Av4hX(wtIyIZg)wAZz7
zygzdgaBvVz3?=?L3n1YV;pqv$*BUs7u!nPp=KzrO{LU$US-|NE1lz#a2I7HkB7Bib
zC_HKj-GeE>CgGy+9whdp4g`BLCvp!8U&;uov(!a2t+bEnRv6HXyi9t`-YglcEo`$K
zI8GTZXYLH1F5YE+b^&9-c%dfYc~N>X1MygiCdpZ8N*OKLV7W5+5rusvVP$KTgd_E;
zV`@J%*flk^Jhjj1)aX9cTQC5ItVZ(2W=FkE;*aH-)|+*kk6SET?pjmWaNEk+>D${o
z_#cmV%sNr-bj$gX%QW$m8{|&wA?SI;%go!uC))SCU%7vKz~jI-L0?1Ap^RZ7;i?hG
zB3+__P9{WW#uUa@#oavB8Q+`m==5;nXwvwZiR6j1<0+%5!{;8Q^`_s>XwIxTUvlAM
z)|rdpmprp=bM$iM@_6#8@((Vr7Q8HcP;{fXs3iGH;8nY8TBRaov}JqcixtC_ZB<vQ
z&(`?YTGlDob2eZb{<=Qi*ws{eBju)Fvqg(SE88vPt>w07?YBCLI#1vB=rX<|d6)j~
z?!9;SA9XkN<n~7QIX}?p7apJ-*nBuW*z>4rDD83J6N{$`!z{xG&lW}=KCd6md=WHe
zF)la3F!5t@`sLkMS6?Sg5vR3gcxTbGOK%>(y*_twKH{Cjg64anMViI^4{J-a%g<LH
ze(YYoyLRud-cLiHC*aZaV}tse#HQ(1*!QCCfgc+?yt~$WY5Tng2XOpFKtGg+ut(A%
z-=j*=VHi8C5{?;<#;=l0k&Y4u$@<72Q1np_Pz_U$(Jatz(CstOF!D30Fk7<(v!=0C
zv-fel=R$LH^XTz<^JVZi3XBSF3Nwow5p@<z60edNko+Kxkr9;Dk#m)gS143$Q5sTy
zqw-A^tHz`*q@kp#r)8miT*qBEKrc)`${^M-!6?x<@o1vS>0=3|@n*5+(H4=G;Z`Bm
z0XDw2UUnY#t`5ZG&W<NeoN#h<c62#GbaHieBf7hKczXJJ`FjWZMEFMgo$^l#ND0gg
zx)_`nQW$zAtSr1TqB^oJs{Uj{bVE#iY<*nqsjB$0ge#{D6LXTzpE;Wxe>ObDFO_)C
zCe0{aEki1k_dNXt+=U-mA1_W_8p^(%Q<IyQmz*DR*{Q&|P@#yYn4);6Wa-M-)$VKc
zrTJw^<pCA;mAX|@)oe8+HQ#IB*NxV9H#A-^X-sd5zTtDzzS*!vsa50_Ya2Pdf^Bpx
zbxz-Y-ZgON&fVsFb@$7<i+XZ;v-&a~r1qx_oPC%wnEEJVDC=?lldDf_hFhL>j|@Mb
z9sM+h7-yIepVWIvd=>Y)XzKR#)XeT1j<?!#-t!mU-CCGlggyu?nJkB_<bCW|eYv*z
ziR!cPy2cmFuWlPb-=a6;w@!ag*pB@X{?m8o#IEU{#=iIg)35lmfin;%)QHeRj3K>H
zI8-@&65hs?W6g0$Tn9b?K9MevmJ{6JljSOT6GbGYHWfM5G<6M41g#z&E8Qx6H$yI?
z50eHn6Z1ODBi1suSavH8F-{EUJXaTYHjh8AJ|73)7XPq7g<!OhsW6Z5mdKE3k(jTz
zvII_ILb6OMKw41-$PCI}kh79!lAl#5RdiS4Q~ID>t>OirQ5IDz)!g7S$y<#pnvPn`
zTCcP(>sag3>W=B<=vx}l7>pa{8`&AN7|$LpGx0noeC)GnyV)so9SefRgyl6WA8Q%w
zeVfO&`F8I1(hk7k+3~B6fhW|RD4pIpx4EPekGo2^q1>k2n?25Xx_BviQ+coYJoGK~
zi}SY&kPV~?{2VkK+z^r;>Jw%VE)ao<Sd8qCsyuln+BHTamLqmAZW>-y@)AN%A4?QY
z!X(X~xtpASHaNvFl_z!g+(cSqdP;^<ru=#G^Gg@HvkESTTr$e$&EC&>mD`$^mG5`i
zpn$&+Rk%>pUtCp^dd2Um*){o6wlZ|t=klqF!OHfk>gs};%-W>7nEHr@(CeX%5lwM7
zQg7xp*S7SwzHLLbOLn+*Uc0?`NAB*$d)wWCJsW)~{h|X4gV%@BpPU*_8L1qd8t0!(
zdySmVd!st{bK%K{=9Rj&=Ffv)KX1|hFxkC)82{hg(&3(fkq6-NB>?O?0kGBtAd?QJ
zm0$~|LIBLj0I*U5i1iA9XzK$|?dCuG2lOlFq=GX}9v}f{nuc(O=>uZH1yBw;!3bD_
zU{(i`gLA_m=mOLPjX+-zbO<GcGvX|w5itp8bfQQ*WD2qkUVrG}Of3%84DUniXj}9p
z^ivET&bv}EkFcb0hLww*!SUjPaCh){ydAy{ze{3HQcDVv9w+S}&=VpFugO%&D#;1t
zQRE9078DOC<tZDexT%V%>8W#QsA+O&>1m7Uxak_`<>>nu%o*kx!<ev4CCswS!z|vc
zXx2(L9kvhb=^T<AuQ(IA#JHxo(|J^QKJ!-bo#1ES9~a0Gv=U+xdMjKf5+SN1MlZG|
z-Y;=YGG59>T2DqomQ{`*59GHMHWa@qZ7S~^!Kl)z@vEz7SZjuAWovinywxMoS2FN7
zEH|1t%4A}H?2754xrD_j%Moi{n>gE7_6iP##}7_;J59Lg5Ifz(-D^B~y{dc!eQ)?H
z1`GsQ2d{)Cgfm98MOmHv9&;s5@6?xs(nO0hxa6LcxN|CLdl`M_GqP+i31t7w9nHU9
zkY40hVt!S*RG^%pl2DDR1@+)Ms)_U_Lks^c#r9*J-d)LeEAIFAEIl9{kQ}rbihXiz
zxOZfJbZ?wtQtXx5l+ld&8>=~scSi5kK8P(dtn9DO{nh=s_)Emb(M`^+uiKA)7VrA)
zEB#tO5ODlSVZM$P@WWh#2Fx+Iz|6u~m`%6|24UXdCqxG`1g0=2kOkd@#-Q&AR(P%P
zMdTpvAy(jBM;jT2tUyk{D~~EF3{{U>K(nFk;T(JdLx-`&6l3PF0@xsI7Y>87!d2q7
z@J9GD{0|aKlAELyq`{<P1SLWl8G$UCY?<7Fe1t-eqMK5MvWrTEsvFK%hiUBK+%$p?
zLsvvEML)#g&IlN<F{v}XV@_oeW0_<<%_hn=1;_jd$2w;fmm@bl_bAT=UK2hFz8Cxj
z0?vX$f}2A9!g(UzqH1CcVqe9dOEgR7NySNf$XLi~$;ru!DDWzBDsd|Fs0baAQPog0
zQ+L;h)y&gs)gIT`*5lMSHi$92ZZv<C*~A7u*Ss=kxA3yOWsS12wQaP=IJh3~J;CFY
z>in5#@A}YP&ZEYQ#XH-V)Gsvv6_^~14ao?j4lj=6k7|w9iW!UZJhhvUlPHq(FxfQ)
zq?V>>q`%8dxgeZ1aw#H*HTOZjUjc35y<*QR6jwV-iRB~}tyPXS=-S45n}+?ysv9OZ
zzqJ(K(rR1j$hs}xHG4PtzG(M&@2Lj@{VyISJQ5#z^W@U7{<DMUcV5Jft50HIPQ0$2
zPMURnYc#L4Ah{^9q_ASN>hV|l=i6Vte3RLV-yYuK+dKCw{z!laG%#N$3ABJM%p<0O
zYA^skKqQbP%m$r-WBwLFh0ujLomRwONMWQ8vL5*f<`CmhgJ?Rm2f718hVj63W7)9r
z*mpQXTq~XnpG|@xNg&xFjU_!Gq>|CVvs#J#1w}9=HDxE2J2egUAWZ`85!yYvKKcv>
zJ4PYKJ*G+KW|m8=VQlv7TJY|}%0<qN;Qq<;jdz`Io&T%Amf)T+R)kiRM@&}SK*Cw_
zq*ShSi_DnpraX&+wqk%%q4J>0wyKDli~41a=UN19Bb{{JVSQ=?d&3H&&qviwE*<+|
zre<Dfv2EpG-EAvqm*;>!9^?4cDF}{Txa*#Kx+jZQvyZXwvVVG@WYFu7)G)>HwaCho
zPBE;pGpDX4cqED@Z6)`nTsY^LE}F4-ek7|Lj+#LpTmF}Vfuf?4z^j_2v}GSEI;v7@
ztn0YySFg7=Mcq_r{?^*qM(m*I?Cd&z=li|$-7G!jeOwO;25=992S<m5AICiD8%90T
z7zuw~Iy(4bWgIcVFe&g-@|E0cxhctM!5QXRlG$%>X5MzsmCeV$vtN*Wk9q%cvGzm6
zlGZYQ`Nc~9M~79`)tR-DzwAEIeH!_EZe4SI`^$~5?i-97Prt=)N^Q<<UHq>3ePg@o
zht*Hi&(<B6UGm+oJ@<X`{k8*#!}y1J%19sX9snX}!Sv&sgM$rz0I=Br>|HuI*eO3a
z*sFk(4fq>KkN@xQ6^F(cm~$_2K14li9;XkV|9<@!M&f%8Nam8p00009a7bBm000XU
z000XU0RWnu7ytkil}SWFRCodHT?u#;Rkr@KbUNvfeG_5`YY-wNfPp{+o{ADgGcxep
z5O;8ydCWk3pWowCbe1RjK4lzy;4&jKqk}U-a1=+ud7z@;LLwlFC>S)v1jwFrI_XY2
zop;WyuIf%_F<t2n-Ic!IS6#QNZr{59Ip>~x?x|CCgE~7q5lBOq0>MKUdH^|7ARquk
zTn+*P5D<VNPJtl~KcE2u0x-lWFvQ`9_7{IpP!Ma2Ag8UrfAvN=8&%31Xc{{x^H{Co
z5Bf3s5%3VAbLh&IH$@c|uDl{TIw}*k&TyNplZ9bWQ*bafH>lHMG@8ELxbaVWHf?&T
znHpfF&E_pZ&^rD;1;7qozi0Q$(`V)7{8<+kI>wdbHk%E>!9AN2eO+^{$KB)hHtVU6
z4;0@%KYw`%{kM%aj|)L>`1``u*EM%B_Ep|f_<QCY-I>7iHT~t6&rZsneaT;XVt##n
z3*O&%0=#!k4Gq$@x_XoAC663)d$?Wm=U<vc73yj8s7C4ye?8%O?|sJH9hZdp8`Rda
zV)AbC6#gCsfWR{`**=IUEL^cLOs^a53WvcKmwX4axn1m?g`lRU8J`}Mu9cL0kEW(p
z*S~w@nA^2_^<7tUJ89F|+dbgAd&6H>XTrha?_sgD)BZa!4dhf)W5g$)o+5f!@!6p=
z7>#E6lGpa0z~7?)*juclePn!mT$U>W2F?VqT7?}(LqHHhL#3+DoNXk5_#Pb{(lwSP
zZ<=X|iSbjYeFoatR`H}3=!RdX3qeSTbc>FTPC<i4yKox}dW8l|7rAq|($3Sz1I6E)
z0PuMxfT6vuJ<7S!uFMY$3&G48V-OV;j??9JIDGgtTjt^~;o&;W${mB481A6nNfvWk
zG}X)k!0qh~kGnVT5kF&`L0DN3I5DFSlBQz~XE21}l1oOBfJjYXO~BgEYVC0M-fop2
zDE{6E#$wJW!`duuVRDQ5Vt#39HNO0^LN1EPNWEbGuwmiKId5c0K$-Du1N3`J+mCki
zh@bx#O+t{9<6v}qdxy{j&CM3X#uzYdYMN4g6Ac50s6((}=&dwR{Cx?4=mpk}4k7Lh
zja}8gbI(8i^%;C~;;e!(COSeeKO#br9p##*Gxqi|^SQmA8UnbUF8H0>&5WKoW3<xS
z+uAzF<~aqE$%3S$Xk2`8s#2WoA?J>vT<}n4p!j<e0B)WjD5-NnRG#WdwxQH=@&k}$
zOR8R0dom`Lc=?)q&RcuB7q?Ti)x-Dbg5R}n#wZVUe2wiO-Z*u-^!Q-=uplma_-CN_
z8Dw8c=~1luSn8d}KeqH68(WbvJ_VB|4Oi+hk~;kO6RAfiC;c;{;CF65PXUWd$`m2j
z>ve)Qtntp05&Y$`N~L&mauhNrjZlt#E%Rdnz*4RdA(~WsS0P~4Cker*^h9K3rID79
zAhx!)2_f*-6tD+E@|~5o_HbR*DQEm#fix64W;xPOIEsuwz3>ej`Mg}wlx+M?%^s;7
zt7<_1|D+24j|zb6{d*Duo)R*nQ%A&N`m}UK6}Gim#oV|jr-^I5{&3u6Y!z0&JjK=N
zf~iA{0UNr_&1RH*=FkdaRxmwXu@ih1pW6b!KwO1@&&hNB<m70YDveg=#B4Xn!v>f0
z=VYU~zns|bF>|Ig{pE8Oi&e4q8Sf>;d>$HnJ*g4^2E{@!BWJXj|MK2>t{)#4iCiKM
z_X3_Wd3!22SVWGECF_5t9Wx1ebd<xtk&rM9vu2GY%sRn*f5AMUxid3;-rS5H$2b;{
z-mVQ6!@O~ym}o<<xL6?O&7IIAJ-pEtPt>Ve1IRabo*K&M<b7Y)<TC!a?+aI7*L)G{
zfCa!8K?T~>e+mp(08G`jsI~A7O*rz=A?*I(Ym_y4*Z<vrc-}ABZsXaPWE_55X~M~}
zTIJaLcOUp224n-2zVikie7yn%YnEfm!5#SRe*5FmzkQB}p11+o{Fp|=>BHj<`2EIL
z@XCfeu<hfsczydWJ6bb}o?S)x0g-o5D5qWnbezFoiTRwunk%qdX1+L6i)|mA5jidY
z{0}jolhGWEzbnk=GvPS&`Ekzpd{Qp}jg2@(@Fp)^ip7aWa9kF~U5c$wF9JUgt8d{b
zE$t=?X?=dwwZ3ox^kn_>GtW8G6RGFlFM<@CjE-OtU#5a;0kB%yXw(N%<3n(~<Qsu8
zZ|+#=eP8Q&gdZK`A?MLQF(CmT9+sYjF8fKIN9rETN>sBeG(H{~)Y9EAyo%kT#Rg2j
zpdOnacnjrpoDswQL%S&=xD)LJZ^c?^7~tUKxVSW2U-+UJ`I8c2{Q|sd4FLUcTr-0M
zaqMa26wFKpz7U~s3AlNV<LRJWL`!?_{RP)vp+K58DGni_Hng{J{G%C}6UQspd^w|9
z18~65>^qhrHMbm9<`9gTLcVV_VCkY<jGCK;u}-q#%LEum+wQu00v693hqg8|E+F?e
zr#1I3!kny;lw%R`{p7X)SelGzum&R`J{*=hGlr!kVHo3VG@_P-B6~JTzO@CWjwnQm
zSR&sif;%k~Gk*Hv$KL#GDAqtA4UzOxf>cW$bp+1aV?*4j`n;5NQvl5P$NHC1)DVqF
ze?14Uta}S5dTDmrRR#Fn;tPAZ>c6M&cw`%zt17X5(`x+mXPZPMYENh$xHA{IIn#Q&
z<WuL|wgH?ccYQ_s*J*Sb=&abC^R8TNFm``&#I=ks9J{Fj!m;GPmSYjgeN40tkN)>^
zG}YF_5*3HIuofIEDMeLB1jc8M#;C+D(d52>)gx`#@~i9ZqkAV_+e~x*&R~QFvHtHw
zX=O8P?QIyJ9<S9QfZh~|rm@{l%9`SLom|1Yuu=8~?mQv6f;~cV->Ss9*B|&g;0hMp
z3Alm-uHb+xn7Ts16&!E{`__2XkJh+p1UhOAxPk+&;D9SQ;0g}7f`^<d*f9ujPKWd{
zNwA(ei(>~4p*<oBsnf?}Y_tY_z+HhWM{(ruZlLc|d{fqlC~^zWp3Kn@&XYx@s8428
z+W?HF@nkVpZ73FT0nX?mGjp-|=hGB_f%7#j<?C8Quwu<K-PpY#ZzNXSa*ba=b=wuJ
zprVklr=6G|NiR*v4kFXBPw?d4Dy67aGq*W0e|U1NJv0<}=fG*4{I=9DV5`;u*19?o
z5{B?pJoNC@$d2dNpxSZdpKs&0Z&je7$@!^)U(mv>Mp`Hum_uHM8Ep9TllPO>m-^Cs
zpVwg1bK6i`-w1z*2vDs7WXVaJJHyU=rk@Vk3#W^iKzdl}7D4^3u#E2B8*>%rGlt8u
z5=Bg)^vMF>N2OW-kTeo=C=#;#Uwg6hiz=At%UPznGuZL$9uX3jIcgXzEoL+}ne7De
zePX!NLIZ__1sfvp<d*$}QF`*n@c|qD`l$#Bn=o#aJJp@HvT{duzX6s;924PRO?YW@
z5nghs!8um&p{H&~w#Fia`1v1)e}&ROaMy!>aY5fTR<P*VpVM{qeXU#N9z3DGRUK^<
z`8ISD3xNMj_Il@&dJSMm!=vkNfHgFv_xE4u6KU4aaHMFg_cJ-U=WN?_Kka)L1Nw>(
zUH5HKQ7-^w@TCk-ATqS$+;^2Y-9Yg{p~En8>~LcE&~OCN2SO-y!qgT7qsff0kWR!$
z^D81!lBm$TfXL;}=Y9YJK+SF{!{d*<g%X#sAaChntiD3qVd2a){!IDu!avzV!TF4o
z9xGgkjB%soL!aKOw*koiw_`(h<}=y@MgPFg@A$O|ZKq#6{$@AkGvc*R?Si#UlQDr#
z3~i~w{&Resf;DUNF*}X?fEFV@D0-Vh!)meXqjD6C9^>=}ZDsk}pA}{0WdF3_)n|T5
zFNK7P(SF;zrP#jx9qieE2>F-K@p;gyHGt(@rI_!hEt)McpP}lbFn3v=a0JCAI=-Ld
z^HfmLKw}#PgVO)j-n&3BpR3@}{)WrPilHHGIK3w22T8R6=u<`rMwjnBh~jFy5zt}A
zN81hv!KkMXNNPDnh1mq7H@>uwma1@k3;2!wtQCOj+9tn%uigkWBw{AL|5)BofhX2&
zA+XZ302%fCsUzg9CimQPVv`f;C6O8|{n>ML#6sZcPqU_9<QKRdEq1%Pk+|_c^D$ZZ
zpTp?8<hlYZ8Sj`5DMd7vdH&JsA-;|ya5+f%=&CEBr`QO8Na{5Jql+JdN1k1QvXkYg
zs3w&k5rOovX-K7)MutMt)1dz*eGsClwG%b<Mnp!CZzRsYY@Y!MK;y}yczWw0oN9FR
ze!+NJMke*DMR;s+x~Aozt~D6zfBzb)s_VMoY^H}`58OE$vvRV#ocF~8^#ULQLiXc}
zUQkuE$PY|=DWHMVx2^eKu~LOA6rK$<8jY@}eezhf09Y%}*qLuY{&mxkmF%SY=ox@6
zF~uhVd#X`JdPSa>DPe!$!>g7c<pOq#ONp0~Ns93gcerW+2#u6Nz)abBxO;&=#J<Sv
z@ytFfgdg<)OXiP7*2GlLRC*q2ahR^|-aI}H3;8oRM~>oyleK6R!5=0O9Kit+4(r(6
ziv6QJ8-P(X4Sa3SakRGjFIv?a0G4_jZD3}d!^RD-cH>&cq5?d2jrKkeAp_;!Ur#;&
z9W7Y4e9epUX=T6m-g%gom8l&2YDT>Vpn#D2K2TLOYC9;D1)wkDRn>N#8T3J_^Lk0W
z2GEDo5^3Wxdgdfd9w7&WOIUcVywJ$#^9sz{H)rNATQUdN%*}+3f?}K#TL)6Cfb&`3
z%&Qjw3IaWJ_$1z;4dDsM&%YQ~=42pUgopbkSWmW!9lu+5e2Bl(Hp~!=)psw#l#5d7
z<59t4!9`Er%bRtn7l4p3WRMY9&31sf7Q0{HC$^-K>G(;07G_Pk5PmWfQbk{$>nD;C
z$aX+;iw(co_@<~Qn^p+B=a%_MiWA>XQ&sn1{z<(6(1#*dufHEF>#Fe8m!&8!F2%dw
zHlg}-8UFYJZG<8tdn)d^eHPNC3G-m$^7_440RBMV3*u1l6Q_-M<NrRRMWepLk*_Lo
z-`bbN{R`%2;?^Z|DbJ;SGxcT1^sB{@@yW=TI0_Rc#C0ow;>ckXuK!rmQ$k)#dR$sG
z@^U71!@qOSF|2)@pOpG;Qm+AE#NKTmpy<6aRJ-8I$ex7UR10>zRSMI&Dx4*+aC%oe
z$>ksZdHCl3@33X-u5M#~!F>8s>bP;(@Z1iZ5DQ57E<W`F&{Exn`H?9pD7b1W{*9JP
zOVBg7&1l$Lik;?q)RUhiU1Oo)Yy>(pe>^RmdH=2Rkv1Y;;r0f4a|kUQI?AO7tZbEf
zJ(*E203jiWBR5FKRnt*$=_L9l06hS)bRb+XpPQ(|6)W>G1u?i-W6WoCJgU<Q$tb+@
zD$Jffj@}LJ{)WH{zrK~CgV-P9o6Kh%ssy0+_)+}f)#J#erD6&w@S}9^J6Kv&%Joq)
zVCl;1aO2ciq)r?I|MikfXnOjsHU;7FXSU<j4Ywhj0zZnsHR6e<UbYvt;`zsJK|)+S
z62!-nogR|)RM~kx=eB4z9^a39Htt8sxmx;YG^zT!qj*qZKBH+t(F@!0k$*VC+$MjO
zDHQiTunA>lRkTWYJ9y;~2lKhQP~5|72z2_#^8q&npdI^OKWZnM4)jd~l<t#{4Y+Cf
zJmk?iep_h!z8~!6w6%M3%WgX$8wEf`;a??H_G2r4_wX%<Bi{h4{7*`b`2}z{1;B(K
z1YRBPTX!e&hj*a(pcMU@bm?3ynVpH1H&2p)w`(&X<ooZ-37-u`n^q8?b=dLG_XXa&
z|9vVMfYDf2c#}A8B~Ph8B=-cMr4A=W1b8$?C6oGZsl=!338XY{hCK?{ZdF8R@clrx
z<=e!HAY|AD>xFIKK%PKOm(9u+`!IG4P>PAtq9@Rh0<!HW6ucJiefk6H>JE!{0DuH!
zkK`y|6ZXDM&ju*fYcM2?dkd?0BQd?AvKl9=rI$l^%Bzo%82<d<-V^wnjLl;2>pwp_
z3!t@d`N^j}MPee&>2}gr!FRvB)4o^~UCPYDMfxiI>b@c+MsVI_ZG?n%#SdILF9)yD
z8iBv~&32h6$j=)^`5;_--)1F7aK==Pycf`JwRRcIa&EjD`NGhX@h9M+TM4YCmA;oJ
zrO3=nv3MeD1n(z%`&dZj&7(JU#eehVv~0XE^yJ%^arZ3+;^s6cinJi_LRv*8MlRsh
z{Xp^er2%-zvwii|iPQND<~cxwB;)S&_u$&{D%8_7aQMh%>8YP30yAe!z=De>;j*0J
zN>6b7(K|VAAJyy)=J$-BZpMp7n5{I{+sN@1<}jm{UYm<6az<l#QupT~-a!ux6n{`>
zC)2KLBDKeY!To$ha&qG2BZqfAotPNM^BbQ^H8u4$*;5z(vZ|_v=c1LgH4&aJ8cR)s
zhZ25=_;#ffO9d0sLd30K^&jiDoI6+3R|Htse-FYDw`bL=buUu;*yY6jR@v$9iMtOO
z{Jm)a77X@ba<Dr`c8-aSjUKyf*>%$<P8>f%7edh>l!!{woQDqvAyLn?wOiY*$B%zo
zv32X~pEWczvH$rLZ56cfy6vr`0a$epDA9d}4E`PkfT>4BU?%e$j!CrfB%e1P1~}M{
zuQ8DZRRHLI>|J6XE5CNbPoY`u^Tv~L_DESt0J@K9biv&;RPgs@1TwMtC4bqg&n_U&
z^RqpU@fmCZV8(Krcxd8Db|Y=v9v+%_sqO*ye5%7a4GH|cY5=AL^#T?U?(IAraOf}Z
znfd(s?_l?Sx}{(;kM%5!ES&ry9?r8?uz9NYQ(Ynr1^j&q08@d8z|&jaWMSaE-1`Sx
z2*lKk?$1KN8*2mJGw(g3`l+riN$dE3Q~;P7L<D9yC)d4ncJO{>Cd=wx?7hW&8J3pu
z_e%g|LIn2Oqk!C_wTCQ#s9zKa2tdEcq}@UR0njdQ`-LnZ0R1A9b_)drK)bx{7qWl=
z^ovZ|Eff#{?eex?$N~b;FEVMjP(T2*%iDe-`+v|7m{y$1dn*6{002ovPDHLkV1lnB
B5rhB$

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-proto-0.1.0.0/tensorflow-proto.txt b/docs/haddock/tensorflow-proto-0.1.0.0/tensorflow-proto.txt
new file mode 100644
index 0000000..095759b
--- /dev/null
+++ b/docs/haddock/tensorflow-proto-0.1.0.0/tensorflow-proto.txt
@@ -0,0 +1,663 @@
+-- Hoogle documentation, generated by Haddock
+-- See Hoogle, http://www.haskell.org/hoogle/
+
+
+-- | TensorFlow protocol buffers.
+--   
+--   Please see README.md
+@package tensorflow-proto
+@version 0.1.0.0
+
+module Proto.Tensorflow.Core.Framework.ResourceHandle
+data ResourceHandle
+ResourceHandle :: Text -> Text -> Text -> Word64 -> Text -> ResourceHandle
+[_ResourceHandle'device] :: ResourceHandle -> Text
+[_ResourceHandle'container] :: ResourceHandle -> Text
+[_ResourceHandle'name] :: ResourceHandle -> Text
+[_ResourceHandle'hashCode] :: ResourceHandle -> Word64
+[_ResourceHandle'maybeTypeName] :: ResourceHandle -> Text
+container :: HasField "container" msg msg' => Lens msg msg' (Field "container" msg) (Field "container" msg')
+device :: HasField "device" msg msg' => Lens msg msg' (Field "device" msg) (Field "device" msg')
+hashCode :: HasField "hashCode" msg msg' => Lens msg msg' (Field "hashCode" msg) (Field "hashCode" msg')
+maybeTypeName :: HasField "maybeTypeName" msg msg' => Lens msg msg' (Field "maybeTypeName" msg) (Field "maybeTypeName" msg')
+name :: HasField "name" msg msg' => Lens msg msg' (Field "name" msg) (Field "name" msg')
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.ResourceHandle.ResourceHandle
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.ResourceHandle.ResourceHandle
+instance Data.ProtoLens.Field.HasField "device" Proto.Tensorflow.Core.Framework.ResourceHandle.ResourceHandle Proto.Tensorflow.Core.Framework.ResourceHandle.ResourceHandle
+instance Data.ProtoLens.Field.HasField "container" Proto.Tensorflow.Core.Framework.ResourceHandle.ResourceHandle Proto.Tensorflow.Core.Framework.ResourceHandle.ResourceHandle
+instance Data.ProtoLens.Field.HasField "name" Proto.Tensorflow.Core.Framework.ResourceHandle.ResourceHandle Proto.Tensorflow.Core.Framework.ResourceHandle.ResourceHandle
+instance Data.ProtoLens.Field.HasField "hashCode" Proto.Tensorflow.Core.Framework.ResourceHandle.ResourceHandle Proto.Tensorflow.Core.Framework.ResourceHandle.ResourceHandle
+instance Data.ProtoLens.Field.HasField "maybeTypeName" Proto.Tensorflow.Core.Framework.ResourceHandle.ResourceHandle Proto.Tensorflow.Core.Framework.ResourceHandle.ResourceHandle
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.ResourceHandle.ResourceHandle
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Framework.ResourceHandle.ResourceHandle
+
+module Proto.Tensorflow.Core.Framework.Types
+data DataType
+DT_INVALID :: DataType
+DT_FLOAT :: DataType
+DT_DOUBLE :: DataType
+DT_INT32 :: DataType
+DT_UINT8 :: DataType
+DT_INT16 :: DataType
+DT_INT8 :: DataType
+DT_STRING :: DataType
+DT_COMPLEX64 :: DataType
+DT_INT64 :: DataType
+DT_BOOL :: DataType
+DT_QINT8 :: DataType
+DT_QUINT8 :: DataType
+DT_QINT32 :: DataType
+DT_BFLOAT16 :: DataType
+DT_QINT16 :: DataType
+DT_QUINT16 :: DataType
+DT_UINT16 :: DataType
+DT_COMPLEX128 :: DataType
+DT_HALF :: DataType
+DT_RESOURCE :: DataType
+DT_FLOAT_REF :: DataType
+DT_DOUBLE_REF :: DataType
+DT_INT32_REF :: DataType
+DT_UINT8_REF :: DataType
+DT_INT16_REF :: DataType
+DT_INT8_REF :: DataType
+DT_STRING_REF :: DataType
+DT_COMPLEX64_REF :: DataType
+DT_INT64_REF :: DataType
+DT_BOOL_REF :: DataType
+DT_QINT8_REF :: DataType
+DT_QUINT8_REF :: DataType
+DT_QINT32_REF :: DataType
+DT_BFLOAT16_REF :: DataType
+DT_QINT16_REF :: DataType
+DT_QUINT16_REF :: DataType
+DT_UINT16_REF :: DataType
+DT_COMPLEX128_REF :: DataType
+DT_HALF_REF :: DataType
+DT_RESOURCE_REF :: DataType
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.Types.DataType
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.Types.DataType
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.Types.DataType
+instance Data.ProtoLens.Message.FieldDefault Proto.Tensorflow.Core.Framework.Types.DataType
+instance Data.ProtoLens.Message.MessageEnum Proto.Tensorflow.Core.Framework.Types.DataType
+instance GHC.Enum.Enum Proto.Tensorflow.Core.Framework.Types.DataType
+instance GHC.Enum.Bounded Proto.Tensorflow.Core.Framework.Types.DataType
+
+module Proto.Tensorflow.Core.Framework.TensorShape
+data TensorShapeProto
+TensorShapeProto :: [TensorShapeProto'Dim] -> Bool -> TensorShapeProto
+[_TensorShapeProto'dim] :: TensorShapeProto -> [TensorShapeProto'Dim]
+[_TensorShapeProto'unknownRank] :: TensorShapeProto -> Bool
+data TensorShapeProto'Dim
+TensorShapeProto'Dim :: Int64 -> Text -> TensorShapeProto'Dim
+[_TensorShapeProto'Dim'size] :: TensorShapeProto'Dim -> Int64
+[_TensorShapeProto'Dim'name] :: TensorShapeProto'Dim -> Text
+dim :: HasField "dim" msg msg' => Lens msg msg' (Field "dim" msg) (Field "dim" msg')
+name :: HasField "name" msg msg' => Lens msg msg' (Field "name" msg) (Field "name" msg')
+size :: HasField "size" msg msg' => Lens msg msg' (Field "size" msg) (Field "size" msg')
+unknownRank :: HasField "unknownRank" msg msg' => Lens msg msg' (Field "unknownRank" msg) (Field "unknownRank" msg')
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.TensorShape.TensorShapeProto
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.TensorShape.TensorShapeProto
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.TensorShape.TensorShapeProto'Dim
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.TensorShape.TensorShapeProto'Dim
+instance Data.ProtoLens.Field.HasField "dim" Proto.Tensorflow.Core.Framework.TensorShape.TensorShapeProto Proto.Tensorflow.Core.Framework.TensorShape.TensorShapeProto
+instance Data.ProtoLens.Field.HasField "unknownRank" Proto.Tensorflow.Core.Framework.TensorShape.TensorShapeProto Proto.Tensorflow.Core.Framework.TensorShape.TensorShapeProto
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.TensorShape.TensorShapeProto
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Framework.TensorShape.TensorShapeProto
+instance Data.ProtoLens.Field.HasField "size" Proto.Tensorflow.Core.Framework.TensorShape.TensorShapeProto'Dim Proto.Tensorflow.Core.Framework.TensorShape.TensorShapeProto'Dim
+instance Data.ProtoLens.Field.HasField "name" Proto.Tensorflow.Core.Framework.TensorShape.TensorShapeProto'Dim Proto.Tensorflow.Core.Framework.TensorShape.TensorShapeProto'Dim
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.TensorShape.TensorShapeProto'Dim
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Framework.TensorShape.TensorShapeProto'Dim
+
+module Proto.Tensorflow.Core.Framework.Tensor
+data TensorProto
+TensorProto :: DataType -> Maybe TensorShapeProto -> Int32 -> ByteString -> [Int32] -> [Float] -> [Double] -> [Int32] -> [ByteString] -> [Float] -> [Int64] -> [Bool] -> [Double] -> [ResourceHandle] -> TensorProto
+[_TensorProto'dtype] :: TensorProto -> DataType
+[_TensorProto'tensorShape] :: TensorProto -> Maybe TensorShapeProto
+[_TensorProto'versionNumber] :: TensorProto -> Int32
+[_TensorProto'tensorContent] :: TensorProto -> ByteString
+[_TensorProto'halfVal] :: TensorProto -> [Int32]
+[_TensorProto'floatVal] :: TensorProto -> [Float]
+[_TensorProto'doubleVal] :: TensorProto -> [Double]
+[_TensorProto'intVal] :: TensorProto -> [Int32]
+[_TensorProto'stringVal] :: TensorProto -> [ByteString]
+[_TensorProto'scomplexVal] :: TensorProto -> [Float]
+[_TensorProto'int64Val] :: TensorProto -> [Int64]
+[_TensorProto'boolVal] :: TensorProto -> [Bool]
+[_TensorProto'dcomplexVal] :: TensorProto -> [Double]
+[_TensorProto'resourceHandleVal] :: TensorProto -> [ResourceHandle]
+boolVal :: HasField "boolVal" msg msg' => Lens msg msg' (Field "boolVal" msg) (Field "boolVal" msg')
+dcomplexVal :: HasField "dcomplexVal" msg msg' => Lens msg msg' (Field "dcomplexVal" msg) (Field "dcomplexVal" msg')
+doubleVal :: HasField "doubleVal" msg msg' => Lens msg msg' (Field "doubleVal" msg) (Field "doubleVal" msg')
+dtype :: HasField "dtype" msg msg' => Lens msg msg' (Field "dtype" msg) (Field "dtype" msg')
+floatVal :: HasField "floatVal" msg msg' => Lens msg msg' (Field "floatVal" msg) (Field "floatVal" msg')
+halfVal :: HasField "halfVal" msg msg' => Lens msg msg' (Field "halfVal" msg) (Field "halfVal" msg')
+int64Val :: HasField "int64Val" msg msg' => Lens msg msg' (Field "int64Val" msg) (Field "int64Val" msg')
+intVal :: HasField "intVal" msg msg' => Lens msg msg' (Field "intVal" msg) (Field "intVal" msg')
+maybe'tensorShape :: HasField "maybe'tensorShape" msg msg' => Lens msg msg' (Field "maybe'tensorShape" msg) (Field "maybe'tensorShape" msg')
+resourceHandleVal :: HasField "resourceHandleVal" msg msg' => Lens msg msg' (Field "resourceHandleVal" msg) (Field "resourceHandleVal" msg')
+scomplexVal :: HasField "scomplexVal" msg msg' => Lens msg msg' (Field "scomplexVal" msg) (Field "scomplexVal" msg')
+stringVal :: HasField "stringVal" msg msg' => Lens msg msg' (Field "stringVal" msg) (Field "stringVal" msg')
+tensorContent :: HasField "tensorContent" msg msg' => Lens msg msg' (Field "tensorContent" msg) (Field "tensorContent" msg')
+tensorShape :: HasField "tensorShape" msg msg' => Lens msg msg' (Field "tensorShape" msg) (Field "tensorShape" msg')
+versionNumber :: HasField "versionNumber" msg msg' => Lens msg msg' (Field "versionNumber" msg) (Field "versionNumber" msg')
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.ProtoLens.Field.HasField "dtype" Proto.Tensorflow.Core.Framework.Tensor.TensorProto Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.ProtoLens.Field.HasField "tensorShape" Proto.Tensorflow.Core.Framework.Tensor.TensorProto Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.ProtoLens.Field.HasField "maybe'tensorShape" Proto.Tensorflow.Core.Framework.Tensor.TensorProto Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.ProtoLens.Field.HasField "versionNumber" Proto.Tensorflow.Core.Framework.Tensor.TensorProto Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.ProtoLens.Field.HasField "tensorContent" Proto.Tensorflow.Core.Framework.Tensor.TensorProto Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.ProtoLens.Field.HasField "halfVal" Proto.Tensorflow.Core.Framework.Tensor.TensorProto Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.ProtoLens.Field.HasField "floatVal" Proto.Tensorflow.Core.Framework.Tensor.TensorProto Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.ProtoLens.Field.HasField "doubleVal" Proto.Tensorflow.Core.Framework.Tensor.TensorProto Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.ProtoLens.Field.HasField "intVal" Proto.Tensorflow.Core.Framework.Tensor.TensorProto Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.ProtoLens.Field.HasField "stringVal" Proto.Tensorflow.Core.Framework.Tensor.TensorProto Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.ProtoLens.Field.HasField "scomplexVal" Proto.Tensorflow.Core.Framework.Tensor.TensorProto Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.ProtoLens.Field.HasField "int64Val" Proto.Tensorflow.Core.Framework.Tensor.TensorProto Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.ProtoLens.Field.HasField "boolVal" Proto.Tensorflow.Core.Framework.Tensor.TensorProto Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.ProtoLens.Field.HasField "dcomplexVal" Proto.Tensorflow.Core.Framework.Tensor.TensorProto Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.ProtoLens.Field.HasField "resourceHandleVal" Proto.Tensorflow.Core.Framework.Tensor.TensorProto Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Framework.Tensor.TensorProto
+
+module Proto.Tensorflow.Core.Framework.AttrValue
+data AttrValue
+AttrValue :: Maybe ByteString -> Maybe Int64 -> Maybe Float -> Maybe Bool -> Maybe DataType -> Maybe TensorShapeProto -> Maybe TensorProto -> Maybe AttrValue'ListValue -> Maybe NameAttrList -> Maybe Text -> AttrValue
+[_AttrValue's] :: AttrValue -> Maybe ByteString
+[_AttrValue'i] :: AttrValue -> Maybe Int64
+[_AttrValue'f] :: AttrValue -> Maybe Float
+[_AttrValue'b] :: AttrValue -> Maybe Bool
+[_AttrValue'type'] :: AttrValue -> Maybe DataType
+[_AttrValue'shape] :: AttrValue -> Maybe TensorShapeProto
+[_AttrValue'tensor] :: AttrValue -> Maybe TensorProto
+[_AttrValue'list] :: AttrValue -> Maybe AttrValue'ListValue
+[_AttrValue'func] :: AttrValue -> Maybe NameAttrList
+[_AttrValue'placeholder] :: AttrValue -> Maybe Text
+data AttrValue'ListValue
+AttrValue'ListValue :: [ByteString] -> [Int64] -> [Float] -> [Bool] -> [DataType] -> [TensorShapeProto] -> [TensorProto] -> AttrValue'ListValue
+[_AttrValue'ListValue's] :: AttrValue'ListValue -> [ByteString]
+[_AttrValue'ListValue'i] :: AttrValue'ListValue -> [Int64]
+[_AttrValue'ListValue'f] :: AttrValue'ListValue -> [Float]
+[_AttrValue'ListValue'b] :: AttrValue'ListValue -> [Bool]
+[_AttrValue'ListValue'type'] :: AttrValue'ListValue -> [DataType]
+[_AttrValue'ListValue'shape] :: AttrValue'ListValue -> [TensorShapeProto]
+[_AttrValue'ListValue'tensor] :: AttrValue'ListValue -> [TensorProto]
+data NameAttrList
+NameAttrList :: Text -> Map Text AttrValue -> NameAttrList
+[_NameAttrList'name] :: NameAttrList -> Text
+[_NameAttrList'attr] :: NameAttrList -> Map Text AttrValue
+data NameAttrList'AttrEntry
+NameAttrList'AttrEntry :: Text -> Maybe AttrValue -> NameAttrList'AttrEntry
+[_NameAttrList'AttrEntry'key] :: NameAttrList'AttrEntry -> Text
+[_NameAttrList'AttrEntry'value] :: NameAttrList'AttrEntry -> Maybe AttrValue
+attr :: HasField "attr" msg msg' => Lens msg msg' (Field "attr" msg) (Field "attr" msg')
+b :: HasField "b" msg msg' => Lens msg msg' (Field "b" msg) (Field "b" msg')
+f :: HasField "f" msg msg' => Lens msg msg' (Field "f" msg) (Field "f" msg')
+func :: HasField "func" msg msg' => Lens msg msg' (Field "func" msg) (Field "func" msg')
+i :: HasField "i" msg msg' => Lens msg msg' (Field "i" msg) (Field "i" msg')
+key :: HasField "key" msg msg' => Lens msg msg' (Field "key" msg) (Field "key" msg')
+list :: HasField "list" msg msg' => Lens msg msg' (Field "list" msg) (Field "list" msg')
+maybe'b :: HasField "maybe'b" msg msg' => Lens msg msg' (Field "maybe'b" msg) (Field "maybe'b" msg')
+maybe'f :: HasField "maybe'f" msg msg' => Lens msg msg' (Field "maybe'f" msg) (Field "maybe'f" msg')
+maybe'func :: HasField "maybe'func" msg msg' => Lens msg msg' (Field "maybe'func" msg) (Field "maybe'func" msg')
+maybe'i :: HasField "maybe'i" msg msg' => Lens msg msg' (Field "maybe'i" msg) (Field "maybe'i" msg')
+maybe'list :: HasField "maybe'list" msg msg' => Lens msg msg' (Field "maybe'list" msg) (Field "maybe'list" msg')
+maybe'placeholder :: HasField "maybe'placeholder" msg msg' => Lens msg msg' (Field "maybe'placeholder" msg) (Field "maybe'placeholder" msg')
+maybe's :: HasField "maybe's" msg msg' => Lens msg msg' (Field "maybe's" msg) (Field "maybe's" msg')
+maybe'shape :: HasField "maybe'shape" msg msg' => Lens msg msg' (Field "maybe'shape" msg) (Field "maybe'shape" msg')
+maybe'tensor :: HasField "maybe'tensor" msg msg' => Lens msg msg' (Field "maybe'tensor" msg) (Field "maybe'tensor" msg')
+maybe'type' :: HasField "maybe'type'" msg msg' => Lens msg msg' (Field "maybe'type'" msg) (Field "maybe'type'" msg')
+maybe'value :: HasField "maybe'value" msg msg' => Lens msg msg' (Field "maybe'value" msg) (Field "maybe'value" msg')
+name :: HasField "name" msg msg' => Lens msg msg' (Field "name" msg) (Field "name" msg')
+placeholder :: HasField "placeholder" msg msg' => Lens msg msg' (Field "placeholder" msg) (Field "placeholder" msg')
+s :: HasField "s" msg msg' => Lens msg msg' (Field "s" msg) (Field "s" msg')
+shape :: HasField "shape" msg msg' => Lens msg msg' (Field "shape" msg) (Field "shape" msg')
+tensor :: HasField "tensor" msg msg' => Lens msg msg' (Field "tensor" msg) (Field "tensor" msg')
+type' :: HasField "type'" msg msg' => Lens msg msg' (Field "type'" msg) (Field "type'" msg')
+value :: HasField "value" msg msg' => Lens msg msg' (Field "value" msg) (Field "value" msg')
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList'AttrEntry
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList'AttrEntry
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue
+instance Data.ProtoLens.Field.HasField "s" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "maybe's" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "i" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "maybe'i" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "f" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "maybe'f" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "b" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "maybe'b" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "type'" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "maybe'type'" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "shape" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "maybe'shape" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "tensor" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "maybe'tensor" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "list" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "maybe'list" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "func" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "maybe'func" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "placeholder" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "maybe'placeholder" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Framework.AttrValue.AttrValue
+instance Data.ProtoLens.Field.HasField "s" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue
+instance Data.ProtoLens.Field.HasField "i" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue
+instance Data.ProtoLens.Field.HasField "f" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue
+instance Data.ProtoLens.Field.HasField "b" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue
+instance Data.ProtoLens.Field.HasField "type'" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue
+instance Data.ProtoLens.Field.HasField "shape" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue
+instance Data.ProtoLens.Field.HasField "tensor" Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Framework.AttrValue.AttrValue'ListValue
+instance Data.ProtoLens.Field.HasField "name" Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList
+instance Data.ProtoLens.Field.HasField "attr" Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList
+instance Data.ProtoLens.Field.HasField "key" Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList'AttrEntry Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList'AttrEntry
+instance Data.ProtoLens.Field.HasField "value" Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList'AttrEntry Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList'AttrEntry
+instance Data.ProtoLens.Field.HasField "maybe'value" Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList'AttrEntry Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList'AttrEntry
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList'AttrEntry
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Framework.AttrValue.NameAttrList'AttrEntry
+
+module Proto.Tensorflow.Core.Framework.NodeDef
+data NodeDef
+NodeDef :: Text -> Text -> [Text] -> Text -> Map Text AttrValue -> NodeDef
+[_NodeDef'name] :: NodeDef -> Text
+[_NodeDef'op] :: NodeDef -> Text
+[_NodeDef'input] :: NodeDef -> [Text]
+[_NodeDef'device] :: NodeDef -> Text
+[_NodeDef'attr] :: NodeDef -> Map Text AttrValue
+data NodeDef'AttrEntry
+NodeDef'AttrEntry :: Text -> Maybe AttrValue -> NodeDef'AttrEntry
+[_NodeDef'AttrEntry'key] :: NodeDef'AttrEntry -> Text
+[_NodeDef'AttrEntry'value] :: NodeDef'AttrEntry -> Maybe AttrValue
+attr :: HasField "attr" msg msg' => Lens msg msg' (Field "attr" msg) (Field "attr" msg')
+device :: HasField "device" msg msg' => Lens msg msg' (Field "device" msg) (Field "device" msg')
+input :: HasField "input" msg msg' => Lens msg msg' (Field "input" msg) (Field "input" msg')
+key :: HasField "key" msg msg' => Lens msg msg' (Field "key" msg) (Field "key" msg')
+maybe'value :: HasField "maybe'value" msg msg' => Lens msg msg' (Field "maybe'value" msg) (Field "maybe'value" msg')
+name :: HasField "name" msg msg' => Lens msg msg' (Field "name" msg) (Field "name" msg')
+op :: HasField "op" msg msg' => Lens msg msg' (Field "op" msg) (Field "op" msg')
+value :: HasField "value" msg msg' => Lens msg msg' (Field "value" msg) (Field "value" msg')
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.NodeDef.NodeDef'AttrEntry
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.NodeDef.NodeDef'AttrEntry
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.NodeDef.NodeDef
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.NodeDef.NodeDef
+instance Data.ProtoLens.Field.HasField "name" Proto.Tensorflow.Core.Framework.NodeDef.NodeDef Proto.Tensorflow.Core.Framework.NodeDef.NodeDef
+instance Data.ProtoLens.Field.HasField "op" Proto.Tensorflow.Core.Framework.NodeDef.NodeDef Proto.Tensorflow.Core.Framework.NodeDef.NodeDef
+instance Data.ProtoLens.Field.HasField "input" Proto.Tensorflow.Core.Framework.NodeDef.NodeDef Proto.Tensorflow.Core.Framework.NodeDef.NodeDef
+instance Data.ProtoLens.Field.HasField "device" Proto.Tensorflow.Core.Framework.NodeDef.NodeDef Proto.Tensorflow.Core.Framework.NodeDef.NodeDef
+instance Data.ProtoLens.Field.HasField "attr" Proto.Tensorflow.Core.Framework.NodeDef.NodeDef Proto.Tensorflow.Core.Framework.NodeDef.NodeDef
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.NodeDef.NodeDef
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Framework.NodeDef.NodeDef
+instance Data.ProtoLens.Field.HasField "key" Proto.Tensorflow.Core.Framework.NodeDef.NodeDef'AttrEntry Proto.Tensorflow.Core.Framework.NodeDef.NodeDef'AttrEntry
+instance Data.ProtoLens.Field.HasField "value" Proto.Tensorflow.Core.Framework.NodeDef.NodeDef'AttrEntry Proto.Tensorflow.Core.Framework.NodeDef.NodeDef'AttrEntry
+instance Data.ProtoLens.Field.HasField "maybe'value" Proto.Tensorflow.Core.Framework.NodeDef.NodeDef'AttrEntry Proto.Tensorflow.Core.Framework.NodeDef.NodeDef'AttrEntry
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.NodeDef.NodeDef'AttrEntry
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Framework.NodeDef.NodeDef'AttrEntry
+
+module Proto.Tensorflow.Core.Framework.OpDef
+data OpDef
+OpDef :: Text -> [OpDef'ArgDef] -> [OpDef'ArgDef] -> [OpDef'AttrDef] -> Maybe OpDeprecation -> Text -> Text -> Bool -> Bool -> Bool -> Bool -> OpDef
+[_OpDef'name] :: OpDef -> Text
+[_OpDef'inputArg] :: OpDef -> [OpDef'ArgDef]
+[_OpDef'outputArg] :: OpDef -> [OpDef'ArgDef]
+[_OpDef'attr] :: OpDef -> [OpDef'AttrDef]
+[_OpDef'deprecation] :: OpDef -> Maybe OpDeprecation
+[_OpDef'summary] :: OpDef -> Text
+[_OpDef'description] :: OpDef -> Text
+[_OpDef'isCommutative] :: OpDef -> Bool
+[_OpDef'isAggregate] :: OpDef -> Bool
+[_OpDef'isStateful] :: OpDef -> Bool
+[_OpDef'allowsUninitializedInput] :: OpDef -> Bool
+data OpDef'ArgDef
+OpDef'ArgDef :: Text -> Text -> DataType -> Text -> Text -> Text -> Bool -> OpDef'ArgDef
+[_OpDef'ArgDef'name] :: OpDef'ArgDef -> Text
+[_OpDef'ArgDef'description] :: OpDef'ArgDef -> Text
+[_OpDef'ArgDef'type'] :: OpDef'ArgDef -> DataType
+[_OpDef'ArgDef'typeAttr] :: OpDef'ArgDef -> Text
+[_OpDef'ArgDef'numberAttr] :: OpDef'ArgDef -> Text
+[_OpDef'ArgDef'typeListAttr] :: OpDef'ArgDef -> Text
+[_OpDef'ArgDef'isRef] :: OpDef'ArgDef -> Bool
+data OpDef'AttrDef
+OpDef'AttrDef :: Text -> Text -> Maybe AttrValue -> Text -> Bool -> Int64 -> Maybe AttrValue -> OpDef'AttrDef
+[_OpDef'AttrDef'name] :: OpDef'AttrDef -> Text
+[_OpDef'AttrDef'type'] :: OpDef'AttrDef -> Text
+[_OpDef'AttrDef'defaultValue] :: OpDef'AttrDef -> Maybe AttrValue
+[_OpDef'AttrDef'description] :: OpDef'AttrDef -> Text
+[_OpDef'AttrDef'hasMinimum] :: OpDef'AttrDef -> Bool
+[_OpDef'AttrDef'minimum] :: OpDef'AttrDef -> Int64
+[_OpDef'AttrDef'allowedValues] :: OpDef'AttrDef -> Maybe AttrValue
+data OpDeprecation
+OpDeprecation :: Int32 -> Text -> OpDeprecation
+[_OpDeprecation'version] :: OpDeprecation -> Int32
+[_OpDeprecation'explanation] :: OpDeprecation -> Text
+data OpList
+OpList :: [OpDef] -> OpList
+[_OpList'op] :: OpList -> [OpDef]
+allowedValues :: HasField "allowedValues" msg msg' => Lens msg msg' (Field "allowedValues" msg) (Field "allowedValues" msg')
+allowsUninitializedInput :: HasField "allowsUninitializedInput" msg msg' => Lens msg msg' (Field "allowsUninitializedInput" msg) (Field "allowsUninitializedInput" msg')
+attr :: HasField "attr" msg msg' => Lens msg msg' (Field "attr" msg) (Field "attr" msg')
+defaultValue :: HasField "defaultValue" msg msg' => Lens msg msg' (Field "defaultValue" msg) (Field "defaultValue" msg')
+deprecation :: HasField "deprecation" msg msg' => Lens msg msg' (Field "deprecation" msg) (Field "deprecation" msg')
+description :: HasField "description" msg msg' => Lens msg msg' (Field "description" msg) (Field "description" msg')
+explanation :: HasField "explanation" msg msg' => Lens msg msg' (Field "explanation" msg) (Field "explanation" msg')
+hasMinimum :: HasField "hasMinimum" msg msg' => Lens msg msg' (Field "hasMinimum" msg) (Field "hasMinimum" msg')
+inputArg :: HasField "inputArg" msg msg' => Lens msg msg' (Field "inputArg" msg) (Field "inputArg" msg')
+isAggregate :: HasField "isAggregate" msg msg' => Lens msg msg' (Field "isAggregate" msg) (Field "isAggregate" msg')
+isCommutative :: HasField "isCommutative" msg msg' => Lens msg msg' (Field "isCommutative" msg) (Field "isCommutative" msg')
+isRef :: HasField "isRef" msg msg' => Lens msg msg' (Field "isRef" msg) (Field "isRef" msg')
+isStateful :: HasField "isStateful" msg msg' => Lens msg msg' (Field "isStateful" msg) (Field "isStateful" msg')
+maybe'allowedValues :: HasField "maybe'allowedValues" msg msg' => Lens msg msg' (Field "maybe'allowedValues" msg) (Field "maybe'allowedValues" msg')
+maybe'defaultValue :: HasField "maybe'defaultValue" msg msg' => Lens msg msg' (Field "maybe'defaultValue" msg) (Field "maybe'defaultValue" msg')
+maybe'deprecation :: HasField "maybe'deprecation" msg msg' => Lens msg msg' (Field "maybe'deprecation" msg) (Field "maybe'deprecation" msg')
+minimum :: HasField "minimum" msg msg' => Lens msg msg' (Field "minimum" msg) (Field "minimum" msg')
+name :: HasField "name" msg msg' => Lens msg msg' (Field "name" msg) (Field "name" msg')
+numberAttr :: HasField "numberAttr" msg msg' => Lens msg msg' (Field "numberAttr" msg) (Field "numberAttr" msg')
+op :: HasField "op" msg msg' => Lens msg msg' (Field "op" msg) (Field "op" msg')
+outputArg :: HasField "outputArg" msg msg' => Lens msg msg' (Field "outputArg" msg) (Field "outputArg" msg')
+summary :: HasField "summary" msg msg' => Lens msg msg' (Field "summary" msg) (Field "summary" msg')
+type' :: HasField "type'" msg msg' => Lens msg msg' (Field "type'" msg) (Field "type'" msg')
+typeAttr :: HasField "typeAttr" msg msg' => Lens msg msg' (Field "typeAttr" msg) (Field "typeAttr" msg')
+typeListAttr :: HasField "typeListAttr" msg msg' => Lens msg msg' (Field "typeListAttr" msg) (Field "typeListAttr" msg')
+version :: HasField "version" msg msg' => Lens msg msg' (Field "version" msg) (Field "version" msg')
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.OpDef.OpList
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.OpDef.OpList
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.OpDef.OpDef
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.OpDef.OpDef
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.OpDef.OpDeprecation
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.OpDef.OpDeprecation
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef
+instance Data.ProtoLens.Field.HasField "name" Proto.Tensorflow.Core.Framework.OpDef.OpDef Proto.Tensorflow.Core.Framework.OpDef.OpDef
+instance Data.ProtoLens.Field.HasField "inputArg" Proto.Tensorflow.Core.Framework.OpDef.OpDef Proto.Tensorflow.Core.Framework.OpDef.OpDef
+instance Data.ProtoLens.Field.HasField "outputArg" Proto.Tensorflow.Core.Framework.OpDef.OpDef Proto.Tensorflow.Core.Framework.OpDef.OpDef
+instance Data.ProtoLens.Field.HasField "attr" Proto.Tensorflow.Core.Framework.OpDef.OpDef Proto.Tensorflow.Core.Framework.OpDef.OpDef
+instance Data.ProtoLens.Field.HasField "deprecation" Proto.Tensorflow.Core.Framework.OpDef.OpDef Proto.Tensorflow.Core.Framework.OpDef.OpDef
+instance Data.ProtoLens.Field.HasField "maybe'deprecation" Proto.Tensorflow.Core.Framework.OpDef.OpDef Proto.Tensorflow.Core.Framework.OpDef.OpDef
+instance Data.ProtoLens.Field.HasField "summary" Proto.Tensorflow.Core.Framework.OpDef.OpDef Proto.Tensorflow.Core.Framework.OpDef.OpDef
+instance Data.ProtoLens.Field.HasField "description" Proto.Tensorflow.Core.Framework.OpDef.OpDef Proto.Tensorflow.Core.Framework.OpDef.OpDef
+instance Data.ProtoLens.Field.HasField "isCommutative" Proto.Tensorflow.Core.Framework.OpDef.OpDef Proto.Tensorflow.Core.Framework.OpDef.OpDef
+instance Data.ProtoLens.Field.HasField "isAggregate" Proto.Tensorflow.Core.Framework.OpDef.OpDef Proto.Tensorflow.Core.Framework.OpDef.OpDef
+instance Data.ProtoLens.Field.HasField "isStateful" Proto.Tensorflow.Core.Framework.OpDef.OpDef Proto.Tensorflow.Core.Framework.OpDef.OpDef
+instance Data.ProtoLens.Field.HasField "allowsUninitializedInput" Proto.Tensorflow.Core.Framework.OpDef.OpDef Proto.Tensorflow.Core.Framework.OpDef.OpDef
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.OpDef.OpDef
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Framework.OpDef.OpDef
+instance Data.ProtoLens.Field.HasField "name" Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef
+instance Data.ProtoLens.Field.HasField "description" Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef
+instance Data.ProtoLens.Field.HasField "type'" Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef
+instance Data.ProtoLens.Field.HasField "typeAttr" Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef
+instance Data.ProtoLens.Field.HasField "numberAttr" Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef
+instance Data.ProtoLens.Field.HasField "typeListAttr" Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef
+instance Data.ProtoLens.Field.HasField "isRef" Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Framework.OpDef.OpDef'ArgDef
+instance Data.ProtoLens.Field.HasField "name" Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef
+instance Data.ProtoLens.Field.HasField "type'" Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef
+instance Data.ProtoLens.Field.HasField "defaultValue" Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef
+instance Data.ProtoLens.Field.HasField "maybe'defaultValue" Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef
+instance Data.ProtoLens.Field.HasField "description" Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef
+instance Data.ProtoLens.Field.HasField "hasMinimum" Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef
+instance Data.ProtoLens.Field.HasField "minimum" Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef
+instance Data.ProtoLens.Field.HasField "allowedValues" Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef
+instance Data.ProtoLens.Field.HasField "maybe'allowedValues" Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Framework.OpDef.OpDef'AttrDef
+instance Data.ProtoLens.Field.HasField "version" Proto.Tensorflow.Core.Framework.OpDef.OpDeprecation Proto.Tensorflow.Core.Framework.OpDef.OpDeprecation
+instance Data.ProtoLens.Field.HasField "explanation" Proto.Tensorflow.Core.Framework.OpDef.OpDeprecation Proto.Tensorflow.Core.Framework.OpDef.OpDeprecation
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.OpDef.OpDeprecation
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Framework.OpDef.OpDeprecation
+instance Data.ProtoLens.Field.HasField "op" Proto.Tensorflow.Core.Framework.OpDef.OpList Proto.Tensorflow.Core.Framework.OpDef.OpList
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.OpDef.OpList
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Framework.OpDef.OpList
+
+module Proto.Tensorflow.Core.Framework.Graph
+data GraphDef
+GraphDef :: [NodeDef] -> Maybe VersionDef -> Int32 -> Maybe FunctionDefLibrary -> GraphDef
+[_GraphDef'node] :: GraphDef -> [NodeDef]
+[_GraphDef'versions] :: GraphDef -> Maybe VersionDef
+[_GraphDef'version] :: GraphDef -> Int32
+[_GraphDef'library] :: GraphDef -> Maybe FunctionDefLibrary
+library :: HasField "library" msg msg' => Lens msg msg' (Field "library" msg) (Field "library" msg')
+maybe'library :: HasField "maybe'library" msg msg' => Lens msg msg' (Field "maybe'library" msg) (Field "maybe'library" msg')
+maybe'versions :: HasField "maybe'versions" msg msg' => Lens msg msg' (Field "maybe'versions" msg) (Field "maybe'versions" msg')
+node :: HasField "node" msg msg' => Lens msg msg' (Field "node" msg) (Field "node" msg')
+version :: HasField "version" msg msg' => Lens msg msg' (Field "version" msg) (Field "version" msg')
+versions :: HasField "versions" msg msg' => Lens msg msg' (Field "versions" msg) (Field "versions" msg')
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Framework.Graph.GraphDef
+instance GHC.Show.Show Proto.Tensorflow.Core.Framework.Graph.GraphDef
+instance Data.ProtoLens.Field.HasField "node" Proto.Tensorflow.Core.Framework.Graph.GraphDef Proto.Tensorflow.Core.Framework.Graph.GraphDef
+instance Data.ProtoLens.Field.HasField "versions" Proto.Tensorflow.Core.Framework.Graph.GraphDef Proto.Tensorflow.Core.Framework.Graph.GraphDef
+instance Data.ProtoLens.Field.HasField "maybe'versions" Proto.Tensorflow.Core.Framework.Graph.GraphDef Proto.Tensorflow.Core.Framework.Graph.GraphDef
+instance Data.ProtoLens.Field.HasField "version" Proto.Tensorflow.Core.Framework.Graph.GraphDef Proto.Tensorflow.Core.Framework.Graph.GraphDef
+instance Data.ProtoLens.Field.HasField "library" Proto.Tensorflow.Core.Framework.Graph.GraphDef Proto.Tensorflow.Core.Framework.Graph.GraphDef
+instance Data.ProtoLens.Field.HasField "maybe'library" Proto.Tensorflow.Core.Framework.Graph.GraphDef Proto.Tensorflow.Core.Framework.Graph.GraphDef
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Framework.Graph.GraphDef
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Framework.Graph.GraphDef
+
+module Proto.Tensorflow.Core.Protobuf.Config
+data ConfigProto
+ConfigProto :: Map Text Int32 -> Int32 -> Int32 -> Bool -> [ThreadPoolOptionProto] -> Int32 -> [Text] -> Maybe GPUOptions -> Bool -> Bool -> Maybe GraphOptions -> Int64 -> ConfigProto
+[_ConfigProto'deviceCount] :: ConfigProto -> Map Text Int32
+[_ConfigProto'intraOpParallelismThreads] :: ConfigProto -> Int32
+[_ConfigProto'interOpParallelismThreads] :: ConfigProto -> Int32
+[_ConfigProto'usePerSessionThreads] :: ConfigProto -> Bool
+[_ConfigProto'sessionInterOpThreadPool] :: ConfigProto -> [ThreadPoolOptionProto]
+[_ConfigProto'placementPeriod] :: ConfigProto -> Int32
+[_ConfigProto'deviceFilters] :: ConfigProto -> [Text]
+[_ConfigProto'gpuOptions] :: ConfigProto -> Maybe GPUOptions
+[_ConfigProto'allowSoftPlacement] :: ConfigProto -> Bool
+[_ConfigProto'logDevicePlacement] :: ConfigProto -> Bool
+[_ConfigProto'graphOptions] :: ConfigProto -> Maybe GraphOptions
+[_ConfigProto'operationTimeoutInMs] :: ConfigProto -> Int64
+data ConfigProto'DeviceCountEntry
+ConfigProto'DeviceCountEntry :: Text -> Int32 -> ConfigProto'DeviceCountEntry
+[_ConfigProto'DeviceCountEntry'key] :: ConfigProto'DeviceCountEntry -> Text
+[_ConfigProto'DeviceCountEntry'value] :: ConfigProto'DeviceCountEntry -> Int32
+data DebugTensorWatch
+DebugTensorWatch :: Text -> Int32 -> [Text] -> [Text] -> DebugTensorWatch
+[_DebugTensorWatch'nodeName] :: DebugTensorWatch -> Text
+[_DebugTensorWatch'outputSlot] :: DebugTensorWatch -> Int32
+[_DebugTensorWatch'debugOps] :: DebugTensorWatch -> [Text]
+[_DebugTensorWatch'debugUrls] :: DebugTensorWatch -> [Text]
+data GPUOptions
+GPUOptions :: Double -> Text -> Int64 -> Bool -> Text -> GPUOptions
+[_GPUOptions'perProcessGpuMemoryFraction] :: GPUOptions -> Double
+[_GPUOptions'allocatorType] :: GPUOptions -> Text
+[_GPUOptions'deferredDeletionBytes] :: GPUOptions -> Int64
+[_GPUOptions'allowGrowth] :: GPUOptions -> Bool
+[_GPUOptions'visibleDeviceList] :: GPUOptions -> Text
+data GraphOptions
+GraphOptions :: Bool -> Maybe OptimizerOptions -> Int64 -> Int64 -> Bool -> Bool -> Bool -> Int32 -> GraphOptions
+[_GraphOptions'enableRecvScheduling] :: GraphOptions -> Bool
+[_GraphOptions'optimizerOptions] :: GraphOptions -> Maybe OptimizerOptions
+[_GraphOptions'buildCostModel] :: GraphOptions -> Int64
+[_GraphOptions'buildCostModelAfter] :: GraphOptions -> Int64
+[_GraphOptions'inferShapes] :: GraphOptions -> Bool
+[_GraphOptions'placePrunedGraph] :: GraphOptions -> Bool
+[_GraphOptions'enableBfloat16Sendrecv] :: GraphOptions -> Bool
+[_GraphOptions'timelineStep] :: GraphOptions -> Int32
+data OptimizerOptions
+OptimizerOptions :: Bool -> Bool -> Bool -> OptimizerOptions'Level -> OptimizerOptions
+[_OptimizerOptions'doCommonSubexpressionElimination] :: OptimizerOptions -> Bool
+[_OptimizerOptions'doConstantFolding] :: OptimizerOptions -> Bool
+[_OptimizerOptions'doFunctionInlining] :: OptimizerOptions -> Bool
+[_OptimizerOptions'optLevel] :: OptimizerOptions -> OptimizerOptions'Level
+data OptimizerOptions'Level
+OptimizerOptions'L0 :: OptimizerOptions'Level
+OptimizerOptions'L1 :: OptimizerOptions'Level
+data RunMetadata
+RunMetadata :: Maybe StepStats -> Maybe CostGraphDef -> [GraphDef] -> RunMetadata
+[_RunMetadata'stepStats] :: RunMetadata -> Maybe StepStats
+[_RunMetadata'costGraph] :: RunMetadata -> Maybe CostGraphDef
+[_RunMetadata'partitionGraphs] :: RunMetadata -> [GraphDef]
+data RunOptions
+RunOptions :: RunOptions'TraceLevel -> Int64 -> Int32 -> [DebugTensorWatch] -> Bool -> RunOptions
+[_RunOptions'traceLevel] :: RunOptions -> RunOptions'TraceLevel
+[_RunOptions'timeoutInMs] :: RunOptions -> Int64
+[_RunOptions'interOpThreadPool] :: RunOptions -> Int32
+[_RunOptions'debugTensorWatchOpts] :: RunOptions -> [DebugTensorWatch]
+[_RunOptions'outputPartitionGraphs] :: RunOptions -> Bool
+data RunOptions'TraceLevel
+RunOptions'NO_TRACE :: RunOptions'TraceLevel
+RunOptions'SOFTWARE_TRACE :: RunOptions'TraceLevel
+RunOptions'HARDWARE_TRACE :: RunOptions'TraceLevel
+RunOptions'FULL_TRACE :: RunOptions'TraceLevel
+data ThreadPoolOptionProto
+ThreadPoolOptionProto :: Int32 -> ThreadPoolOptionProto
+[_ThreadPoolOptionProto'numThreads] :: ThreadPoolOptionProto -> Int32
+allocatorType :: HasField "allocatorType" msg msg' => Lens msg msg' (Field "allocatorType" msg) (Field "allocatorType" msg')
+allowGrowth :: HasField "allowGrowth" msg msg' => Lens msg msg' (Field "allowGrowth" msg) (Field "allowGrowth" msg')
+allowSoftPlacement :: HasField "allowSoftPlacement" msg msg' => Lens msg msg' (Field "allowSoftPlacement" msg) (Field "allowSoftPlacement" msg')
+buildCostModel :: HasField "buildCostModel" msg msg' => Lens msg msg' (Field "buildCostModel" msg) (Field "buildCostModel" msg')
+buildCostModelAfter :: HasField "buildCostModelAfter" msg msg' => Lens msg msg' (Field "buildCostModelAfter" msg) (Field "buildCostModelAfter" msg')
+costGraph :: HasField "costGraph" msg msg' => Lens msg msg' (Field "costGraph" msg) (Field "costGraph" msg')
+debugOps :: HasField "debugOps" msg msg' => Lens msg msg' (Field "debugOps" msg) (Field "debugOps" msg')
+debugTensorWatchOpts :: HasField "debugTensorWatchOpts" msg msg' => Lens msg msg' (Field "debugTensorWatchOpts" msg) (Field "debugTensorWatchOpts" msg')
+debugUrls :: HasField "debugUrls" msg msg' => Lens msg msg' (Field "debugUrls" msg) (Field "debugUrls" msg')
+deferredDeletionBytes :: HasField "deferredDeletionBytes" msg msg' => Lens msg msg' (Field "deferredDeletionBytes" msg) (Field "deferredDeletionBytes" msg')
+deviceCount :: HasField "deviceCount" msg msg' => Lens msg msg' (Field "deviceCount" msg) (Field "deviceCount" msg')
+deviceFilters :: HasField "deviceFilters" msg msg' => Lens msg msg' (Field "deviceFilters" msg) (Field "deviceFilters" msg')
+doCommonSubexpressionElimination :: HasField "doCommonSubexpressionElimination" msg msg' => Lens msg msg' (Field "doCommonSubexpressionElimination" msg) (Field "doCommonSubexpressionElimination" msg')
+doConstantFolding :: HasField "doConstantFolding" msg msg' => Lens msg msg' (Field "doConstantFolding" msg) (Field "doConstantFolding" msg')
+doFunctionInlining :: HasField "doFunctionInlining" msg msg' => Lens msg msg' (Field "doFunctionInlining" msg) (Field "doFunctionInlining" msg')
+enableBfloat16Sendrecv :: HasField "enableBfloat16Sendrecv" msg msg' => Lens msg msg' (Field "enableBfloat16Sendrecv" msg) (Field "enableBfloat16Sendrecv" msg')
+enableRecvScheduling :: HasField "enableRecvScheduling" msg msg' => Lens msg msg' (Field "enableRecvScheduling" msg) (Field "enableRecvScheduling" msg')
+gpuOptions :: HasField "gpuOptions" msg msg' => Lens msg msg' (Field "gpuOptions" msg) (Field "gpuOptions" msg')
+graphOptions :: HasField "graphOptions" msg msg' => Lens msg msg' (Field "graphOptions" msg) (Field "graphOptions" msg')
+inferShapes :: HasField "inferShapes" msg msg' => Lens msg msg' (Field "inferShapes" msg) (Field "inferShapes" msg')
+interOpParallelismThreads :: HasField "interOpParallelismThreads" msg msg' => Lens msg msg' (Field "interOpParallelismThreads" msg) (Field "interOpParallelismThreads" msg')
+interOpThreadPool :: HasField "interOpThreadPool" msg msg' => Lens msg msg' (Field "interOpThreadPool" msg) (Field "interOpThreadPool" msg')
+intraOpParallelismThreads :: HasField "intraOpParallelismThreads" msg msg' => Lens msg msg' (Field "intraOpParallelismThreads" msg) (Field "intraOpParallelismThreads" msg')
+key :: HasField "key" msg msg' => Lens msg msg' (Field "key" msg) (Field "key" msg')
+logDevicePlacement :: HasField "logDevicePlacement" msg msg' => Lens msg msg' (Field "logDevicePlacement" msg) (Field "logDevicePlacement" msg')
+maybe'costGraph :: HasField "maybe'costGraph" msg msg' => Lens msg msg' (Field "maybe'costGraph" msg) (Field "maybe'costGraph" msg')
+maybe'gpuOptions :: HasField "maybe'gpuOptions" msg msg' => Lens msg msg' (Field "maybe'gpuOptions" msg) (Field "maybe'gpuOptions" msg')
+maybe'graphOptions :: HasField "maybe'graphOptions" msg msg' => Lens msg msg' (Field "maybe'graphOptions" msg) (Field "maybe'graphOptions" msg')
+maybe'optimizerOptions :: HasField "maybe'optimizerOptions" msg msg' => Lens msg msg' (Field "maybe'optimizerOptions" msg) (Field "maybe'optimizerOptions" msg')
+maybe'stepStats :: HasField "maybe'stepStats" msg msg' => Lens msg msg' (Field "maybe'stepStats" msg) (Field "maybe'stepStats" msg')
+nodeName :: HasField "nodeName" msg msg' => Lens msg msg' (Field "nodeName" msg) (Field "nodeName" msg')
+numThreads :: HasField "numThreads" msg msg' => Lens msg msg' (Field "numThreads" msg) (Field "numThreads" msg')
+operationTimeoutInMs :: HasField "operationTimeoutInMs" msg msg' => Lens msg msg' (Field "operationTimeoutInMs" msg) (Field "operationTimeoutInMs" msg')
+optLevel :: HasField "optLevel" msg msg' => Lens msg msg' (Field "optLevel" msg) (Field "optLevel" msg')
+optimizerOptions :: HasField "optimizerOptions" msg msg' => Lens msg msg' (Field "optimizerOptions" msg) (Field "optimizerOptions" msg')
+outputPartitionGraphs :: HasField "outputPartitionGraphs" msg msg' => Lens msg msg' (Field "outputPartitionGraphs" msg) (Field "outputPartitionGraphs" msg')
+outputSlot :: HasField "outputSlot" msg msg' => Lens msg msg' (Field "outputSlot" msg) (Field "outputSlot" msg')
+partitionGraphs :: HasField "partitionGraphs" msg msg' => Lens msg msg' (Field "partitionGraphs" msg) (Field "partitionGraphs" msg')
+perProcessGpuMemoryFraction :: HasField "perProcessGpuMemoryFraction" msg msg' => Lens msg msg' (Field "perProcessGpuMemoryFraction" msg) (Field "perProcessGpuMemoryFraction" msg')
+placePrunedGraph :: HasField "placePrunedGraph" msg msg' => Lens msg msg' (Field "placePrunedGraph" msg) (Field "placePrunedGraph" msg')
+placementPeriod :: HasField "placementPeriod" msg msg' => Lens msg msg' (Field "placementPeriod" msg) (Field "placementPeriod" msg')
+sessionInterOpThreadPool :: HasField "sessionInterOpThreadPool" msg msg' => Lens msg msg' (Field "sessionInterOpThreadPool" msg) (Field "sessionInterOpThreadPool" msg')
+stepStats :: HasField "stepStats" msg msg' => Lens msg msg' (Field "stepStats" msg) (Field "stepStats" msg')
+timelineStep :: HasField "timelineStep" msg msg' => Lens msg msg' (Field "timelineStep" msg) (Field "timelineStep" msg')
+timeoutInMs :: HasField "timeoutInMs" msg msg' => Lens msg msg' (Field "timeoutInMs" msg) (Field "timeoutInMs" msg')
+traceLevel :: HasField "traceLevel" msg msg' => Lens msg msg' (Field "traceLevel" msg) (Field "traceLevel" msg')
+usePerSessionThreads :: HasField "usePerSessionThreads" msg msg' => Lens msg msg' (Field "usePerSessionThreads" msg) (Field "usePerSessionThreads" msg')
+value :: HasField "value" msg msg' => Lens msg msg' (Field "value" msg) (Field "value" msg')
+visibleDeviceList :: HasField "visibleDeviceList" msg msg' => Lens msg msg' (Field "visibleDeviceList" msg) (Field "visibleDeviceList" msg')
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance GHC.Show.Show Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Protobuf.Config.ThreadPoolOptionProto
+instance GHC.Show.Show Proto.Tensorflow.Core.Protobuf.Config.ThreadPoolOptionProto
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Protobuf.Config.RunOptions
+instance GHC.Show.Show Proto.Tensorflow.Core.Protobuf.Config.RunOptions
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Protobuf.Config.RunOptions'TraceLevel
+instance GHC.Show.Show Proto.Tensorflow.Core.Protobuf.Config.RunOptions'TraceLevel
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Protobuf.Config.RunMetadata
+instance GHC.Show.Show Proto.Tensorflow.Core.Protobuf.Config.RunMetadata
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Protobuf.Config.GraphOptions
+instance GHC.Show.Show Proto.Tensorflow.Core.Protobuf.Config.GraphOptions
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions
+instance GHC.Show.Show Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions'Level
+instance GHC.Show.Show Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions'Level
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Protobuf.Config.GPUOptions
+instance GHC.Show.Show Proto.Tensorflow.Core.Protobuf.Config.GPUOptions
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Protobuf.Config.DebugTensorWatch
+instance GHC.Show.Show Proto.Tensorflow.Core.Protobuf.Config.DebugTensorWatch
+instance GHC.Classes.Eq Proto.Tensorflow.Core.Protobuf.Config.ConfigProto'DeviceCountEntry
+instance GHC.Show.Show Proto.Tensorflow.Core.Protobuf.Config.ConfigProto'DeviceCountEntry
+instance Data.ProtoLens.Field.HasField "deviceCount" Proto.Tensorflow.Core.Protobuf.Config.ConfigProto Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance Data.ProtoLens.Field.HasField "intraOpParallelismThreads" Proto.Tensorflow.Core.Protobuf.Config.ConfigProto Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance Data.ProtoLens.Field.HasField "interOpParallelismThreads" Proto.Tensorflow.Core.Protobuf.Config.ConfigProto Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance Data.ProtoLens.Field.HasField "usePerSessionThreads" Proto.Tensorflow.Core.Protobuf.Config.ConfigProto Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance Data.ProtoLens.Field.HasField "sessionInterOpThreadPool" Proto.Tensorflow.Core.Protobuf.Config.ConfigProto Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance Data.ProtoLens.Field.HasField "placementPeriod" Proto.Tensorflow.Core.Protobuf.Config.ConfigProto Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance Data.ProtoLens.Field.HasField "deviceFilters" Proto.Tensorflow.Core.Protobuf.Config.ConfigProto Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance Data.ProtoLens.Field.HasField "gpuOptions" Proto.Tensorflow.Core.Protobuf.Config.ConfigProto Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance Data.ProtoLens.Field.HasField "maybe'gpuOptions" Proto.Tensorflow.Core.Protobuf.Config.ConfigProto Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance Data.ProtoLens.Field.HasField "allowSoftPlacement" Proto.Tensorflow.Core.Protobuf.Config.ConfigProto Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance Data.ProtoLens.Field.HasField "logDevicePlacement" Proto.Tensorflow.Core.Protobuf.Config.ConfigProto Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance Data.ProtoLens.Field.HasField "graphOptions" Proto.Tensorflow.Core.Protobuf.Config.ConfigProto Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance Data.ProtoLens.Field.HasField "maybe'graphOptions" Proto.Tensorflow.Core.Protobuf.Config.ConfigProto Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance Data.ProtoLens.Field.HasField "operationTimeoutInMs" Proto.Tensorflow.Core.Protobuf.Config.ConfigProto Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Protobuf.Config.ConfigProto
+instance Data.ProtoLens.Field.HasField "key" Proto.Tensorflow.Core.Protobuf.Config.ConfigProto'DeviceCountEntry Proto.Tensorflow.Core.Protobuf.Config.ConfigProto'DeviceCountEntry
+instance Data.ProtoLens.Field.HasField "value" Proto.Tensorflow.Core.Protobuf.Config.ConfigProto'DeviceCountEntry Proto.Tensorflow.Core.Protobuf.Config.ConfigProto'DeviceCountEntry
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Protobuf.Config.ConfigProto'DeviceCountEntry
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Protobuf.Config.ConfigProto'DeviceCountEntry
+instance Data.ProtoLens.Field.HasField "nodeName" Proto.Tensorflow.Core.Protobuf.Config.DebugTensorWatch Proto.Tensorflow.Core.Protobuf.Config.DebugTensorWatch
+instance Data.ProtoLens.Field.HasField "outputSlot" Proto.Tensorflow.Core.Protobuf.Config.DebugTensorWatch Proto.Tensorflow.Core.Protobuf.Config.DebugTensorWatch
+instance Data.ProtoLens.Field.HasField "debugOps" Proto.Tensorflow.Core.Protobuf.Config.DebugTensorWatch Proto.Tensorflow.Core.Protobuf.Config.DebugTensorWatch
+instance Data.ProtoLens.Field.HasField "debugUrls" Proto.Tensorflow.Core.Protobuf.Config.DebugTensorWatch Proto.Tensorflow.Core.Protobuf.Config.DebugTensorWatch
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Protobuf.Config.DebugTensorWatch
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Protobuf.Config.DebugTensorWatch
+instance Data.ProtoLens.Field.HasField "perProcessGpuMemoryFraction" Proto.Tensorflow.Core.Protobuf.Config.GPUOptions Proto.Tensorflow.Core.Protobuf.Config.GPUOptions
+instance Data.ProtoLens.Field.HasField "allocatorType" Proto.Tensorflow.Core.Protobuf.Config.GPUOptions Proto.Tensorflow.Core.Protobuf.Config.GPUOptions
+instance Data.ProtoLens.Field.HasField "deferredDeletionBytes" Proto.Tensorflow.Core.Protobuf.Config.GPUOptions Proto.Tensorflow.Core.Protobuf.Config.GPUOptions
+instance Data.ProtoLens.Field.HasField "allowGrowth" Proto.Tensorflow.Core.Protobuf.Config.GPUOptions Proto.Tensorflow.Core.Protobuf.Config.GPUOptions
+instance Data.ProtoLens.Field.HasField "visibleDeviceList" Proto.Tensorflow.Core.Protobuf.Config.GPUOptions Proto.Tensorflow.Core.Protobuf.Config.GPUOptions
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Protobuf.Config.GPUOptions
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Protobuf.Config.GPUOptions
+instance Data.ProtoLens.Field.HasField "enableRecvScheduling" Proto.Tensorflow.Core.Protobuf.Config.GraphOptions Proto.Tensorflow.Core.Protobuf.Config.GraphOptions
+instance Data.ProtoLens.Field.HasField "optimizerOptions" Proto.Tensorflow.Core.Protobuf.Config.GraphOptions Proto.Tensorflow.Core.Protobuf.Config.GraphOptions
+instance Data.ProtoLens.Field.HasField "maybe'optimizerOptions" Proto.Tensorflow.Core.Protobuf.Config.GraphOptions Proto.Tensorflow.Core.Protobuf.Config.GraphOptions
+instance Data.ProtoLens.Field.HasField "buildCostModel" Proto.Tensorflow.Core.Protobuf.Config.GraphOptions Proto.Tensorflow.Core.Protobuf.Config.GraphOptions
+instance Data.ProtoLens.Field.HasField "buildCostModelAfter" Proto.Tensorflow.Core.Protobuf.Config.GraphOptions Proto.Tensorflow.Core.Protobuf.Config.GraphOptions
+instance Data.ProtoLens.Field.HasField "inferShapes" Proto.Tensorflow.Core.Protobuf.Config.GraphOptions Proto.Tensorflow.Core.Protobuf.Config.GraphOptions
+instance Data.ProtoLens.Field.HasField "placePrunedGraph" Proto.Tensorflow.Core.Protobuf.Config.GraphOptions Proto.Tensorflow.Core.Protobuf.Config.GraphOptions
+instance Data.ProtoLens.Field.HasField "enableBfloat16Sendrecv" Proto.Tensorflow.Core.Protobuf.Config.GraphOptions Proto.Tensorflow.Core.Protobuf.Config.GraphOptions
+instance Data.ProtoLens.Field.HasField "timelineStep" Proto.Tensorflow.Core.Protobuf.Config.GraphOptions Proto.Tensorflow.Core.Protobuf.Config.GraphOptions
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Protobuf.Config.GraphOptions
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Protobuf.Config.GraphOptions
+instance Data.ProtoLens.Field.HasField "doCommonSubexpressionElimination" Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions
+instance Data.ProtoLens.Field.HasField "doConstantFolding" Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions
+instance Data.ProtoLens.Field.HasField "doFunctionInlining" Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions
+instance Data.ProtoLens.Field.HasField "optLevel" Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions'Level
+instance Data.ProtoLens.Message.FieldDefault Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions'Level
+instance Data.ProtoLens.Message.MessageEnum Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions'Level
+instance GHC.Enum.Enum Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions'Level
+instance GHC.Enum.Bounded Proto.Tensorflow.Core.Protobuf.Config.OptimizerOptions'Level
+instance Data.ProtoLens.Field.HasField "stepStats" Proto.Tensorflow.Core.Protobuf.Config.RunMetadata Proto.Tensorflow.Core.Protobuf.Config.RunMetadata
+instance Data.ProtoLens.Field.HasField "maybe'stepStats" Proto.Tensorflow.Core.Protobuf.Config.RunMetadata Proto.Tensorflow.Core.Protobuf.Config.RunMetadata
+instance Data.ProtoLens.Field.HasField "costGraph" Proto.Tensorflow.Core.Protobuf.Config.RunMetadata Proto.Tensorflow.Core.Protobuf.Config.RunMetadata
+instance Data.ProtoLens.Field.HasField "maybe'costGraph" Proto.Tensorflow.Core.Protobuf.Config.RunMetadata Proto.Tensorflow.Core.Protobuf.Config.RunMetadata
+instance Data.ProtoLens.Field.HasField "partitionGraphs" Proto.Tensorflow.Core.Protobuf.Config.RunMetadata Proto.Tensorflow.Core.Protobuf.Config.RunMetadata
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Protobuf.Config.RunMetadata
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Protobuf.Config.RunMetadata
+instance Data.ProtoLens.Field.HasField "traceLevel" Proto.Tensorflow.Core.Protobuf.Config.RunOptions Proto.Tensorflow.Core.Protobuf.Config.RunOptions
+instance Data.ProtoLens.Field.HasField "timeoutInMs" Proto.Tensorflow.Core.Protobuf.Config.RunOptions Proto.Tensorflow.Core.Protobuf.Config.RunOptions
+instance Data.ProtoLens.Field.HasField "interOpThreadPool" Proto.Tensorflow.Core.Protobuf.Config.RunOptions Proto.Tensorflow.Core.Protobuf.Config.RunOptions
+instance Data.ProtoLens.Field.HasField "debugTensorWatchOpts" Proto.Tensorflow.Core.Protobuf.Config.RunOptions Proto.Tensorflow.Core.Protobuf.Config.RunOptions
+instance Data.ProtoLens.Field.HasField "outputPartitionGraphs" Proto.Tensorflow.Core.Protobuf.Config.RunOptions Proto.Tensorflow.Core.Protobuf.Config.RunOptions
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Protobuf.Config.RunOptions
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Protobuf.Config.RunOptions
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Protobuf.Config.RunOptions'TraceLevel
+instance Data.ProtoLens.Message.FieldDefault Proto.Tensorflow.Core.Protobuf.Config.RunOptions'TraceLevel
+instance Data.ProtoLens.Message.MessageEnum Proto.Tensorflow.Core.Protobuf.Config.RunOptions'TraceLevel
+instance GHC.Enum.Enum Proto.Tensorflow.Core.Protobuf.Config.RunOptions'TraceLevel
+instance GHC.Enum.Bounded Proto.Tensorflow.Core.Protobuf.Config.RunOptions'TraceLevel
+instance Data.ProtoLens.Field.HasField "numThreads" Proto.Tensorflow.Core.Protobuf.Config.ThreadPoolOptionProto Proto.Tensorflow.Core.Protobuf.Config.ThreadPoolOptionProto
+instance Data.Default.Class.Default Proto.Tensorflow.Core.Protobuf.Config.ThreadPoolOptionProto
+instance Data.ProtoLens.Message.Message Proto.Tensorflow.Core.Protobuf.Config.ThreadPoolOptionProto
diff --git a/docs/haddock/tensorflow-queue-0.1.0.0/TensorFlow-Queue.html b/docs/haddock/tensorflow-queue-0.1.0.0/TensorFlow-Queue.html
new file mode 100644
index 0000000..f8673b4
--- /dev/null
+++ b/docs/haddock/tensorflow-queue-0.1.0.0/TensorFlow-Queue.html
@@ -0,0 +1,9 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Queue</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-Queue.html");};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="src/TensorFlow-Queue.html">Source</a></li><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-queue-0.1.0.0: Basic access to TensorFlow queues.</p></div><div id="content"><div id="module-header"><table class="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><p class="caption">TensorFlow.Queue</p></div><div id="description"><p class="caption">Description</p><div class="doc"><p>Queues in TensorFlow graph. Very limited support for now.</p></div></div><div id="synopsis"><p id="control.syn" class="caption expander" onclick="toggleSection('syn')">Synopsis</p><ul id="section.syn" class="hide" onclick="toggleSection('syn')"><li class="src short"><span class="keyword">data</span> <a href="#t:Queue2">Queue2</a> a b</li><li class="src short"><a href="#v:makeQueue2">makeQueue2</a> :: <span class="keyword">forall</span> a b. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> b) =&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> -&gt; <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a> -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="TensorFlow-Queue.html#t:Queue2">Queue2</a> a b)</li><li class="src short"><a href="#v:enqueue">enqueue</a> :: <span class="keyword">forall</span> a b v1 v2. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> b) =&gt; <a href="TensorFlow-Queue.html#t:Queue2">Queue2</a> a b -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 a -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 b -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></li><li class="src short"><a href="#v:dequeue">dequeue</a> :: <span class="keyword">forall</span> a b. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> b) =&gt; <a href="TensorFlow-Queue.html#t:Queue2">Queue2</a> a b -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> b)</li></ul></div><div id="interface"><h1>Documentation</h1><div class="top"><p class="src"><span class="keyword">data</span> <a name="t:Queue2" class="def">Queue2</a> a b <a href="src/TensorFlow-Queue.html#Queue2" class="link">Source</a></p><div class="doc"><p>A queue carrying tuples. The underlying structure is more
+ versatile and can be made to support arbitrary tuples.</p></div></div><div class="top"><p class="src"><a name="v:makeQueue2" class="def">makeQueue2</a> <a href="src/TensorFlow-Queue.html#makeQueue2" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> b)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><td class="doc"><p>The upper bound on the number of elements in
+  this queue. Negative numbers mean no limit.</p></td></tr><tr><td class="src">-&gt; <a href="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><td class="doc"><p>If non-empty, this queue will be shared
+ under the given name across multiple sessions.</p></td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="TensorFlow-Queue.html#t:Queue2">Queue2</a> a b)</td><td class="doc empty">&nbsp;</td></tr></table></div><div class="doc"><p>Creates a new queue with the given capacity and shared name.</p></div></div><div class="top"><p class="src"><a name="v:enqueue" class="def">enqueue</a> :: <span class="keyword">forall</span> a b v1 v2. (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> b) =&gt; <a href="TensorFlow-Queue.html#t:Queue2">Queue2</a> a b -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 a -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 b -&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a> <a href="src/TensorFlow-Queue.html#enqueue" class="link">Source</a></p><div class="doc"><p>Adds the given values to the queue.</p></div></div><div class="top"><p class="src"><a name="v:dequeue" class="def">dequeue</a> <a href="src/TensorFlow-Queue.html#dequeue" class="link">Source</a></p><div class="subs arguments"><p class="caption">Arguments</p><table><tr><td class="src">:: (<a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a, <a href="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> b)</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">=&gt; <a href="TensorFlow-Queue.html#t:Queue2">Queue2</a> a b</td><td class="doc empty">&nbsp;</td></tr><tr><td class="src">-&gt; <a href="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a, <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> <a href="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> b)</td><td class="doc"><p>Dequeued tensors. They are paired in a sense
+ that values appear together, even if they are
+ not consumed together.</p></td></tr></table></div><div class="doc"><p>Retrieves the values from the queue.</p></div></div></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-queue-0.1.0.0/doc-index.html b/docs/haddock/tensorflow-queue-0.1.0.0/doc-index.html
new file mode 100644
index 0000000..f9ac0fb
--- /dev/null
+++ b/docs/haddock/tensorflow-queue-0.1.0.0/doc-index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-queue-0.1.0.0: Basic access to TensorFlow queues. (Index)</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-queue-0.1.0.0: Basic access to TensorFlow queues.</p></div><div id="content"><div id="index"><p class="caption">Index</p><table><tr><td class="src">dequeue</td><td class="module"><a href="TensorFlow-Queue.html#v:dequeue">TensorFlow.Queue</a></td></tr><tr><td class="src">enqueue</td><td class="module"><a href="TensorFlow-Queue.html#v:enqueue">TensorFlow.Queue</a></td></tr><tr><td class="src">makeQueue2</td><td class="module"><a href="TensorFlow-Queue.html#v:makeQueue2">TensorFlow.Queue</a></td></tr><tr><td class="src">Queue2</td><td class="module"><a href="TensorFlow-Queue.html#t:Queue2">TensorFlow.Queue</a></td></tr></table></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-queue-0.1.0.0/frames.html b/docs/haddock/tensorflow-queue-0.1.0.0/frames.html
new file mode 100644
index 0000000..1b4e38d
--- /dev/null
+++ b/docs/haddock/tensorflow-queue-0.1.0.0/frames.html
@@ -0,0 +1,30 @@
+<!DOCTYPE html 
+     PUBLIC "-//W3C//DTD XHTML 1.0 Frameset//EN"
+     "http://www.w3.org/TR/xhtml1/DTD/xhtml1-frameset.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<title></title>
+<script src="haddock-util.js" type="text/javascript"></script>
+<script type="text/javascript"><!--
+/*
+
+  The synopsis frame needs to be updated using javascript, so we hide
+  it by default and only show it if javascript is enabled.
+
+  TODO: provide some means to disable it.
+*/
+function load() {
+  var d = document.getElementById("inner-fs");
+  d.rows = "50%,50%";
+  postReframe();
+}
+--></script>
+</head>
+<frameset id="outer-fs" cols="25%,75%" onload="load()">
+  <frameset id="inner-fs" rows="100%,0%">
+    <frame src="index-frames.html" name="modules" />
+    <frame src="" name="synopsis" />
+  </frameset>
+  <frame src="index.html" name="main" />
+</frameset>
+</html>
diff --git a/docs/haddock/tensorflow-queue-0.1.0.0/haddock-util.js b/docs/haddock/tensorflow-queue-0.1.0.0/haddock-util.js
new file mode 100644
index 0000000..9a6fccf
--- /dev/null
+++ b/docs/haddock/tensorflow-queue-0.1.0.0/haddock-util.js
@@ -0,0 +1,344 @@
+// Haddock JavaScript utilities
+
+var rspace = /\s\s+/g,
+	  rtrim = /^\s+|\s+$/g;
+
+function spaced(s) { return (" " + s + " ").replace(rspace, " "); }
+function trim(s)   { return s.replace(rtrim, ""); }
+
+function hasClass(elem, value) {
+  var className = spaced(elem.className || "");
+  return className.indexOf( " " + value + " " ) >= 0;
+}
+
+function addClass(elem, value) {
+  var className = spaced(elem.className || "");
+  if ( className.indexOf( " " + value + " " ) < 0 ) {
+    elem.className = trim(className + " " + value);
+  }
+}
+
+function removeClass(elem, value) {
+  var className = spaced(elem.className || "");
+  className = className.replace(" " + value + " ", " ");
+  elem.className = trim(className);
+}
+
+function toggleClass(elem, valueOn, valueOff, bool) {
+  if (bool == null) { bool = ! hasClass(elem, valueOn); }
+  if (bool) {
+    removeClass(elem, valueOff);
+    addClass(elem, valueOn);
+  }
+  else {
+    removeClass(elem, valueOn);
+    addClass(elem, valueOff);
+  }
+  return bool;
+}
+
+
+function makeClassToggle(valueOn, valueOff)
+{
+  return function(elem, bool) {
+    return toggleClass(elem, valueOn, valueOff, bool);
+  }
+}
+
+toggleShow = makeClassToggle("show", "hide");
+toggleCollapser = makeClassToggle("collapser", "expander");
+
+function toggleSection(id)
+{
+  var b = toggleShow(document.getElementById("section." + id));
+  toggleCollapser(document.getElementById("control." + id), b);
+  rememberCollapsed(id, b);
+  return b;
+}
+
+var collapsed = {};
+function rememberCollapsed(id, b)
+{
+  if(b)
+    delete collapsed[id]
+  else
+    collapsed[id] = null;
+
+  var sections = [];
+  for(var i in collapsed)
+  {
+    if(collapsed.hasOwnProperty(i))
+      sections.push(i);
+  }
+  // cookie specific to this page; don't use setCookie which sets path=/
+  document.cookie = "collapsed=" + escape(sections.join('+'));
+}
+
+function restoreCollapsed()
+{
+  var cookie = getCookie("collapsed");
+  if(!cookie)
+    return;
+
+  var ids = cookie.split('+');
+  for(var i in ids)
+  {
+    if(document.getElementById("section." + ids[i]))
+      toggleSection(ids[i]);
+  }
+}
+
+function setCookie(name, value) {
+  document.cookie = name + "=" + escape(value) + ";path=/;";
+}
+
+function clearCookie(name) {
+  document.cookie = name + "=;path=/;expires=Thu, 01-Jan-1970 00:00:01 GMT;";
+}
+
+function getCookie(name) {
+  var nameEQ = name + "=";
+  var ca = document.cookie.split(';');
+  for(var i=0;i < ca.length;i++) {
+    var c = ca[i];
+    while (c.charAt(0)==' ') c = c.substring(1,c.length);
+    if (c.indexOf(nameEQ) == 0) {
+      return unescape(c.substring(nameEQ.length,c.length));
+    }
+  }
+  return null;
+}
+
+
+
+var max_results = 75; // 50 is not enough to search for map in the base libraries
+var shown_range = null;
+var last_search = null;
+
+function quick_search()
+{
+    perform_search(false);
+}
+
+function full_search()
+{
+    perform_search(true);
+}
+
+
+function perform_search(full)
+{
+    var text = document.getElementById("searchbox").value.toLowerCase();
+    if (text == last_search && !full) return;
+    last_search = text;
+    
+    var table = document.getElementById("indexlist");
+    var status = document.getElementById("searchmsg");
+    var children = table.firstChild.childNodes;
+    
+    // first figure out the first node with the prefix
+    var first = bisect(-1);
+    var last = (first == -1 ? -1 : bisect(1));
+
+    if (first == -1)
+    {
+        table.className = "";
+        status.innerHTML = "No results found, displaying all";
+    }
+    else if (first == 0 && last == children.length - 1)
+    {
+        table.className = "";
+        status.innerHTML = "";
+    }
+    else if (last - first >= max_results && !full)
+    {
+        table.className = "";
+        status.innerHTML = "More than " + max_results + ", press Search to display";
+    }
+    else
+    {
+        // decide what you need to clear/show
+        if (shown_range)
+            setclass(shown_range[0], shown_range[1], "indexrow");
+        setclass(first, last, "indexshow");
+        shown_range = [first, last];
+        table.className = "indexsearch";
+        status.innerHTML = "";
+    }
+
+    
+    function setclass(first, last, status)
+    {
+        for (var i = first; i <= last; i++)
+        {
+            children[i].className = status;
+        }
+    }
+    
+    
+    // do a binary search, treating 0 as ...
+    // return either -1 (no 0's found) or location of most far match
+    function bisect(dir)
+    {
+        var first = 0, finish = children.length - 1;
+        var mid, success = false;
+
+        while (finish - first > 3)
+        {
+            mid = Math.floor((finish + first) / 2);
+
+            var i = checkitem(mid);
+            if (i == 0) i = dir;
+            if (i == -1)
+                finish = mid;
+            else
+                first = mid;
+        }
+        var a = (dir == 1 ? first : finish);
+        var b = (dir == 1 ? finish : first);
+        for (var i = b; i != a - dir; i -= dir)
+        {
+            if (checkitem(i) == 0) return i;
+        }
+        return -1;
+    }    
+    
+    
+    // from an index, decide what the result is
+    // 0 = match, -1 is lower, 1 is higher
+    function checkitem(i)
+    {
+        var s = getitem(i).toLowerCase().substr(0, text.length);
+        if (s == text) return 0;
+        else return (s > text ? -1 : 1);
+    }
+    
+    
+    // from an index, get its string
+    // this abstracts over alternates
+    function getitem(i)
+    {
+        for ( ; i >= 0; i--)
+        {
+            var s = children[i].firstChild.firstChild.data;
+            if (s.indexOf(' ') == -1)
+                return s;
+        }
+        return ""; // should never be reached
+    }
+}
+
+function setSynopsis(filename) {
+    if (parent.window.synopsis) {
+        if (parent.window.synopsis.location.replace) {
+            // In Firefox this avoids adding the change to the history.
+            parent.window.synopsis.location.replace(filename);
+        } else {
+            parent.window.synopsis.location = filename;
+        }
+    }
+}
+
+function addMenuItem(html) {
+  var menu = document.getElementById("page-menu");
+  if (menu) {
+    var btn = menu.firstChild.cloneNode(false);
+    btn.innerHTML = html;
+    menu.appendChild(btn);
+  }
+}
+
+function adjustForFrames() {
+  var bodyCls;
+  
+  if (parent.location.href == window.location.href) {
+    // not in frames, so add Frames button
+    addMenuItem("<a href='#' onclick='reframe();return true;'>Frames</a>");
+    bodyCls = "no-frame";
+  }
+  else {
+    bodyCls = "in-frame";
+  }
+  addClass(document.body, bodyCls);
+}
+
+function reframe() {
+  setCookie("haddock-reframe", document.URL);
+  window.location = "frames.html";
+}
+
+function postReframe() {
+  var s = getCookie("haddock-reframe");
+  if (s) {
+    parent.window.main.location = s;
+    clearCookie("haddock-reframe");
+  }
+}
+
+function styles() {
+  var i, a, es = document.getElementsByTagName("link"), rs = [];
+  for (i = 0; a = es[i]; i++) {
+    if(a.rel.indexOf("style") != -1 && a.title) {
+      rs.push(a);
+    }
+  }
+  return rs;
+}
+
+function addStyleMenu() {
+  var as = styles();
+  var i, a, btns = "";
+  for(i=0; a = as[i]; i++) {
+    btns += "<li><a href='#' onclick=\"setActiveStyleSheet('"
+      + a.title + "'); return false;\">"
+      + a.title + "</a></li>"
+  }
+  if (as.length > 1) {
+    var h = "<div id='style-menu-holder'>"
+      + "<a href='#' onclick='styleMenu(); return false;'>Style &#9662;</a>"
+      + "<ul id='style-menu' class='hide'>" + btns + "</ul>"
+      + "</div>";
+    addMenuItem(h);
+  }
+}
+
+function setActiveStyleSheet(title) {
+  var as = styles();
+  var i, a, found;
+  for(i=0; a = as[i]; i++) {
+    a.disabled = true;
+          // need to do this always, some browsers are edge triggered
+    if(a.title == title) {
+      found = a;
+    }
+  }
+  if (found) {
+    found.disabled = false;
+    setCookie("haddock-style", title);
+  }
+  else {
+    as[0].disabled = false;
+    clearCookie("haddock-style");
+  }
+  styleMenu(false);
+}
+
+function resetStyle() {
+  var s = getCookie("haddock-style");
+  if (s) setActiveStyleSheet(s);
+}
+
+
+function styleMenu(show) {
+  var m = document.getElementById('style-menu');
+  if (m) toggleShow(m, show);
+}
+
+
+function pageLoad() {
+  addStyleMenu();
+  adjustForFrames();
+  resetStyle();
+  restoreCollapsed();
+}
+
diff --git a/docs/haddock/tensorflow-queue-0.1.0.0/hslogo-16.png b/docs/haddock/tensorflow-queue-0.1.0.0/hslogo-16.png
new file mode 100644
index 0000000000000000000000000000000000000000..0ff8579fbd897417b0d6dad6e920f8882138a7c0
GIT binary patch
literal 1684
zcmV;F25b3=P)<h;3K|Lk000e1NJLTq000*N000mO1^@s6{V(Jg0008+X+uL$Nkc;*
zP;zf(X>4Tx0C)j~RL^S@K@|QrZmG~B2wH0nvUrdpNm;9CMbtL^5n^i$+aI<?kYqDS
z)^0ZI?k2URh~(nYD&9ryp$Gqf9(wR(FG_`4MJV2@C}N=(E%D86Otz)!z|Ma2-Z$@k
zZ+71R4RX>n^?(HA4aZWV5ov6ELTdbo0FI&wK{O>*+w4vx20?>!`FrQsdJlnHR>OPy
zcd~b_n$otK2Za4V;76L-DzNVtaSB-y0*E}{p()372;bw_^6ZZ}PI-92wGS&j#91PI
zKs7DSe@(bk%_Y-7gGe}(^>I=@oY#w#*Bu9GZf3^F5WP>3rn}7Ut74&?PWBFvy`A)a
zPP5)V!Xd&78LdA?xQ(9mjMYElVd13a#D+Z_7&Y|xU=_C-srWU*6kiZcC!$nw*)9$7
zn6CX+@=AhmkT}X@VSsa5NKe;HZuq)~1$`#h6R+ZTR#D-3j}vF!)ZOnz+5)dI4jl{{
z44Mr{P!L4~VVJN`K!!XTF*LGrKO?IK8<Tr7btG!LbYeuYL3=jbJ-1P$-8}v%B5{;M
zwFr{@LH;VQ$xr2Z`O93e*jD$Ht(%&<^58qg<(at}9@o>z<8w`3e3jI8lUGNUta*C8
zn(P`s>{pjD=7Kek#B;Fw@hxAK%$F&Q6vg9J^Xf~4by_hu-=A!MJ3Znq&n~srbFGPs
zH&&aMXZ>nO`|hf|ljc?VPhR!${AbO?W8x_>CU%PFA&Hm8F7cAsOREdwU~R_;ot1_u
z(ruCYB-LPGn!NQdT|ZlRy+(fw^-+`=%+gee_kY4FWHg<*4sZI8+sFJD<oAl_pC|$^
zY~aY5x@}W&?+~G7rEYVs0vEs0eekw!YomTR`~+A$s}`+NHJ>270UUORdLHO0nA4V)
z%{fwsET5CQ>B?eK%uw4yQc~9?*JVo<vzb+5>2}ze(;aRcp*ceL#HUJSllrgm5wQKR
zQu+C;QrUh^8rFfA`ftFz{YAidi-`aL010qNS#tmY4c7nw4c7reD4Tcy00T@(L_t(I
z5sj2vNEA^R$7gqDc6T=2^@fUA2(c`MltuL5<|<Z1YDGjBEr^N`3VJpQ$`C4&9(*Z6
z=|*B$y;-1^F6vXRAd7+(;)77Qps2euv)AeWmj>KW>RWz$&YbU@|M|{$E*8Tu-Ux!w
z1Y*Dr&Ubfr&v-nZaaB{3ilRumrjPmk{sZvQEWlW+{o~IH|8)=s6c#X9S5s5d%<Agu
zgM$153Ttk++ZPA~ZqgqN1|^pNV-|?9-zo48n9+Yiy$*su5czgp*VFj6!1h&DRZR>J
z4@)&QH5|xQY-)^L1n<EM!01(Q6kM8}ojndgRD~2KVIa@>0pTRu0Lx9`08YTjTwn^6
z0;b1+aQ@)n;Em$q;=7BBi)v0zj&o^g>0Whp^_^5IbxIUP8C@y9;R?*Ouu}rmfxbU=
zwtWVNke-m!=`7bYEhWpcI5#)9qp`8E0lr6IQ)ARL3Ui}Af@grj8aN1=r>Cb+prlzO
zNfJs*N_tUm2ZL<f%*@E?H`Ak_J`Q#Qri<;K{{H@Bg@uKWS5{WE+}vCx91geD*4DNm
zJ?D^~dGz7lWtbWH`o<0O#|{R6h9M*GMpjnV?Y_RgatOb~1le#ikPKTNIn%<;=>%5*
zPmL2??da$TR904gL(VDAQ<BE%k@5ZA>-Fv_Dk}Pdw*4T(%*f4MKLRg=4ekMjhe2mW
zMFsBwg%ftWT}0kxRaIk1k7qJ8*#cKB;Ft{i`zVIs-Nqge;!!Ld7#O&Qqu7e0sJmP)
z$MW*>L$vSB&dxp@iA3U9fo)-7!Czlr{|o7<LYh`Fd@X2-b0fIa*Wdy$({Zt@t7`&L
zhoaGF!l5{1P1Qm;<GaSj#=Oa7atf1JZ8n=F9*@uC?b(X48O9kPoH~h;!o2=WM?ZLl
z$Lj;4K~)fAdW@^N%nzT>Hv{1oyg3xsu%gn@(b1>$;SM-ZaQ`HV=V0s;lr%d8bd;xY
zGw<i+<$17ay36gdrqk(Lxc1|{y}dSeM5OIV(FjhBg5#OM<!Z*8(@<7crlVw+B-8R1
zl{3J6w9fYQ^vojauaRgZ=~SE&CK!R3pCe!v3>Nvm3=Iu=tyXIgtJnf@A(2S@M140N
ew{UA~tMxaJq;$xaSSi*30000<MNUMnLSTa0p)&^n

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-queue-0.1.0.0/index-frames.html b/docs/haddock/tensorflow-queue-0.1.0.0/index-frames.html
new file mode 100644
index 0000000..684117d
--- /dev/null
+++ b/docs/haddock/tensorflow-queue-0.1.0.0/index-frames.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-queue-0.1.0.0: Basic access to TensorFlow queues.</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-list"><p class="caption">Modules</p><ul><li class="module"><a href="TensorFlow-Queue.html" target="main">TensorFlow.Queue</a></li></ul></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-queue-0.1.0.0/index.html b/docs/haddock/tensorflow-queue-0.1.0.0/index.html
new file mode 100644
index 0000000..a905134
--- /dev/null
+++ b/docs/haddock/tensorflow-queue-0.1.0.0/index.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>tensorflow-queue-0.1.0.0: Basic access to TensorFlow queues.</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body><div id="package-header"><ul class="links" id="page-menu"><li><a href="index.html">Contents</a></li><li><a href="doc-index.html">Index</a></li></ul><p class="caption">tensorflow-queue-0.1.0.0: Basic access to TensorFlow queues.</p></div><div id="content"><div id="description"><h1>tensorflow-queue-0.1.0.0: Basic access to TensorFlow queues.</h1><div class="doc"><p>Please see README.md</p></div></div><div id="module-list"><p class="caption">Modules</p><ul><li><span id="control.n.1" class="module collapser" onclick="toggleSection('n.1')">TensorFlow</span><ul id="section.n.1" class="show"><li><span class="module"><a href="TensorFlow-Queue.html">TensorFlow.Queue</a></span></li></ul></li></ul></div></div><div id="footer"><p>Produced by <a href="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-queue-0.1.0.0/mini_TensorFlow-Queue.html b/docs/haddock/tensorflow-queue-0.1.0.0/mini_TensorFlow-Queue.html
new file mode 100644
index 0000000..2e16ab1
--- /dev/null
+++ b/docs/haddock/tensorflow-queue-0.1.0.0/mini_TensorFlow-Queue.html
@@ -0,0 +1,4 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>TensorFlow.Queue</title><link href="ocean.css" rel="stylesheet" type="text/css" title="Ocean" /><script src="haddock-util.js" type="text/javascript"></script><script type="text/javascript">//<![CDATA[
+window.onload = function () {pageLoad();};
+//]]>
+</script></head><body id="mini"><div id="module-header"><p class="caption">TensorFlow.Queue</p></div><div id="interface"><div class="top"><p class="src"><span class="keyword">data</span> <a href="TensorFlow-Queue.html#t:Queue2" target="main">Queue2</a> a b</p></div><div class="top"><p class="src"><a href="TensorFlow-Queue.html#v:makeQueue2" target="main">makeQueue2</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Queue.html#v:enqueue" target="main">enqueue</a></p></div><div class="top"><p class="src"><a href="TensorFlow-Queue.html#v:dequeue" target="main">dequeue</a></p></div></div></body></html>
\ No newline at end of file
diff --git a/docs/haddock/tensorflow-queue-0.1.0.0/minus.gif b/docs/haddock/tensorflow-queue-0.1.0.0/minus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..1deac2fe1a42e35b994f1b855488f392c50f6a89
GIT binary patch
literal 56
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS53HJ0a;6C;`j@u!^I(6E|
Jh4LZ{)&P#)4$%Mr

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-queue-0.1.0.0/ocean.css b/docs/haddock/tensorflow-queue-0.1.0.0/ocean.css
new file mode 100644
index 0000000..1110b40
--- /dev/null
+++ b/docs/haddock/tensorflow-queue-0.1.0.0/ocean.css
@@ -0,0 +1,600 @@
+/* @group Fundamentals */
+
+* { margin: 0; padding: 0 }
+
+/* Is this portable? */
+html {
+  background-color: white;
+  width: 100%;
+  height: 100%;
+}
+
+body {
+  background: white;
+  color: black;
+  text-align: left;
+  min-height: 100%;
+  position: relative;
+}
+
+p {
+  margin: 0.8em 0;
+}
+
+ul, ol {
+  margin: 0.8em 0 0.8em 2em;
+}
+
+dl {
+  margin: 0.8em 0;
+}
+
+dt {
+  font-weight: bold;
+}
+dd {
+  margin-left: 2em;
+}
+
+a { text-decoration: none; }
+a[href]:link { color: rgb(196,69,29); }
+a[href]:visited { color: rgb(171,105,84); }
+a[href]:hover { text-decoration:underline; }
+
+/* @end */
+
+/* @group Fonts & Sizes */
+
+/* Basic technique & IE workarounds from YUI 3
+   For reasons, see:
+      http://yui.yahooapis.com/3.1.1/build/cssfonts/fonts.css
+ */
+
+body {
+	font:13px/1.4 sans-serif;
+	*font-size:small; /* for IE */
+	*font:x-small; /* for IE in quirks mode */
+}
+
+h1 { font-size: 146.5%; /* 19pt */ }
+h2 { font-size: 131%;   /* 17pt */ }
+h3 { font-size: 116%;   /* 15pt */ }
+h4 { font-size: 100%;   /* 13pt */ }
+h5 { font-size: 100%;   /* 13pt */ }
+
+select, input, button, textarea {
+	font:99% sans-serif;
+}
+
+table {
+	font-size:inherit;
+	font:100%;
+}
+
+pre, code, kbd, samp, tt, .src {
+	font-family:monospace;
+	*font-size:108%;
+	line-height: 124%;
+}
+
+.links, .link {
+  font-size: 85%; /* 11pt */
+}
+
+#module-header .caption {
+  font-size: 182%; /* 24pt */
+}
+
+.info  {
+  font-size: 85%; /* 11pt */
+}
+
+#table-of-contents, #synopsis  {
+  /* font-size: 85%; /* 11pt */
+}
+
+
+/* @end */
+
+/* @group Common */
+
+.caption, h1, h2, h3, h4, h5, h6 {
+  font-weight: bold;
+  color: rgb(78,98,114);
+  margin: 0.8em 0 0.4em;
+}
+
+* + h1, * + h2, * + h3, * + h4, * + h5, * + h6 {
+  margin-top: 2em;
+}
+
+h1 + h2, h2 + h3, h3 + h4, h4 + h5, h5 + h6 {
+  margin-top: inherit;
+}
+
+ul.links {
+  list-style: none;
+  text-align: left;
+  float: right;
+  display: inline-table;
+  margin: 0 0 0 1em;
+}
+
+ul.links li {
+  display: inline;
+  border-left: 1px solid #d5d5d5;
+  white-space: nowrap;
+  padding: 0;
+}
+
+ul.links li a {
+  padding: 0.2em 0.5em;
+}
+
+.hide { display: none; }
+.show { display: inherit; }
+.clear { clear: both; }
+
+.collapser {
+  background-image: url(minus.gif);
+  background-repeat: no-repeat;
+}
+.expander {
+  background-image: url(plus.gif);
+  background-repeat: no-repeat;
+}
+p.caption.collapser,
+p.caption.expander {
+  background-position: 0 0.4em;
+}
+.collapser, .expander {
+  padding-left: 14px;
+  margin-left: -14px;
+  cursor: pointer;
+}
+
+pre {
+  padding: 0.25em;
+  margin: 0.8em 0;
+  background: rgb(229,237,244);
+  overflow: auto;
+  border-bottom: 0.25em solid white;
+  /* white border adds some space below the box to compensate
+     for visual extra space that paragraphs have between baseline
+     and the bounding box */
+}
+
+.src {
+  background: #f0f0f0;
+  padding: 0.2em 0.5em;
+}
+
+.keyword { font-weight: normal; }
+.def { font-weight: bold; }
+
+
+/* @end */
+
+/* @group Page Structure */
+
+#content {
+  margin: 0 auto;
+  padding: 0 2em 6em;
+}
+
+#package-header {
+  background: rgb(41,56,69);
+  border-top: 5px solid rgb(78,98,114);
+  color: #ddd;
+  padding: 0.2em;
+  position: relative;
+  text-align: left;
+}
+
+#package-header .caption {
+  background: url(hslogo-16.png) no-repeat 0em;
+  color: white;
+  margin: 0 2em;
+  font-weight: normal;
+  font-style: normal;
+  padding-left: 2em;
+}
+
+#package-header a:link, #package-header a:visited { color: white; }
+#package-header a:hover { background: rgb(78,98,114); }
+
+#module-header .caption {
+  color: rgb(78,98,114);
+  font-weight: bold;
+  border-bottom: 1px solid #ddd;
+}
+
+table.info {
+  float: right;
+  padding: 0.5em 1em;
+  border: 1px solid #ddd;
+  color: rgb(78,98,114);
+  background-color: #fff;
+  max-width: 40%;
+  border-spacing: 0;
+  position: relative;
+  top: -0.5em;
+  margin: 0 0 0 2em;
+}
+
+.info th {
+	padding: 0 1em 0 0;
+}
+
+div#style-menu-holder {
+  position: relative;
+  z-index: 2;
+  display: inline;
+}
+
+#style-menu {
+  position: absolute;
+  z-index: 1;
+  overflow: visible;
+  background: #374c5e;
+  margin: 0;
+  text-align: center;
+  right: 0;
+  padding: 0;
+  top: 1.25em;
+}
+
+#style-menu li {
+	display: list-item;
+	border-style: none;
+	margin: 0;
+	padding: 0;
+	color: #000;
+	list-style-type: none;
+}
+
+#style-menu li + li {
+	border-top: 1px solid #919191;
+}
+
+#style-menu a {
+  width: 6em;
+  padding: 3px;
+  display: block;
+}
+
+#footer {
+  background: #ddd;
+  border-top: 1px solid #aaa;
+  padding: 0.5em 0;
+  color: #666;
+  text-align: center;
+  position: absolute;
+  bottom: 0;
+  width: 100%;
+  height: 3em;
+}
+
+/* @end */
+
+/* @group Front Matter */
+
+#table-of-contents {
+  float: right;
+  clear: right;
+  background: #faf9dc;
+  border: 1px solid #d8d7ad;
+  padding: 0.5em 1em;
+  max-width: 20em;
+  margin: 0.5em 0 1em 1em;
+}
+
+#table-of-contents .caption {
+  text-align: center;
+  margin: 0;
+}
+
+#table-of-contents ul {
+  list-style: none;
+  margin: 0;
+}
+
+#table-of-contents ul ul {
+  margin-left: 2em;
+}
+
+#description .caption {
+  display: none;
+}
+
+#synopsis {
+  display: none;
+}
+
+.no-frame #synopsis {
+  display: block;
+  position: fixed;
+  right: 0;
+  height: 80%;
+  top: 10%;
+  padding: 0;
+  max-width: 75%;
+}
+
+#synopsis .caption {
+  float: left;
+  width: 29px;
+  color: rgba(255,255,255,0);
+  height: 110px;
+  margin: 0;
+  font-size: 1px;
+  padding: 0;
+}
+
+#synopsis p.caption.collapser {
+  background: url(synopsis.png) no-repeat -64px -8px;
+}
+
+#synopsis p.caption.expander {
+  background: url(synopsis.png) no-repeat 0px -8px;
+}
+
+#synopsis ul {
+  height: 100%;
+  overflow: auto;
+  padding: 0.5em;
+  margin: 0;
+}
+
+#synopsis ul ul {
+  overflow: hidden;
+}
+
+#synopsis ul,
+#synopsis ul li.src {
+  background-color: #faf9dc;
+  white-space: nowrap;
+  list-style: none;
+  margin-left: 0;
+}
+
+/* @end */
+
+/* @group Main Content */
+
+#interface div.top { margin: 2em 0; }
+#interface h1 + div.top,
+#interface h2 + div.top,
+#interface h3 + div.top,
+#interface h4 + div.top,
+#interface h5 + div.top {
+ 	margin-top: 1em;
+}
+#interface p.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface td.src .link {
+  float: right;
+  color: #919191;
+  border-left: 1px solid #919191;
+  background: #f0f0f0;
+  padding: 0 0.5em 0.2em;
+  margin: 0 -0.5em 0 0.5em;
+}
+
+#interface span.fixity {
+  color: #919191;
+  border-left: 1px solid #919191;
+  padding: 0.2em 0.5em 0.2em 0.5em;
+  margin: 0 -1em 0 1em;
+}
+
+#interface span.rightedge {
+  border-left: 1px solid #919191;
+  padding: 0.2em 0 0.2em 0;
+  margin: 0 0 0 1em;
+}
+
+#interface table { border-spacing: 2px; }
+#interface td {
+  vertical-align: top;
+  padding-left: 0.5em;
+}
+#interface td.src {
+  white-space: nowrap;
+}
+#interface td.doc p {
+  margin: 0;
+}
+#interface td.doc p + p {
+  margin-top: 0.8em;
+}
+
+.clearfix:after {
+  clear: both;
+  content: " ";
+  display: block;
+  height: 0;
+  visibility: hidden;
+}
+
+.subs dl {
+  margin: 0;
+}
+
+.subs dt {
+  float: left;
+  clear: left;
+  display: block;
+  margin: 1px 0;
+}
+
+.subs dd {
+  float: right;
+  width: 90%;
+  display: block;
+  padding-left: 0.5em;
+  margin-bottom: 0.5em;
+}
+
+.subs dd.empty {
+  display: none;
+}
+
+.subs dd p {
+  margin: 0;
+}
+
+/* Render short-style data instances */
+.inst ul {
+  height: 100%;
+  padding: 0.5em;
+  margin: 0;
+}
+
+.inst, .inst li {
+  list-style: none;
+  margin-left: 1em;
+}
+
+/* Workaround for bug in Firefox (issue #384) */
+.inst-left {
+  float: left;
+}
+
+.top p.src {
+  border-top: 1px solid #ccc;
+}
+
+.subs, .doc {
+  /* use this selector for one level of indent */
+  padding-left: 2em;
+}
+
+.warning {
+  color: red;
+}
+
+.arguments {
+  margin-top: -0.4em;
+}
+.arguments .caption {
+  display: none;
+}
+
+.fields { padding-left: 1em; }
+
+.fields .caption { display: none; }
+
+.fields p { margin: 0 0; }
+
+/* this seems bulky to me
+.methods, .constructors {
+  background: #f8f8f8;
+  border: 1px solid #eee;
+}
+*/
+
+/* @end */
+
+/* @group Auxillary Pages */
+
+
+.extension-list {
+    list-style-type: none;
+    margin-left: 0;
+}
+
+#mini {
+  margin: 0 auto;
+  padding: 0 1em 1em;
+}
+
+#mini > * {
+  font-size: 93%; /* 12pt */
+}
+
+#mini #module-list .caption,
+#mini #module-header .caption {
+  font-size: 125%; /* 15pt */
+}
+
+#mini #interface h1,
+#mini #interface h2,
+#mini #interface h3,
+#mini #interface h4 {
+  font-size: 109%; /* 13pt */
+  margin: 1em 0 0;
+}
+
+#mini #interface .top,
+#mini #interface .src {
+  margin: 0;
+}
+
+#mini #module-list ul {
+  list-style: none;
+  margin: 0;
+}
+
+#alphabet ul {
+	list-style: none;
+	padding: 0;
+	margin: 0.5em 0 0;
+	text-align: center;
+}
+
+#alphabet li {
+	display: inline;
+	margin: 0 0.25em;
+}
+
+#alphabet a {
+	font-weight: bold;
+}
+
+#index .caption,
+#module-list .caption { font-size: 131%; /* 17pt */ }
+
+#index table {
+  margin-left: 2em;
+}
+
+#index .src {
+  font-weight: bold;
+}
+#index .alt {
+  font-size: 77%; /* 10pt */
+  font-style: italic;
+  padding-left: 2em;
+}
+
+#index td + td {
+  padding-left: 1em;
+}
+
+#module-list ul {
+  list-style: none;
+  margin: 0 0 0 2em;
+}
+
+#module-list li {
+  clear: right;
+}
+
+#module-list span.collapser,
+#module-list span.expander {
+  background-position: 0 0.3em;
+}
+
+#module-list .package {
+  float: right;
+}
+
+/* @end */
diff --git a/docs/haddock/tensorflow-queue-0.1.0.0/plus.gif b/docs/haddock/tensorflow-queue-0.1.0.0/plus.gif
new file mode 100644
index 0000000000000000000000000000000000000000..2d15c14173d23f664b955cd24f51c82f5f09d91d
GIT binary patch
literal 59
zcmZ?wbhEHb<YeGvn8?8J@87?srY0sPCI$uu9UumZaxyS5iS+a@;NG@rPSdH*>gbBX
M^XE!9f*2UA0nx1yDgXcg

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-queue-0.1.0.0/src/TensorFlow-Queue.html b/docs/haddock/tensorflow-queue-0.1.0.0/src/TensorFlow-Queue.html
new file mode 100644
index 0000000..df19140
--- /dev/null
+++ b/docs/haddock/tensorflow-queue-0.1.0.0/src/TensorFlow-Queue.html
@@ -0,0 +1,89 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+<html>
+<head>
+<!-- Generated by HsColour, http://code.haskell.org/~malcolm/hscolour/ -->
+<title>src/TensorFlow/Queue.hs</title>
+<link type='text/css' rel='stylesheet' href='hscolour.css' />
+</head>
+<body>
+<pre><a name="line-1"></a><span class='hs-comment'>-- Copyright 2016 TensorFlow authors.</span>
+<a name="line-2"></a><span class='hs-comment'>--</span>
+<a name="line-3"></a><span class='hs-comment'>-- Licensed under the Apache License, Version 2.0 (the "License");</span>
+<a name="line-4"></a><span class='hs-comment'>-- you may not use this file except in compliance with the License.</span>
+<a name="line-5"></a><span class='hs-comment'>-- You may obtain a copy of the License at</span>
+<a name="line-6"></a><span class='hs-comment'>--</span>
+<a name="line-7"></a><span class='hs-comment'>--     <a href="http://www.apache.org/licenses/LICENSE-2.0">http://www.apache.org/licenses/LICENSE-2.0</a></span>
+<a name="line-8"></a><span class='hs-comment'>--</span>
+<a name="line-9"></a><span class='hs-comment'>-- Unless required by applicable law or agreed to in writing, software</span>
+<a name="line-10"></a><span class='hs-comment'>-- distributed under the License is distributed on an "AS IS" BASIS,</span>
+<a name="line-11"></a><span class='hs-comment'>-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
+<a name="line-12"></a><span class='hs-comment'>-- See the License for the specific language governing permissions and</span>
+<a name="line-13"></a><span class='hs-comment'>-- limitations under the License.</span>
+<a name="line-14"></a>
+<a name="line-15"></a><span class='hs-comment'>{-# LANGUAGE OverloadedStrings #-}</span>
+<a name="line-16"></a><span class='hs-comment'>{-# LANGUAGE ScopedTypeVariables #-}</span>
+<a name="line-17"></a>
+<a name="line-18"></a><span class='hs-comment'>-- | Queues in TensorFlow graph. Very limited support for now.</span>
+<a name="line-19"></a><span class='hs-keyword'>module</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Queue</span> <span class='hs-layout'>(</span><span class='hs-conid'>Queue2</span><span class='hs-layout'>,</span> <span class='hs-varid'>makeQueue2</span><span class='hs-layout'>,</span> <span class='hs-varid'>enqueue</span><span class='hs-layout'>,</span> <span class='hs-varid'>dequeue</span><span class='hs-layout'>)</span> <span class='hs-keyword'>where</span>
+<a name="line-20"></a>
+<a name="line-21"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>ByteString</span> <span class='hs-layout'>(</span><span class='hs-conid'>ByteString</span><span class='hs-layout'>)</span>
+<a name="line-22"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Data</span><span class='hs-varop'>.</span><span class='hs-conid'>Int</span> <span class='hs-layout'>(</span><span class='hs-conid'>Int64</span><span class='hs-layout'>)</span>
+<a name="line-23"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>Lens</span><span class='hs-varop'>.</span><span class='hs-conid'>Family2</span> <span class='hs-layout'>(</span><span class='hs-layout'>(</span><span class='hs-varop'>.~</span><span class='hs-layout'>)</span><span class='hs-layout'>,</span> <span class='hs-layout'>(</span><span class='hs-varop'>&amp;</span><span class='hs-layout'>)</span><span class='hs-layout'>)</span>
+<a name="line-24"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>ControlNode</span><span class='hs-layout'>,</span> <span class='hs-conid'>Build</span><span class='hs-layout'>,</span> <span class='hs-varid'>addInitializer</span><span class='hs-layout'>,</span> <span class='hs-varid'>opAttr</span><span class='hs-layout'>,</span> <span class='hs-varid'>opDef</span><span class='hs-layout'>)</span>
+<a name="line-25"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>BuildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>buildOp</span><span class='hs-layout'>)</span>
+<a name="line-26"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>ControlFlow</span> <span class='hs-layout'>(</span><span class='hs-varid'>group</span><span class='hs-layout'>)</span>
+<a name="line-27"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Tensor</span> <span class='hs-layout'>(</span><span class='hs-conid'>Ref</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span><span class='hs-layout'>)</span>
+<a name="line-28"></a><span class='hs-keyword'>import</span> <span class='hs-conid'>TensorFlow</span><span class='hs-varop'>.</span><span class='hs-conid'>Types</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span><span class='hs-layout'>,</span> <span class='hs-varid'>tensorType</span><span class='hs-layout'>)</span>
+<a name="line-29"></a>
+<a name="line-30"></a><a name="Queue2"></a><span class='hs-comment'>-- | A queue carrying tuples. The underlying structure is more</span>
+<a name="line-31"></a><a name="Queue2"></a><span class='hs-comment'>-- versatile and can be made to support arbitrary tuples.</span>
+<a name="line-32"></a><a name="Queue2"></a><span class='hs-keyword'>data</span> <span class='hs-conid'>Queue2</span> <span class='hs-varid'>a</span> <span class='hs-varid'>b</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Queue2</span> <span class='hs-layout'>{</span> <span class='hs-varid'>handle</span> <span class='hs-keyglyph'>::</span> <span class='hs-conid'>Handle</span> <span class='hs-layout'>}</span>
+<a name="line-33"></a>
+<a name="line-34"></a><a name="Handle"></a><span class='hs-keyword'>type</span> <span class='hs-conid'>Handle</span> <span class='hs-keyglyph'>=</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Ref</span> <span class='hs-conid'>ByteString</span>
+<a name="line-35"></a>
+<a name="line-36"></a><a name="enqueue"></a><span class='hs-comment'>-- | Adds the given values to the queue.</span>
+<a name="line-37"></a><span class='hs-definition'>enqueue</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varid'>b</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>v2</span><span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>b</span><span class='hs-layout'>)</span>
+<a name="line-38"></a>           <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Queue2</span> <span class='hs-varid'>a</span> <span class='hs-varid'>b</span>
+<a name="line-39"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v1</span> <span class='hs-varid'>a</span>
+<a name="line-40"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Tensor</span> <span class='hs-varid'>v2</span> <span class='hs-varid'>b</span>
+<a name="line-41"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-conid'>ControlNode</span>
+<a name="line-42"></a><span class='hs-definition'>enqueue</span> <span class='hs-varid'>q</span> <span class='hs-keyglyph'>=</span>
+<a name="line-43"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"QueueEnqueue"</span>
+<a name="line-44"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"Tcomponents"</span> <span class='hs-varop'>.~</span> <span class='hs-keyglyph'>[</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-45"></a>                                       <span class='hs-layout'>,</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>b</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-46"></a>    <span class='hs-layout'>(</span><span class='hs-varid'>handle</span> <span class='hs-varid'>q</span><span class='hs-layout'>)</span>
+<a name="line-47"></a>
+<a name="line-48"></a><a name="dequeue"></a><span class='hs-comment'>-- | Retrieves the values from the queue.</span>
+<a name="line-49"></a><span class='hs-definition'>dequeue</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varid'>b</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>b</span><span class='hs-layout'>)</span>
+<a name="line-50"></a>           <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Queue2</span> <span class='hs-varid'>a</span> <span class='hs-varid'>b</span>
+<a name="line-51"></a>           <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>Tensor</span> <span class='hs-conid'>Ref</span> <span class='hs-varid'>a</span><span class='hs-layout'>,</span> <span class='hs-conid'>Tensor</span> <span class='hs-conid'>Ref</span> <span class='hs-varid'>b</span><span class='hs-layout'>)</span>
+<a name="line-52"></a>           <span class='hs-comment'>-- ^ Dequeued tensors. They are paired in a sense</span>
+<a name="line-53"></a>           <span class='hs-comment'>-- that values appear together, even if they are</span>
+<a name="line-54"></a>           <span class='hs-comment'>-- not consumed together.</span>
+<a name="line-55"></a><span class='hs-definition'>dequeue</span> <span class='hs-varid'>q</span> <span class='hs-keyglyph'>=</span>
+<a name="line-56"></a>    <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"QueueDequeue"</span>
+<a name="line-57"></a>             <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"component_types"</span> <span class='hs-varop'>.~</span> <span class='hs-keyglyph'>[</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-58"></a>                                           <span class='hs-layout'>,</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>b</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span><span class='hs-layout'>)</span>
+<a name="line-59"></a>    <span class='hs-layout'>(</span><span class='hs-varid'>handle</span> <span class='hs-varid'>q</span><span class='hs-layout'>)</span>
+<a name="line-60"></a>
+<a name="line-61"></a><a name="makeQueue2"></a><span class='hs-comment'>-- | Creates a new queue with the given capacity and shared name.</span>
+<a name="line-62"></a><span class='hs-definition'>makeQueue2</span> <span class='hs-keyglyph'>::</span> <span class='hs-keyword'>forall</span> <span class='hs-varid'>a</span> <span class='hs-varid'>b</span> <span class='hs-varop'>.</span> <span class='hs-layout'>(</span><span class='hs-conid'>TensorType</span> <span class='hs-varid'>a</span><span class='hs-layout'>,</span> <span class='hs-conid'>TensorType</span> <span class='hs-varid'>b</span><span class='hs-layout'>)</span>
+<a name="line-63"></a>              <span class='hs-keyglyph'>=&gt;</span> <span class='hs-conid'>Int64</span>  <span class='hs-comment'>-- ^ The upper bound on the number of elements in</span>
+<a name="line-64"></a>                        <span class='hs-comment'>--  this queue. Negative numbers mean no limit.</span>
+<a name="line-65"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>ByteString</span> <span class='hs-comment'>-- ^ If non-empty, this queue will be shared</span>
+<a name="line-66"></a>                            <span class='hs-comment'>-- under the given name across multiple sessions.</span>
+<a name="line-67"></a>              <span class='hs-keyglyph'>-&gt;</span> <span class='hs-conid'>Build</span> <span class='hs-layout'>(</span><span class='hs-conid'>Queue2</span> <span class='hs-varid'>a</span> <span class='hs-varid'>b</span><span class='hs-layout'>)</span>
+<a name="line-68"></a><span class='hs-definition'>makeQueue2</span> <span class='hs-varid'>capacity</span> <span class='hs-varid'>sharedName</span> <span class='hs-keyglyph'>=</span> <span class='hs-keyword'>do</span>
+<a name="line-69"></a>    <span class='hs-varid'>q</span> <span class='hs-keyglyph'>&lt;-</span> <span class='hs-varid'>buildOp</span> <span class='hs-layout'>(</span><span class='hs-varid'>opDef</span> <span class='hs-str'>"FIFOQueue"</span>
+<a name="line-70"></a>                     <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"component_types"</span> <span class='hs-varop'>.~</span> <span class='hs-keyglyph'>[</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>a</span><span class='hs-layout'>)</span>
+<a name="line-71"></a>                                                   <span class='hs-layout'>,</span> <span class='hs-varid'>tensorType</span> <span class='hs-layout'>(</span><span class='hs-varid'>undefined</span> <span class='hs-keyglyph'>::</span> <span class='hs-varid'>b</span><span class='hs-layout'>)</span><span class='hs-keyglyph'>]</span>
+<a name="line-72"></a>                     <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"shared_name"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>sharedName</span>
+<a name="line-73"></a>                     <span class='hs-varop'>&amp;</span> <span class='hs-varid'>opAttr</span> <span class='hs-str'>"capacity"</span> <span class='hs-varop'>.~</span> <span class='hs-varid'>capacity</span>
+<a name="line-74"></a>                    <span class='hs-layout'>)</span>
+<a name="line-75"></a>    <span class='hs-varid'>group</span> <span class='hs-varid'>q</span> <span class='hs-varop'>&gt;&gt;=</span> <span class='hs-varid'>addInitializer</span>
+<a name="line-76"></a>    <span class='hs-varid'>return</span> <span class='hs-layout'>(</span><span class='hs-conid'>Queue2</span> <span class='hs-varid'>q</span><span class='hs-layout'>)</span>
+<a name="line-77"></a>
+<a name="line-78"></a><span class='hs-comment'>-- TODO(gnezdo): Figure out the closing story for queues.</span>
+</pre></body>
+</html>
diff --git a/docs/haddock/tensorflow-queue-0.1.0.0/src/hscolour.css b/docs/haddock/tensorflow-queue-0.1.0.0/src/hscolour.css
new file mode 100644
index 0000000..c15919e
--- /dev/null
+++ b/docs/haddock/tensorflow-queue-0.1.0.0/src/hscolour.css
@@ -0,0 +1,5 @@
+.hs-keyglyph, .hs-layout {color: red;}
+.hs-keyword {color: blue;}
+.hs-comment, .hs-comment a {color: green;}
+.hs-str, .hs-chr {color: teal;}
+.hs-keyword, .hs-conid, .hs-varid, .hs-conop, .hs-varop, .hs-num, .hs-cpp, .hs-sel, .hs-definition {}
diff --git a/docs/haddock/tensorflow-queue-0.1.0.0/synopsis.png b/docs/haddock/tensorflow-queue-0.1.0.0/synopsis.png
new file mode 100644
index 0000000000000000000000000000000000000000..85fb86ec84907bcc86531dc82871948ff4d471fa
GIT binary patch
literal 11327
zcmV-FEWp!=P)<h;3K|Lk000e1NJLTq004jh003wR1^@s6IVt!g000+IX+uL$Nkc;*
zP;zf(X>4Tx0C)k_S!GyNTeqHT_l8Y(cXyX`gGi?cY`Qxn1VID|MJXwjPC)?)F$h6K
zMMOd+6hs7sqbPzXbr*U(-*=zy-hcPcUC*=TdiNM(jyd-lv&OpsU|J&v2m2!^0SE{T
z54F(O;<U4O;KWV=44?vZKnREv-NJ%3&CHI1|Ml9M1kj;!OzhCt|7!L>E2!K(!rTCW
z%wV;vdzf1QjBf#e&~gh74F>?Z4a=WLg$KhJ^$5nap>PLbJadS>e&h8+?D`9%QNL`g
zEVKbYGXj7k5Q(8)0Fd#*a?VIMFW3*64geVHKzE-&0BG!BtmfuTbO(T`0Jaeg2nagF
z{V*1E{Wm{e|AvV~*MEExiC+KU-~R=!2{)|c6Bg`GjQ;iG|FQ`1kAUCTuZtQk34#8{
z4r4<P-^=bq9b@>(3g7#|{=Z@d+d#}7f!3C=>=26vx*jwA8>@MS>RG@Tt_zt3hie^T
z_?<zX!Qp0Bf6L=zytR+Pm;lRjJ;HSV_$l=z8Xd-r9G16)Mp#(Gm>0%9VUd=)Fos7I
z^ghPh%Jy%YZ|)vCf6EaFPai$Q-!=$ppK!y&wrJs)bNdAuANB!m3n34Tfj{s75g-&U
z1A!Pg3bcXF-=!Gv1VmU93G2duANT;{0JugFTqg*|oPXPC|A$2HS3NJd-hcPV3EW`Y
zh=1Dr-5Mv{<{zIvz#Yba<V3iQL)~HT!km4Q{?+Cmi$j0EdL!8&xgxpyr%xz;lr%~X
zrG-*OsiBkrCyE}$48%~fC`FVeN*$#F>y&^Vcn^E_`qRfl{{bzYkp)4~$~NAx_VB;E
z{?P)PU)DbV{Qi#~0H0@T9czDj06@6MNq8OrpdAz(9qQxd9nPr<&s+~tPQySqaZyfb
zNh!%g_5YjeaLxMN*$s<dYY{|0@-La=c>v_p;d%b#U$Wpz0Geb0U>E+EOsEQ;I<Re%
zqFY30<Zpkd!v`9W019}7G68nL0|epsEDhv=3eW($zzCQED_{?t;PK-PkHJt71>!&=
zNC6q(BFF<p;2Nj^wV)BSfDUjM^n!=r33v`Bz%-Zxi(nP3gDtQFK@b)qgJ>Wohy&t-
zL?CHM5mJM6p`(xmWDmJOUQi$u0mVUQpbRJ*DuT+OI;a`C4fR4p&?xj8nuk`Puh35f
z5<x)FBG?f82nmEDLK9((utJ<bcp*X%F^FWu1w;X&9C01dj_5@^K};a#5UYqyBtViP
z>5*JWF{C0=8)=GkKzbrWk@3iMWInPS*@Wyu4kE{pbI3L14-^JPgW^Pq!Q<2bWsPz}
zg`nb5nW!REEvg;Wj~YYGqt;RTXfiY_S_G|(HbmQ@z0gtU6m&ki8r_B-Ku@3-(OVb{
zh8`n;QNS2r>@mKWSWG773g!l;2Q!LUz-(f%SSG9pRuyZCC1S&|DcC~nb!<2G1$Gg;
zjU&Zz;G}VSI0sxHE(w>9tH<5Py}&KucJP#VKD;vC6z`6Y#%JLx@m=^4{33pbgo;Ff
zM3uyf#Fr$Iq=2M}WPoIbWP_BHl$%tE)ST3Z^fYM!=}po{r1PXd2-E~&f;PdC5J9*=
zs3G(aUK2LR$jJD~G{_vt!pSa>)sa0QdqcKOPD3tEZb<G<o<Lqq-cJ64{1XM9f{#Lz
z!ige=BA=p_VuWIi5>LrbsZB|wjHfK7yiNI%a+8XNN{Y&qDu61Js-9|yYMB~K%}=dM
z?M|IcT|xbTdVvN>!$YG@<3@9arjllWW|0;{D?n>V>r0zK+erJ2cAbuzPL|Gw?j&6?
z-95TFdL%tRy&=6neHMKS{UrTQ1~vvw1`mcbh9-s=4Br`97&RC@7}FVVFitT3Wa4Df
zW%6UX#MHqw%Zy?cW;SPzV!p~ez`Vvn%c8>K#*)s`!ZO8*U=?PyV2x$1V13HE$;Qs6
z&lb#9$o7D3jh&udgWZ=sm;FBb3I`2`8ix-@E=M=VM@~9UO-_H#0?vNUbuLye1Fi_J
zGOlM_JKO@?*4#<lH@IhcNO%-^Jb7|?`gqoPIe1NY<9Hi*r}*%EihSODm-z<yzVi$4
z+w!OKcknL@FbWt7#0WG9%m|VRstbk)mJ5yxVT2Tg{DiIuJr_m@%L#i67YmPwAVlOv
zd_=B@jEbT~l|%zY%SB&`5yUjbBE=fS=Edp7kBTRWcZh$I;FWNYxFj(su`ekn=`UF+
zIW0vkWh9j()g|>+T3Fgmx>$N#hD=6JCPAiC=8LR|tcUDX*;jHjawc-Aa(!}p@(S{y
z@=fw93cLy~3MC3J6=@aC6f+ecDWR3LloFKgD*aHFR}NQhQU0tVrsAhkud;kZ;E2bO
z$|DP^+^R&?GSxXXPBj;`QnfjCE_I@Mx%xW|9u0SmYK<jLAx&S+>zbdmB(*}d+O)oF
zD{G(9?$JT&=<B5GJkzDtwbm`teWS;%=cU)6x2`XzAFJPEfHp8TxMc9skj>D|u+DJZ
zNWtioQNJ<4*wVPj_}x+AqoGH;Ob{kUCOIZE$M}u~9_ug#riP|Drn6=OW+7&G%rWL>
z=Ede8ETk;rECwxUES)XuEw`++tg@`8tp%+ktov*zY#eRsY`)v-*k;?#*-6-)vU_6B
zZ0}>=>40^xaj16KJg$2@@A#sloMVdP<OzWjCr=DHF+2G?-E*dJCOWsepj~WT>Ron;
zro?jMrmLZAiR-$Xw%cX5Rd)^dT=x|ZRgY|sB~Mk)Y|mvcRj(Yc6>oL#eD5_MZJ#2a
zFTMu8*L=VGnflfE9r)Y&-w413xCGn|qz?28><?lOIvF$?EE1d&JQt!Ik{7ZbY8+Y}
z28Epn>kOxb4~I`91S8Hy%txw47DsMJ*+jLTq&gXR@@ceibXxRMj9yGtEGpJ5wl9t=
zE-`NYl;)|jcqraAzAu3%Avt03wEpSZM3O|m#Ni~#r0k?`XKc@OC9@@;PF^^xf3_io
zJS8;cWvWW*wR5O*KIfjL$)pvg?Wen^KhBWM$j{i#bjy5vUg~_o`GX6d7oKIwXI;IB
zxfpnH@{;<c+H9)qnCyidlbrTk&fN6eZ+XsnL;14#*DjM>j<`HmaI~Pakhkz+;ck(4
z(L}LU@r@GJlC+ZVSKP0>xT<lr=^ER$jB7tjeM(=I>6f*a^OxsWU@9UjK2+LN4pu2v
z)m1ZBXH@Ui1lG*eTGaN}Db&@~v({%dAQ~bXR<1i<f8MCm*x4l3RCa^@M#fEWGvemQ
zW|!v47UPz_R>jt)TYR@l+GyI++oAU8_Vo_$j=4_z&e7XOxBI$Oy4voD->JFFb+`B)
z-My^)B=?i=A9TlbZ}tTDto3^JF7!F~O+T=EFy3$8|7^f`;L$_9hYtod2fH7sKDs-k
zJaqf9;^U4d@=w~I$~|oxmK$z+CjYE`L}8@!xzh8l(IcbxU#P$69n%?mIBq!pWa8Mw
z=%n@JtCx;1=U%zLT7K>S`pZ=0)Xwzj8T3s0Eahze8`d}FZ-w68n3JEoH?K4Q^qu9q
z=>@li)%RiVcNddCkbTHs;#jI%mR`QQqPOz=<MEG+tHG-~Ybk$G{Z;Zw@YAi&8lRu8
z+pRBt3H@^LHFJY$qvo6Jx4uoY%{N;CTRY#=x0$wUe<=PK{Av4hX(wtIyIZg)wAZz7
zygzdgaBvVz3?=?L3n1YV;pqv$*BUs7u!nPp=KzrO{LU$US-|NE1lz#a2I7HkB7Bib
zC_HKj-GeE>CgGy+9whdp4g`BLCvp!8U&;uov(!a2t+bEnRv6HXyi9t`-YglcEo`$K
zI8GTZXYLH1F5YE+b^&9-c%dfYc~N>X1MygiCdpZ8N*OKLV7W5+5rusvVP$KTgd_E;
zV`@J%*flk^Jhjj1)aX9cTQC5ItVZ(2W=FkE;*aH-)|+*kk6SET?pjmWaNEk+>D${o
z_#cmV%sNr-bj$gX%QW$m8{|&wA?SI;%go!uC))SCU%7vKz~jI-L0?1Ap^RZ7;i?hG
zB3+__P9{WW#uUa@#oavB8Q+`m==5;nXwvwZiR6j1<0+%5!{;8Q^`_s>XwIxTUvlAM
z)|rdpmprp=bM$iM@_6#8@((Vr7Q8HcP;{fXs3iGH;8nY8TBRaov}JqcixtC_ZB<vQ
z&(`?YTGlDob2eZb{<=Qi*ws{eBju)Fvqg(SE88vPt>w07?YBCLI#1vB=rX<|d6)j~
z?!9;SA9XkN<n~7QIX}?p7apJ-*nBuW*z>4rDD83J6N{$`!z{xG&lW}=KCd6md=WHe
zF)la3F!5t@`sLkMS6?Sg5vR3gcxTbGOK%>(y*_twKH{Cjg64anMViI^4{J-a%g<LH
ze(YYoyLRud-cLiHC*aZaV}tse#HQ(1*!QCCfgc+?yt~$WY5Tng2XOpFKtGg+ut(A%
z-=j*=VHi8C5{?;<#;=l0k&Y4u$@<72Q1np_Pz_U$(Jatz(CstOF!D30Fk7<(v!=0C
zv-fel=R$LH^XTz<^JVZi3XBSF3Nwow5p@<z60edNko+Kxkr9;Dk#m)gS143$Q5sTy
zqw-A^tHz`*q@kp#r)8miT*qBEKrc)`${^M-!6?x<@o1vS>0=3|@n*5+(H4=G;Z`Bm
z0XDw2UUnY#t`5ZG&W<NeoN#h<c62#GbaHieBf7hKczXJJ`FjWZMEFMgo$^l#ND0gg
zx)_`nQW$zAtSr1TqB^oJs{Uj{bVE#iY<*nqsjB$0ge#{D6LXTzpE;Wxe>ObDFO_)C
zCe0{aEki1k_dNXt+=U-mA1_W_8p^(%Q<IyQmz*DR*{Q&|P@#yYn4);6Wa-M-)$VKc
zrTJw^<pCA;mAX|@)oe8+HQ#IB*NxV9H#A-^X-sd5zTtDzzS*!vsa50_Ya2Pdf^Bpx
zbxz-Y-ZgON&fVsFb@$7<i+XZ;v-&a~r1qx_oPC%wnEEJVDC=?lldDf_hFhL>j|@Mb
z9sM+h7-yIepVWIvd=>Y)XzKR#)XeT1j<?!#-t!mU-CCGlggyu?nJkB_<bCW|eYv*z
ziR!cPy2cmFuWlPb-=a6;w@!ag*pB@X{?m8o#IEU{#=iIg)35lmfin;%)QHeRj3K>H
zI8-@&65hs?W6g0$Tn9b?K9MevmJ{6JljSOT6GbGYHWfM5G<6M41g#z&E8Qx6H$yI?
z50eHn6Z1ODBi1suSavH8F-{EUJXaTYHjh8AJ|73)7XPq7g<!OhsW6Z5mdKE3k(jTz
zvII_ILb6OMKw41-$PCI}kh79!lAl#5RdiS4Q~ID>t>OirQ5IDz)!g7S$y<#pnvPn`
zTCcP(>sag3>W=B<=vx}l7>pa{8`&AN7|$LpGx0noeC)GnyV)so9SefRgyl6WA8Q%w
zeVfO&`F8I1(hk7k+3~B6fhW|RD4pIpx4EPekGo2^q1>k2n?25Xx_BviQ+coYJoGK~
zi}SY&kPV~?{2VkK+z^r;>Jw%VE)ao<Sd8qCsyuln+BHTamLqmAZW>-y@)AN%A4?QY
z!X(X~xtpASHaNvFl_z!g+(cSqdP;^<ru=#G^Gg@HvkESTTr$e$&EC&>mD`$^mG5`i
zpn$&+Rk%>pUtCp^dd2Um*){o6wlZ|t=klqF!OHfk>gs};%-W>7nEHr@(CeX%5lwM7
zQg7xp*S7SwzHLLbOLn+*Uc0?`NAB*$d)wWCJsW)~{h|X4gV%@BpPU*_8L1qd8t0!(
zdySmVd!st{bK%K{=9Rj&=Ffv)KX1|hFxkC)82{hg(&3(fkq6-NB>?O?0kGBtAd?QJ
zm0$~|LIBLj0I*U5i1iA9XzK$|?dCuG2lOlFq=GX}9v}f{nuc(O=>uZH1yBw;!3bD_
zU{(i`gLA_m=mOLPjX+-zbO<GcGvX|w5itp8bfQQ*WD2qkUVrG}Of3%84DUniXj}9p
z^ivET&bv}EkFcb0hLww*!SUjPaCh){ydAy{ze{3HQcDVv9w+S}&=VpFugO%&D#;1t
zQRE9078DOC<tZDexT%V%>8W#QsA+O&>1m7Uxak_`<>>nu%o*kx!<ev4CCswS!z|vc
zXx2(L9kvhb=^T<AuQ(IA#JHxo(|J^QKJ!-bo#1ES9~a0Gv=U+xdMjKf5+SN1MlZG|
z-Y;=YGG59>T2DqomQ{`*59GHMHWa@qZ7S~^!Kl)z@vEz7SZjuAWovinywxMoS2FN7
zEH|1t%4A}H?2754xrD_j%Moi{n>gE7_6iP##}7_;J59Lg5Ifz(-D^B~y{dc!eQ)?H
z1`GsQ2d{)Cgfm98MOmHv9&;s5@6?xs(nO0hxa6LcxN|CLdl`M_GqP+i31t7w9nHU9
zkY40hVt!S*RG^%pl2DDR1@+)Ms)_U_Lks^c#r9*J-d)LeEAIFAEIl9{kQ}rbihXiz
zxOZfJbZ?wtQtXx5l+ld&8>=~scSi5kK8P(dtn9DO{nh=s_)Emb(M`^+uiKA)7VrA)
zEB#tO5ODlSVZM$P@WWh#2Fx+Iz|6u~m`%6|24UXdCqxG`1g0=2kOkd@#-Q&AR(P%P
zMdTpvAy(jBM;jT2tUyk{D~~EF3{{U>K(nFk;T(JdLx-`&6l3PF0@xsI7Y>87!d2q7
z@J9GD{0|aKlAELyq`{<P1SLWl8G$UCY?<7Fe1t-eqMK5MvWrTEsvFK%hiUBK+%$p?
zLsvvEML)#g&IlN<F{v}XV@_oeW0_<<%_hn=1;_jd$2w;fmm@bl_bAT=UK2hFz8Cxj
z0?vX$f}2A9!g(UzqH1CcVqe9dOEgR7NySNf$XLi~$;ru!DDWzBDsd|Fs0baAQPog0
zQ+L;h)y&gs)gIT`*5lMSHi$92ZZv<C*~A7u*Ss=kxA3yOWsS12wQaP=IJh3~J;CFY
z>in5#@A}YP&ZEYQ#XH-V)Gsvv6_^~14ao?j4lj=6k7|w9iW!UZJhhvUlPHq(FxfQ)
zq?V>>q`%8dxgeZ1aw#H*HTOZjUjc35y<*QR6jwV-iRB~}tyPXS=-S45n}+?ysv9OZ
zzqJ(K(rR1j$hs}xHG4PtzG(M&@2Lj@{VyISJQ5#z^W@U7{<DMUcV5Jft50HIPQ0$2
zPMURnYc#L4Ah{^9q_ASN>hV|l=i6Vte3RLV-yYuK+dKCw{z!laG%#N$3ABJM%p<0O
zYA^skKqQbP%m$r-WBwLFh0ujLomRwONMWQ8vL5*f<`CmhgJ?Rm2f718hVj63W7)9r
z*mpQXTq~XnpG|@xNg&xFjU_!Gq>|CVvs#J#1w}9=HDxE2J2egUAWZ`85!yYvKKcv>
zJ4PYKJ*G+KW|m8=VQlv7TJY|}%0<qN;Qq<;jdz`Io&T%Amf)T+R)kiRM@&}SK*Cw_
zq*ShSi_DnpraX&+wqk%%q4J>0wyKDli~41a=UN19Bb{{JVSQ=?d&3H&&qviwE*<+|
zre<Dfv2EpG-EAvqm*;>!9^?4cDF}{Txa*#Kx+jZQvyZXwvVVG@WYFu7)G)>HwaCho
zPBE;pGpDX4cqED@Z6)`nTsY^LE}F4-ek7|Lj+#LpTmF}Vfuf?4z^j_2v}GSEI;v7@
ztn0YySFg7=Mcq_r{?^*qM(m*I?Cd&z=li|$-7G!jeOwO;25=992S<m5AICiD8%90T
z7zuw~Iy(4bWgIcVFe&g-@|E0cxhctM!5QXRlG$%>X5MzsmCeV$vtN*Wk9q%cvGzm6
zlGZYQ`Nc~9M~79`)tR-DzwAEIeH!_EZe4SI`^$~5?i-97Prt=)N^Q<<UHq>3ePg@o
zht*Hi&(<B6UGm+oJ@<X`{k8*#!}y1J%19sX9snX}!Sv&sgM$rz0I=Br>|HuI*eO3a
z*sFk(4fq>KkN@xQ6^F(cm~$_2K14li9;XkV|9<@!M&f%8Nam8p00009a7bBm000XU
z000XU0RWnu7ytkil}SWFRCodHT?u#;Rkr@KbUNvfeG_5`YY-wNfPp{+o{ADgGcxep
z5O;8ydCWk3pWowCbe1RjK4lzy;4&jKqk}U-a1=+ud7z@;LLwlFC>S)v1jwFrI_XY2
zop;WyuIf%_F<t2n-Ic!IS6#QNZr{59Ip>~x?x|CCgE~7q5lBOq0>MKUdH^|7ARquk
zTn+*P5D<VNPJtl~KcE2u0x-lWFvQ`9_7{IpP!Ma2Ag8UrfAvN=8&%31Xc{{x^H{Co
z5Bf3s5%3VAbLh&IH$@c|uDl{TIw}*k&TyNplZ9bWQ*bafH>lHMG@8ELxbaVWHf?&T
znHpfF&E_pZ&^rD;1;7qozi0Q$(`V)7{8<+kI>wdbHk%E>!9AN2eO+^{$KB)hHtVU6
z4;0@%KYw`%{kM%aj|)L>`1``u*EM%B_Ep|f_<QCY-I>7iHT~t6&rZsneaT;XVt##n
z3*O&%0=#!k4Gq$@x_XoAC663)d$?Wm=U<vc73yj8s7C4ye?8%O?|sJH9hZdp8`Rda
zV)AbC6#gCsfWR{`**=IUEL^cLOs^a53WvcKmwX4axn1m?g`lRU8J`}Mu9cL0kEW(p
z*S~w@nA^2_^<7tUJ89F|+dbgAd&6H>XTrha?_sgD)BZa!4dhf)W5g$)o+5f!@!6p=
z7>#E6lGpa0z~7?)*juclePn!mT$U>W2F?VqT7?}(LqHHhL#3+DoNXk5_#Pb{(lwSP
zZ<=X|iSbjYeFoatR`H}3=!RdX3qeSTbc>FTPC<i4yKox}dW8l|7rAq|($3Sz1I6E)
z0PuMxfT6vuJ<7S!uFMY$3&G48V-OV;j??9JIDGgtTjt^~;o&;W${mB481A6nNfvWk
zG}X)k!0qh~kGnVT5kF&`L0DN3I5DFSlBQz~XE21}l1oOBfJjYXO~BgEYVC0M-fop2
zDE{6E#$wJW!`duuVRDQ5Vt#39HNO0^LN1EPNWEbGuwmiKId5c0K$-Du1N3`J+mCki
zh@bx#O+t{9<6v}qdxy{j&CM3X#uzYdYMN4g6Ac50s6((}=&dwR{Cx?4=mpk}4k7Lh
zja}8gbI(8i^%;C~;;e!(COSeeKO#br9p##*Gxqi|^SQmA8UnbUF8H0>&5WKoW3<xS
z+uAzF<~aqE$%3S$Xk2`8s#2WoA?J>vT<}n4p!j<e0B)WjD5-NnRG#WdwxQH=@&k}$
zOR8R0dom`Lc=?)q&RcuB7q?Ti)x-Dbg5R}n#wZVUe2wiO-Z*u-^!Q-=uplma_-CN_
z8Dw8c=~1luSn8d}KeqH68(WbvJ_VB|4Oi+hk~;kO6RAfiC;c;{;CF65PXUWd$`m2j
z>ve)Qtntp05&Y$`N~L&mauhNrjZlt#E%Rdnz*4RdA(~WsS0P~4Cker*^h9K3rID79
zAhx!)2_f*-6tD+E@|~5o_HbR*DQEm#fix64W;xPOIEsuwz3>ej`Mg}wlx+M?%^s;7
zt7<_1|D+24j|zb6{d*Duo)R*nQ%A&N`m}UK6}Gim#oV|jr-^I5{&3u6Y!z0&JjK=N
zf~iA{0UNr_&1RH*=FkdaRxmwXu@ih1pW6b!KwO1@&&hNB<m70YDveg=#B4Xn!v>f0
z=VYU~zns|bF>|Ig{pE8Oi&e4q8Sf>;d>$HnJ*g4^2E{@!BWJXj|MK2>t{)#4iCiKM
z_X3_Wd3!22SVWGECF_5t9Wx1ebd<xtk&rM9vu2GY%sRn*f5AMUxid3;-rS5H$2b;{
z-mVQ6!@O~ym}o<<xL6?O&7IIAJ-pEtPt>Ve1IRabo*K&M<b7Y)<TC!a?+aI7*L)G{
zfCa!8K?T~>e+mp(08G`jsI~A7O*rz=A?*I(Ym_y4*Z<vrc-}ABZsXaPWE_55X~M~}
zTIJaLcOUp224n-2zVikie7yn%YnEfm!5#SRe*5FmzkQB}p11+o{Fp|=>BHj<`2EIL
z@XCfeu<hfsczydWJ6bb}o?S)x0g-o5D5qWnbezFoiTRwunk%qdX1+L6i)|mA5jidY
z{0}jolhGWEzbnk=GvPS&`Ekzpd{Qp}jg2@(@Fp)^ip7aWa9kF~U5c$wF9JUgt8d{b
zE$t=?X?=dwwZ3ox^kn_>GtW8G6RGFlFM<@CjE-OtU#5a;0kB%yXw(N%<3n(~<Qsu8
zZ|+#=eP8Q&gdZK`A?MLQF(CmT9+sYjF8fKIN9rETN>sBeG(H{~)Y9EAyo%kT#Rg2j
zpdOnacnjrpoDswQL%S&=xD)LJZ^c?^7~tUKxVSW2U-+UJ`I8c2{Q|sd4FLUcTr-0M
zaqMa26wFKpz7U~s3AlNV<LRJWL`!?_{RP)vp+K58DGni_Hng{J{G%C}6UQspd^w|9
z18~65>^qhrHMbm9<`9gTLcVV_VCkY<jGCK;u}-q#%LEum+wQu00v693hqg8|E+F?e
zr#1I3!kny;lw%R`{p7X)SelGzum&R`J{*=hGlr!kVHo3VG@_P-B6~JTzO@CWjwnQm
zSR&sif;%k~Gk*Hv$KL#GDAqtA4UzOxf>cW$bp+1aV?*4j`n;5NQvl5P$NHC1)DVqF
ze?14Uta}S5dTDmrRR#Fn;tPAZ>c6M&cw`%zt17X5(`x+mXPZPMYENh$xHA{IIn#Q&
z<WuL|wgH?ccYQ_s*J*Sb=&abC^R8TNFm``&#I=ks9J{Fj!m;GPmSYjgeN40tkN)>^
zG}YF_5*3HIuofIEDMeLB1jc8M#;C+D(d52>)gx`#@~i9ZqkAV_+e~x*&R~QFvHtHw
zX=O8P?QIyJ9<S9QfZh~|rm@{l%9`SLom|1Yuu=8~?mQv6f;~cV->Ss9*B|&g;0hMp
z3Alm-uHb+xn7Ts16&!E{`__2XkJh+p1UhOAxPk+&;D9SQ;0g}7f`^<d*f9ujPKWd{
zNwA(ei(>~4p*<oBsnf?}Y_tY_z+HhWM{(ruZlLc|d{fqlC~^zWp3Kn@&XYx@s8428
z+W?HF@nkVpZ73FT0nX?mGjp-|=hGB_f%7#j<?C8Quwu<K-PpY#ZzNXSa*ba=b=wuJ
zprVklr=6G|NiR*v4kFXBPw?d4Dy67aGq*W0e|U1NJv0<}=fG*4{I=9DV5`;u*19?o
z5{B?pJoNC@$d2dNpxSZdpKs&0Z&je7$@!^)U(mv>Mp`Hum_uHM8Ep9TllPO>m-^Cs
zpVwg1bK6i`-w1z*2vDs7WXVaJJHyU=rk@Vk3#W^iKzdl}7D4^3u#E2B8*>%rGlt8u
z5=Bg)^vMF>N2OW-kTeo=C=#;#Uwg6hiz=At%UPznGuZL$9uX3jIcgXzEoL+}ne7De
zePX!NLIZ__1sfvp<d*$}QF`*n@c|qD`l$#Bn=o#aJJp@HvT{duzX6s;924PRO?YW@
z5nghs!8um&p{H&~w#Fia`1v1)e}&ROaMy!>aY5fTR<P*VpVM{qeXU#N9z3DGRUK^<
z`8ISD3xNMj_Il@&dJSMm!=vkNfHgFv_xE4u6KU4aaHMFg_cJ-U=WN?_Kka)L1Nw>(
zUH5HKQ7-^w@TCk-ATqS$+;^2Y-9Yg{p~En8>~LcE&~OCN2SO-y!qgT7qsff0kWR!$
z^D81!lBm$TfXL;}=Y9YJK+SF{!{d*<g%X#sAaChntiD3qVd2a){!IDu!avzV!TF4o
z9xGgkjB%soL!aKOw*koiw_`(h<}=y@MgPFg@A$O|ZKq#6{$@AkGvc*R?Si#UlQDr#
z3~i~w{&Resf;DUNF*}X?fEFV@D0-Vh!)meXqjD6C9^>=}ZDsk}pA}{0WdF3_)n|T5
zFNK7P(SF;zrP#jx9qieE2>F-K@p;gyHGt(@rI_!hEt)McpP}lbFn3v=a0JCAI=-Ld
z^HfmLKw}#PgVO)j-n&3BpR3@}{)WrPilHHGIK3w22T8R6=u<`rMwjnBh~jFy5zt}A
zN81hv!KkMXNNPDnh1mq7H@>uwma1@k3;2!wtQCOj+9tn%uigkWBw{AL|5)BofhX2&
zA+XZ302%fCsUzg9CimQPVv`f;C6O8|{n>ML#6sZcPqU_9<QKRdEq1%Pk+|_c^D$ZZ
zpTp?8<hlYZ8Sj`5DMd7vdH&JsA-;|ya5+f%=&CEBr`QO8Na{5Jql+JdN1k1QvXkYg
zs3w&k5rOovX-K7)MutMt)1dz*eGsClwG%b<Mnp!CZzRsYY@Y!MK;y}yczWw0oN9FR
ze!+NJMke*DMR;s+x~Aozt~D6zfBzb)s_VMoY^H}`58OE$vvRV#ocF~8^#ULQLiXc}
zUQkuE$PY|=DWHMVx2^eKu~LOA6rK$<8jY@}eezhf09Y%}*qLuY{&mxkmF%SY=ox@6
zF~uhVd#X`JdPSa>DPe!$!>g7c<pOq#ONp0~Ns93gcerW+2#u6Nz)abBxO;&=#J<Sv
z@ytFfgdg<)OXiP7*2GlLRC*q2ahR^|-aI}H3;8oRM~>oyleK6R!5=0O9Kit+4(r(6
ziv6QJ8-P(X4Sa3SakRGjFIv?a0G4_jZD3}d!^RD-cH>&cq5?d2jrKkeAp_;!Ur#;&
z9W7Y4e9epUX=T6m-g%gom8l&2YDT>Vpn#D2K2TLOYC9;D1)wkDRn>N#8T3J_^Lk0W
z2GEDo5^3Wxdgdfd9w7&WOIUcVywJ$#^9sz{H)rNATQUdN%*}+3f?}K#TL)6Cfb&`3
z%&Qjw3IaWJ_$1z;4dDsM&%YQ~=42pUgopbkSWmW!9lu+5e2Bl(Hp~!=)psw#l#5d7
z<59t4!9`Er%bRtn7l4p3WRMY9&31sf7Q0{HC$^-K>G(;07G_Pk5PmWfQbk{$>nD;C
z$aX+;iw(co_@<~Qn^p+B=a%_MiWA>XQ&sn1{z<(6(1#*dufHEF>#Fe8m!&8!F2%dw
zHlg}-8UFYJZG<8tdn)d^eHPNC3G-m$^7_440RBMV3*u1l6Q_-M<NrRRMWepLk*_Lo
z-`bbN{R`%2;?^Z|DbJ;SGxcT1^sB{@@yW=TI0_Rc#C0ow;>ckXuK!rmQ$k)#dR$sG
z@^U71!@qOSF|2)@pOpG;Qm+AE#NKTmpy<6aRJ-8I$ex7UR10>zRSMI&Dx4*+aC%oe
z$>ksZdHCl3@33X-u5M#~!F>8s>bP;(@Z1iZ5DQ57E<W`F&{Exn`H?9pD7b1W{*9JP
zOVBg7&1l$Lik;?q)RUhiU1Oo)Yy>(pe>^RmdH=2Rkv1Y;;r0f4a|kUQI?AO7tZbEf
zJ(*E203jiWBR5FKRnt*$=_L9l06hS)bRb+XpPQ(|6)W>G1u?i-W6WoCJgU<Q$tb+@
zD$Jffj@}LJ{)WH{zrK~CgV-P9o6Kh%ssy0+_)+}f)#J#erD6&w@S}9^J6Kv&%Joq)
zVCl;1aO2ciq)r?I|MikfXnOjsHU;7FXSU<j4Ywhj0zZnsHR6e<UbYvt;`zsJK|)+S
z62!-nogR|)RM~kx=eB4z9^a39Htt8sxmx;YG^zT!qj*qZKBH+t(F@!0k$*VC+$MjO
zDHQiTunA>lRkTWYJ9y;~2lKhQP~5|72z2_#^8q&npdI^OKWZnM4)jd~l<t#{4Y+Cf
zJmk?iep_h!z8~!6w6%M3%WgX$8wEf`;a??H_G2r4_wX%<Bi{h4{7*`b`2}z{1;B(K
z1YRBPTX!e&hj*a(pcMU@bm?3ynVpH1H&2p)w`(&X<ooZ-37-u`n^q8?b=dLG_XXa&
z|9vVMfYDf2c#}A8B~Ph8B=-cMr4A=W1b8$?C6oGZsl=!338XY{hCK?{ZdF8R@clrx
z<=e!HAY|AD>xFIKK%PKOm(9u+`!IG4P>PAtq9@Rh0<!HW6ucJiefk6H>JE!{0DuH!
zkK`y|6ZXDM&ju*fYcM2?dkd?0BQd?AvKl9=rI$l^%Bzo%82<d<-V^wnjLl;2>pwp_
z3!t@d`N^j}MPee&>2}gr!FRvB)4o^~UCPYDMfxiI>b@c+MsVI_ZG?n%#SdILF9)yD
z8iBv~&32h6$j=)^`5;_--)1F7aK==Pycf`JwRRcIa&EjD`NGhX@h9M+TM4YCmA;oJ
zrO3=nv3MeD1n(z%`&dZj&7(JU#eehVv~0XE^yJ%^arZ3+;^s6cinJi_LRv*8MlRsh
z{Xp^er2%-zvwii|iPQND<~cxwB;)S&_u$&{D%8_7aQMh%>8YP30yAe!z=De>;j*0J
zN>6b7(K|VAAJyy)=J$-BZpMp7n5{I{+sN@1<}jm{UYm<6az<l#QupT~-a!ux6n{`>
zC)2KLBDKeY!To$ha&qG2BZqfAotPNM^BbQ^H8u4$*;5z(vZ|_v=c1LgH4&aJ8cR)s
zhZ25=_;#ffO9d0sLd30K^&jiDoI6+3R|Htse-FYDw`bL=buUu;*yY6jR@v$9iMtOO
z{Jm)a77X@ba<Dr`c8-aSjUKyf*>%$<P8>f%7edh>l!!{woQDqvAyLn?wOiY*$B%zo
zv32X~pEWczvH$rLZ56cfy6vr`0a$epDA9d}4E`PkfT>4BU?%e$j!CrfB%e1P1~}M{
zuQ8DZRRHLI>|J6XE5CNbPoY`u^Tv~L_DESt0J@K9biv&;RPgs@1TwMtC4bqg&n_U&
z^RqpU@fmCZV8(Krcxd8Db|Y=v9v+%_sqO*ye5%7a4GH|cY5=AL^#T?U?(IAraOf}Z
znfd(s?_l?Sx}{(;kM%5!ES&ry9?r8?uz9NYQ(Ynr1^j&q08@d8z|&jaWMSaE-1`Sx
z2*lKk?$1KN8*2mJGw(g3`l+riN$dE3Q~;P7L<D9yC)d4ncJO{>Cd=wx?7hW&8J3pu
z_e%g|LIn2Oqk!C_wTCQ#s9zKa2tdEcq}@UR0njdQ`-LnZ0R1A9b_)drK)bx{7qWl=
z^ovZ|Eff#{?eex?$N~b;FEVMjP(T2*%iDe-`+v|7m{y$1dn*6{002ovPDHLkV1lnB
B5rhB$

literal 0
HcmV?d00001

diff --git a/docs/haddock/tensorflow-queue-0.1.0.0/tensorflow-queue.txt b/docs/haddock/tensorflow-queue-0.1.0.0/tensorflow-queue.txt
new file mode 100644
index 0000000..2db224c
--- /dev/null
+++ b/docs/haddock/tensorflow-queue-0.1.0.0/tensorflow-queue.txt
@@ -0,0 +1,26 @@
+-- Hoogle documentation, generated by Haddock
+-- See Hoogle, http://www.haskell.org/hoogle/
+
+
+-- | Basic access to TensorFlow queues.
+--   
+--   Please see README.md
+@package tensorflow-queue
+@version 0.1.0.0
+
+
+-- | Queues in TensorFlow graph. Very limited support for now.
+module TensorFlow.Queue
+
+-- | A queue carrying tuples. The underlying structure is more versatile
+--   and can be made to support arbitrary tuples.
+data Queue2 a b
+
+-- | Creates a new queue with the given capacity and shared name.
+makeQueue2 :: (TensorType a, TensorType b) => Int64 -> ByteString -> Build (Queue2 a b)
+
+-- | Adds the given values to the queue.
+enqueue :: (TensorType a, TensorType b) => Queue2 a b -> Tensor v1 a -> Tensor v2 b -> Build ControlNode
+
+-- | Retrieves the values from the queue.
+dequeue :: (TensorType a, TensorType b) => Queue2 a b -> Build (Tensor Ref a, Tensor Ref b)
diff --git a/tools/haddock.sh b/tools/haddock.sh
index f31248d..5f1b7ed 100755
--- a/tools/haddock.sh
+++ b/tools/haddock.sh
@@ -10,8 +10,9 @@ STACK="stack --docker --docker-image=$IMAGE_NAME"
 $STACK haddock --no-haddock-deps tensorflow*
 DOC_ROOT=$($STACK path --local-doc-root)
 DOCS=docs/haddock
-git rm -r $DOCS
+git rm -fr $DOCS
 mkdir -p $DOCS
 cp $DOC_ROOT/{*.html,*js,*.png,*.gif,*.css} $DOCS
 cp -a $DOC_ROOT/tensorflow* $DOCS
+rm -f $DOCS/*/*.haddock
 git add $DOCS